t t,',' * '. 1 '. -I 'v C, r ; , ill;': '■.•,'..! ■ ,■ ELECTRON MICROSCOPY Proceedings of the Stockholm Conference SEPTEMBER 1956 ELECTRON MICROSCOPY ELECTRON MICROSCOPY Proceedings of the Stockholm Conference September 1956 EDITORS F.S.Sjostrand and J. Rhodin ACADEMIC PRESS INC., PUBLISHERS, NEW YORK, N.Y. PRINTED IN SWEDEN Alinqvist & Wiksells BOKTRYCKERI AKTIEBOLAG UPPSALA 1957 To the Memory of BODO VON BORRIES CONTENTS Preface F. S. Sjostrand Zum Gedenken an Bodo v. Borries E. RUSKA Commemorative address in honour of Bodo v. Bor- ries V. E. COSSLETT I. INSTRUMENTATION Observation directe des surfaces metalliques par re- flexion (Invited paper). Ch. Pert Objekticiihlung im Elektronenmikroskop. O. SCHOTT UND S. LeISEGANG Richtstrahlwerte der kalten Kathode. L. Wegmann UND M. Gribi 1 II. ELECTRON OPTICS tJber asymptotische Bildfchlcr. ■X P. Lenz 48 Zur Errechnung elektronenoptischer Peldvertei- lungen mit geforderten Abbildungscigenschaftcn. 52 5 K.W. J. PicuT Lentilles cicctroniques magnetiques. Regies de leur construction et expressions uni\crscllcs de leurs caractcristiques electro-optiqucs. 55 P. DURANDEAU 8 III. ELECTRON-SPECIMEN INTERACTION A Scanning Microscope with either Electron or X- Ray Recording. J 2 V. E. CoSSLETT AND P. DUNCUMB Imaging Elements Operating with Permanent Mag- nets. 14 B. V. Borries, G. Langner and W. Scheffels LJber magnetostatische Linsenanordnungen mit me- chanischen Regelgliedern. 17 K. MULLER. Der EinfluB der Bestrahlungsbedingungen auf die Objektverschmutzung. 20 S. Leisegang und O. Schott 27 ijber ein Elektronenmikroskop mit universeller An- wendbarkeit fiir Elektronenbeugung. 30 H. Bethge A New Universal Electron Microscope of High Resolving Power: Metro-Vick Type EM6. 32 M. E. Haine and R. S. Page Uber ein neues elektrostatisches Gebrauchs-Elek- tronenmikroskop. 34 H. Mahl, H. Volkmann und W. Weitsch Bolzenkathode als Objekt im Elektronenemissions- mikroskop. 37 E. B. Bas 41 Summary of the Proceedings of a Symposium on X-Ray Microscopy and Microradiography, Cam- bridge University, England, August 16-21, 1956. 42 W. C. Nixon Uber die Entstehung des Kontrastes im elektronen- mikroskopischcn Bild (Invited paper). 60 B. v. Borrils und p. Lenz The DitVerential Scattering Cross Section of Atoms at Small Angles. 64 M. E. Haine and A. W. Agar Experimentelle Untersuchung dor Strcuung \on 70-kV-Elektronen an KohlenstotTin klcinstc \\ inkcl. 67 G. Kempf und p. Lenz Experimentelle Untersuchungen zum Kontrast diJnner Schichten im Elektronenmikroskop. 73 W. Lipfert Zur Veriinderung des Streuvermogens eincs Fest- korpers gegeniiber mittelschnellen Elektronen infolge lonisation und Anregung. 76 W. Scheffels Contraste de phase et contraste interchromatique. Etude comparee des methodes. 78 M. Locquin Der EinfluB von Temperatur. Unterlage und Be- deckung auf die Vcranderungelektronenmikroskopi- schcr Priiparate. 79 K. J. Hanszen IV. HIGH RESOLUTION ELECTRON MICRO- SCOPY AND ELECTRON DIFFRACTION Der Durchgang von Elektronenstrahlen durch das Kristallgitter und seine Folgen fiir das elektronen- mikroskopische Bild. 86 H. NiEHRS The Resolution of Crystal Lattices (Invited paper). 88 J. W. Menter Elektronenmikroskopische Abbildung von Kristall- gitterstrukturen. 93 R. Neider VIII Contents 107 Investigation of High Resolution Electron DitTrac- tion Patterns from Individual Micro-Crystals by Using a Three-Stage Electron Microscope. 98 W. D. RiECKE An Electron Microscope Examination of Freshly Prepared Silver loide Sols. R. Ottewill and R. W. Horne Y. SPECIMEN PREPARATION TECHNIQUES IN BIOLOGY AND MEDICINE Problems of Osmium Fixation. 106 G. F. Bahr, G. Bloom and U. Friberg The Quantitative Assay of Lipids Extracted from Untreated and OsOj-fixed Beef Brain. G. F. Bahr The Fixation of Nuclei in Locust Testis. 108 I. R. Gibbons and J. R. G. Bradfield Ultra-thin Sections of Avian Tubercle Bacilli in a New Embedding Medium. HI A. M. Glauert and E. M. Brieger The Use of Gelatin for Embedding Biological Ob- jects in Preparation of Ultrathin Sections for Elec- tron Microscopy. V. P. GiLiiv On the Preparation of Ultrathin Serial Sections by Means of a Watchmaker's Lathe. W. NiKLOWITZ How to Prepare Ultrathin Sections of Tissue Cul- tures. V. DOSTAL Eine einfache Vorrichtung zum Anspitzen von plexiglaseingebetteten Objekten. 1 1 8 A. Maas An Improved Method to Prepare Formvar Nets for Mounting Thin Sections for Electron Microscopy. 120 F. S. Sjostrand Experiments on Staining Thin-Sections for Electron 1 ''1 Microscopy. I. R. Gibbons and J. R. G. Bradfield Die Eignung und Anwendung von Phosphor- wolframsaure und Thalliumnitrat als Kontrastmittel zur Darstellung cytoplasmatischer Strukturen. 124 K. E. Wohlfarth-Bottermann The Use of the Electron Microscope to Control Preparation of Cellular Constituents. 125 M. S. C. BiRBECK and E. H. Mercher A Micro-manipulation Method for the Preparation of Calibrated Microdroplets over the Range 10-'=- 10-1" ml in Electron Microscopy. 127 Irene Sugar ijber die quantitative Spreitung von Zellen. Eine Untersuchung dunnster Filme mit dem Elektronen- mikroskop. 12° A. Kleinschmidt 113 115 117 The Effect of Acridine Type Dyes on the Submicro- scopical Structure of Large Molecules. 131 F. GuBA, G. Hajossi-Kerek and G. Romhanyi VI. CELL ULTRASTRUCTURE, GENERAL Mitochondria Elektronenmikroskopische Studien an Leberschnit- ten von Thyroxin-behandelten Ratten. 134 H. ScHULZ, H. Low, L. Ernster und F. S. Sjo- strand Die Entstehung, die Vermehrung und die Abschei- dung geformter Sekrete der Mitochondrien von 137 Paramecium. K. E. Wohlfarth-Bottermann Golgi Apparatus Changes in the Ultrastructure of the Ciliary Epi- thelium during Inhibition of the Secretion of 1 39 Aqueous Humour. A. HOLMBERG L'appareil de Golgi des protozoaires et son ultra- structure comparee a celle des metazoaires. 143 P. -P. Grasse The Morphology of the Golgi Apparatus in Neu- rones and Epithelial Cells of the Common Limpet Patella vulgata. 1^^ D. Lacy Cytoplasmic Basophilia Basophilic Structures in the Cytoplasm of the Sea Urchin Egg. B. A. Afzelius 147 150 151 Plasma Membrane Functional Changes of the Free Cell Surface Mem- brane of the Intestinal Absorbing Cell. F. S. Sjostrand and H. Zetterqvist The Hepatic Sinusoidal Endothelial Cell and Its Histological Relationships. H. F. Parks Observations on Early Stages of Phagocytosis of Collodial Particles by Hepatic Phagocytes of the Mouse. H. F. Parks and A. D. Chiquoine The Role of Cell Membranes in Morphogenesis. 156 M. S. C. BiRBECK AND E. H. Mercer Electron Microscopic, X-Ray, and Birefringence Studies on the Proteins of the Hair Follicle. M. S. C. BiRBECK AND E. H. Mercer The Mechanism of Hemolysis Caused by Ultra- sonic Irradiation I. W. ROMANOWSKI, A. FeLTYNOWSKI AND J. LlTWIN The Mechanism of Hemolysis Caused by Ultra- sonic Irradiation II. W. ROMANOWSKI AND A. FELTYNOWSKI 158 161 Contents IX Cell Nucleus L'ultrastructure de la membrane nuclcairc dcs ovocytes de I'Araignee {Tegenariu doDicsiica C lerk). J. Andre et Ch. Rouiller Electron Microscopy on Grasshopper Spermatids. F. S. Sjostrand and B. A. Afzelius The Acrosomal Reaction of the Sea Urchin Sper- matozoon. B. A. Afzelius The Structure of Galea Capitis in Human Sperm. J. Schultz-Larsen and R. Hammen Thyroid Gland The Ultrastructure of the Thyroid Gland of the Mouse. R. Ekholm and F. S. Sjostrand Electron Microscopy of Chick Embryo Thyroid. R. Stole, P. Blanquet, A. P. Lachapele, R. Maraud and A. Magimel Kidney Preliminary Studies on the Development and Dif- ferentiation of Cells and Structures of the Renal Corpuscle. B. V. Hall and L. E. Roth Further Studies on the Nephron Ultrastructure in Mouse: Terminal Part of Proximal Convolution. J. Rhodin Adipose Tissue The Fine Structure of Brown Adipose Tissue in the Rat: with Observations on the Cytological Changes Following Starvation and Adrenalectomy. J. D. Lever VII. NERVE CELLS AND RECEPTORS Neurofilaments et neurofibrilles dans les fibres nerveuses de la Sangsue. R. COUTEAUX Elektronenmikroskopische Untersuchungen an Grenzstrangganglien von nienschiichem Operations- material. Hedi Gansler Ultra-structure des cellules visuelles du Gecko. Mise en evidence de prolongements cytoplasmi- ques infra-microscopiques au niveau du segment interne. Nina Carasso Some Observations on the Structure of the Retinal Receptors of the Toad Eye as Revealed by the Electron Microscope. F. S. Sjostrand and L. G. Elfvin Submicroscopic Morphology of the Retinal Pig- ment Epithelium. G. Lion, C. Maertens and G. Vandermeerssche Preliminary Observations on the Ultrastructure of a Frog Muscle Spindle. 197 162 .1. D. Robertson 164 167 169 171 173 176 180 182 190 192 194 196 VIII. MUSCLE AND OTHER CONTRACTILE ELEMENTS Preliminary Observations on the Structure of Insect Flight Muscle. 202 H. E. Huxley and Jean Hanson The Ultrastructure of Skeletal Muscle Myofila- ments. 204 F. S. Sjostrand and Ebba Andersson The Tubular System in the Striated Muscle Cell. 208 Ebba Andersson Elektronenmikroskopische Untersuchungen am Ute- rusmuskel der Ratte. 210 Hedi Gansler Elektronenmikroskopische Untersuchungen iiber das Gewebe des glatten Muskels. 212 F. GuBA UND G. Hajossi-Kerek Mechanism of Pigment Migration uithin Teleost Melanophores. 213 S. Falk and J. Rhodin The Pharyngeal Protein Fibres of the Ciliates. 216 Ch. Rouiller, E. Faure-Fremiet and M. Gau- chery IX. COLLAGEN, CARTILAGE, BONE KoUagen (Invited paper). U. HoFMANN UND K. Kuhn 220 Neue Befunde zur Struktur der Sehnenfibrille an Hand von Dunnschnitten. 223 E. Kuhnke UND K. E. Wohlfarth-Bottermann The Proteolytic Action of Papain Studied by Means of the Electron Microscope. 225 G. Lelli and G. Arangio-Ruiz Electron Microscopic Observations on Frozen Col- lagen. 227 G. Lelli and G. Arangio-Ruiz Electron Microscopic Obser\a:ions on Collagen Exposed to X-Rays. 228 G. Lelli, U. Marotta and A. D'Amore Studies on the Fibrogenesis of Collagen. Sylvia Fitton Jackson 229 Further Observations on the Transformation of Collagen Fibrils into "Llastin". 230 M. K. Kfech and R. Reed Osteoarthritis of the Hip .loint. K. Little and L. H. Pimm 233 Correlation of Electron Microscopy with X-Ray DitTraction and Optical Birefringence in the Study of the Bone. 234 V. Caglioti, a. Ascenzi and A. Santoro Contents X. PATHOLOGY Elektronenmikroskopische Untersuchungen des ex- perimentellen Lungenodems. H. SCHULZ 240 Electron Microscope Studies on Alveolar Cells from Mammals. 244 A. PoLiCARD, A. Collet and S. Pregermain Vergleichende Untersuchungen der Mitochondrien in Rattenlungen nach intratrachealer Injektion von Kieselsaure. 246 W. KiKUTH, H. W. SCHLIPKOTER UND P. SCHROE- TELER The Lung Tissue in Mice Infected by Tubercle Bacilli. 248 B. Cedergren The Importance of an Accurate Size Determination of Fine Particles when Investigating Their Biological Effects. 250 G. Bloom, J. Glomme and A. Swensson Elektronenoptische Untersuchungen von Staub- korngroBen in Staublungen. 251 H. W. SCHLIPKOTER UND A. COLLI Electron Microscopy of the Glomerular Basement Membrane in Experimental Amyloidosis of the Mouse. 254 F. Miller and A. Bohle Electron Microscope Investigation on Biopsy Material from Patients with Renal Diseases: A Case of Subacute Glomerulonephritis. 256 A. Bergstrand and H. Bucht XI. MICROBIOLOGY Some Observations on the Structure of Tobacco Mosaic Virus. 260 H. E. Huxley Electron Microscope Studies on the Periodicity in Tobacco Mosaic Virus. 261 R. E. F. Matthews, R. W. Horne and E. M. Green Filamentous Forms of Influenza Viruses. 262 A. Feltynowski Fastening of Phage Particles to Bacterial Cell. 264 A. S. TiKHONENKO AND A. E. KrISS Filamentous Form of Bacteriophage at the Earliest Stages of Their Formation in a Bacterial Cell. 265 A. S. TiKHONENKO AND A. E. KrISS Comparative Studies on Sections of Intact Cells, Protoplasts, and "Ghosts" of a Bacillus Species. 266 C. Weibull AND K. G. Thorsson A Study of the T>\\\i\ono^ Saccharomyces cerevisiae Using Carbon Replicas. 268 D. E. Bradley XII. BOTANY Die Ontogenese der Chloroplasten von Cliloro- phytiiin comosum. 272 E. S. Perner Some Botanical Applications of the Carbon Re- plica Technique. 274 D. E. Bradley On the Ultrastructure of a Fungus: The Gametes of AUomyces. 276 G. TuRiAN and E. Kellenberger Elektronenmikroskopische Beobachtungen iiber die Warzenstruktur bei den Koniferen. 276 W. LlESE XIII. PAPER AND TEXTILE RESEARCH X-Ray Microscopy of Paper. 282 J. Isings, Ong Sin Poen, J. B. Le Poole and G. VAN NeDERVEEN Partial Embedding Technique for Replication of Fibres. 283 J. Dlugosz A Method for the Carbon Replication of Extensive Areas of Very Irregular Surfaces, with Particular Application to the Study of Pulp Fibres, Wood and Paper. 285 D. H. Page Further Reflection Electron Microscopy of Pulp Fibres and Paper. 287 H. W. Emerton, D. H. Page and J. Watts Application of Ultra-Microtomy to the Fine Structure Study of Rayon Viscose Fibre. 290 P. Kassenbeck Structural Details of Natural Fibers as Observed with the Electron Microscope. 292 C. Maertens, G. Raes and G. Vandermeerssche Contact Region between Two Fibres. 293 Saara Asunmaa Die elektronenmikroskopische Darstellung groBer Perioden in Cellulosefasern durch Einlagerung schwerer Atome. 295 K. Hess On the Submicroscopic Structure of Mannans. 298 H. Meier A Contribution to the Structure of Keratin. 300 M. W. Andrews and J. Sikorski The Electron Microscopy of Pigmented Keratinous Materials. ^^^ J. Hope, J. Sikorski and C. S. Whewell iJber den Feinbau der Spinnenfiiden. 307 R. LeHMENSICK UND E. KULLMANN Contents XI XIV. METALLOGRAPHY AND OTHLR INDUSTRIAL APPLICATIONS Direct Observation of Dislocations and Their Movement in Metal Foils. 312 P. B. HiRSCH, R. W. HoRNE AND M. .1. Whelan Dislocations in Stainless Steel. 316 W. BOLLMANN Migrations of Grain Boundaries Studied with the Electronic Emission Microscope. 318 R. Arnal and M. Sorel Perlit- und Bainitgefuge in drei KohlenstoflTstahlen mit 0, 1 8 "o, 0,50 "„ und 0,86 % C. 319 S. MODIN Selective Oxidation due to the Heating of the Evaporated Film of a-Brass. 322 N. TaKAHASHI AND K. MlHAMA The Electropolishing of Aluminium. 324 A. W. Agar and R. S. M. Revell Zur Anwendung eines neuartigen elektrolytischen Poliergerates. 326 R. Zetzsche, E. Guyenot und H. J. Proger Llektronenoptische Untersuchungen an metallur- gischen Stauben, insbesondere deren Praparation und physikalische Differenzierung. 328 A. M. D'Ans und L. von Bogdandy The Electron Microscope in the Study of Wear. 331 D. Scott and H. M. Scott Anwendungen der Mikrotomschnitt-Technik auf elektronenmikroskopische Mineraluntersuchungen. 333 G. Prefferkorn, H. Themann und H. Urban The Use of a Freeze-drying Technique in the In- vestigation of Sodium-Montmorillonite by Electron Microscopy. 334 H. C. Corbet and J. Wolffes Zur Kenntnis der Glasoberfliiche. E. BkiJCHE und H. Poppa 336 Elektronenoptische Untersuchung natiirlichcr Opale in Verbindung mit ditVcrcntiallhermischen und ront- genographischen Studien Liber die Polytypic des SiO.,. 339 A. Maa.s A Reflection Electron Microscope Study of Dia- mond Cleavage Surfaces. 341 M. Seal The Direct Observation in the Scanning Micro- scope of Chemical Reactions. 343 J. H. L. McAusLAN AND K. C. A. Smith Uber das Teilchenwachstum sublimierbarer StofTe, dargestelll am Beispicl des Zinksulfids. 346 W. M Ciller und W. Jaenicke Electron-microscopical Investigations of Calcium Hydroxide and Calcium Carbonate. 347 G. SCHIMMEL Further Investigations of Photographic Develop- ment by Means of the Electron Microscope. 349 E. Klein Gelatin in the Photographic Process. 351 G. Vandermeerssche, C. Maertens and G. Lion The Microstructure of Photoconducti\e PbTe Layers. 352 A. Feltvnowski, J. Glass and L. Grelewicz Identification of Minerals Present in Mine Dusts by Electron Diflfraction and Electron Microscopy. 353 .1. H. Talbot ,<.^>^^^^ *'^^A/ M1C\-^"'" During four busy days, from the 17th to the 20th of September 1956, a conference on electron mi- croscopy was held at the Karolinska Institute in Stockholm. The conference was organized in about seven months by the Scandinavian Electron Mi- croscope Society at the request of the International Federation of Electron Microscope Societies. It represented the first European regional conference arranged under the auspices of this federation. The organizing committee consisted of the Com- mittee of the Scandinavian Electron Microscope Society with the addition of a few other members of that society. Financial support for the conference has been granted by the Swedish Government through the Minister of Education, Mr. I. Persson, and through the State Research Councils on Agriculture, Natural Sciences, and Technology. The members of the organizing committee express their appreciation for this important support. When the conference was opened by the Rector of the Karolinska Institute, Professor S. Friberg, it proved to have attracted 370 participants from 27 countries (see the list overleaf). The number of participants and of papers by far exceeded what had seemed reasonable to expect for a regional conference. In fact, with 176 papers read, this conference represented the largest meeting of electron microscopists ever held. It proved to be a difficult problem to squeeze this program into the time limits. An unavoidable consequence was to run mostly two and, temporarily, three parallel sessions. An extensive and representative exhibition of electron micrographs and of electron microscopes and accessories had been arranged. The participants had the opportunity of seeing the following types of electron microscopes in performance: Akashi Troncoscope, Philips EM 100 B and EM 75 B, RCA EMU 2 and EMU 3c, Siemens Elmiskop I, and the prototype of the new Zeiss electrostatic microscope. The tragic death of Professor Bodo von Borrics on July 17, 1956, overshadowed the conference. In commemorative addresses. Dr. V. E. Cosslett, Sec- retary of the International Federation of Electron Microscope Societies, and Professor E. Ruska, hon- oured his pioneer work in electron microscopy. On a proposal of Dr. J. Hillier, this conference will be referred to as the "Bodo von Borries Memorial Conference". The Proceedings of the Stockholm conference have been edited with two major aims: to have the proceedings in print in the shortest possible time, and to keep their size within reasonable limits. We have therefore had to cut down the editorial work, to exclude summaries, to restrict the space available for pictures, and to shorten the texts of several papers. We feel that a great deal more could have been done from an editorial point of view. These Proceedings are dedicated to the Memory of Bodo von Borries, in accordance with a proposal made by Dr. J. Hillier and by the decision of the Committee of the Scandinavian Electron Microscope Society. Stockholm, February 1957 Fritiof S. Sjostrand 1 — 568204 Electron Microscopy ORGANIZING COMMITTEE of the Stockholm Conference on Electron Microscopy President Dr. F. S. Sjostrand Secretary Dr. J. Rhodin Assistant Secretary Mrs. Elisabeth Johannesson Treasurer Mr. Ove Nilsson Members Mr. A. Birch-Andersen Prof. A. Engstrom Dr. H. Engstrom Prof. G. Glimstedt COUNTRIES REPRESENTED at the Stockholm Conference on Electron Microscopy Number of participants in parentheses Argentina (1) Australia (1) Austria (4) Belgium (5) Czechoslovakia (3) Denmark (9) Finland (4) France (31) Germany [Eastern] (13) Germany [Western] (99) Great Britain (40) Holland (20) Hungary (2) Ireland (1) Italy (11) Japan (2) Norway (3) Poland (1) Portugal (1) Soviet Union (8) South Africa (1) Spain (6) Sweden (72) Switzerland (12) Turkey (1) United States (17> Yugoslavia (2) BODO VON BORRTES 1905-1956 Am 17. Juli 1956 verschied unerwartet in seinem 52. Lebensjahr nach kurzer schwerer Krankheit der Pionier der Elektronenmikroskopie Prof. Dr.-Ing. habil. BoDO v. Borries. Der Verstorbene wurde am 22. Mai 1905 in Her- ford als Sohn des dortigen Landrats geboren und wandte sich — wohl durch sein mUtterliches Erbe bestimmt — der technischen Wissenschaft zu. Er studierte in Karlsruhe, Danzig und Munchen Elek- trotechnik und wurde 1930 wissenschaftlicher As- sistent von Prof. A. Matthias am Hochspannungs- institut der Technischen Hochschule Berlin, an dem eine von M. Knoll geleitete Arbeitsgruppe sich mit der Physik und Technik des Elektronenstrahl- oszillographen beschiiftigte. 1932 promovicrte er dort mit einer Arbeit Liber die ,,AuBenaufnahme am Kathodenstrahloszillographen". Wahrend seiner Doktorandenzeit begannen ihn die Aussichten zu fesseln, die sich damals der Forschung durch die Moglichkeit erolTneten, Elektro- nenstrahlen zur Mikroskopie zu verwcnden. Er wandte sich daher bald elektronenoptischen Unter- suchungen zu und betrat so in jungen Jahren den Weg, auf dem er bis zuletzt in uncrmiidlicher wissen- schaftlicher, technischer und organisatorischer Tiitig- keit so fruchtbare Arbeit leisten solltc. Die zu jenerZeit von verschiedener Seite auf dem Gebiet der Elektronenmikroskopie erzielten — zumindest ermutigenden — theoretischen und experimentellen Ergebnisse hatten noch kcineswegs vermocht, die Bedcnken und psychologischen Hcmmungcn zu iiberwinden, die sowohl bei Phy- sikern als auch bei anderen das Lichtmikroskop benutzcndcn Forschcrn gegenijber der neuen Mc- thodc bestanden. Dadurch war es zuniichst nicht moglich, die erfordcrlichen erheblichcn Mittel fur einc intensive Entwicklung dieses Gebietes von wis- senschaftlicher odcr industricller Seite zu crhalten. B. V. BoRRiLS hat schon fruh die grol3c Bedcutung der Elektronenmikroskopie mit klarem Blick erkannt und sich in Wort und Schrift mit seiner ganzen iiberzeugenden Pcrsonlichkcit fiir ihre Forderung eingesctzt. Ein erster entscheidender Erfolg war diesen Bemuhungen beschieden, als das Haus Siemens ihm und dem Referenten Ende 1936 die Moglichkeit Bodo von Barries bot, Elektronenmikroskope fiir die Praxis zu entwickeln. Unsere 1937 bei Siemens beginnende gemeinsame Arbeit wurde bald darauf noch durch H. RusKA unterstiitzt, der als Mediziner die sich abzeichnende Erweiterung der mikroskopischen Moglichkeiten schon seit Jahren verfolgt hatte. Als Frucht dieser Zusammenarbeit konnte das Friah- stadium der Experimente und Versuchsanordnungen iiberwunden und Ende 1939 das erste serienmaBige Elektronenmikroskop in einem Forschungslabora- torium der IG-Farben in Hochst in Betrieb genom- men werden. Neben der Vervollkommnung des Durchstrah- lungsmikroskops widmete sich v. Borries besonders auch der Verbesserung des Riickstrahlungsprinzips. Es gelang ihm dabei erstmals, Oberflachen von Metallproben mit besserer als lichtmikroskopischer Auflosung elektronenmikroskopisch sichtbar zu machen, wodurch er die bisher auf durchstrahlbare, d.h. sehr diinne Objekte beschrankte Elektronen- mikroskopie aufs glijcklichste erganzte. In den letz- ten Kriegsjahren begann er dann — ankniipfend an seine friihen Arbeiten uber die Elektronenemp- findlichkeit photographischer Flatten — mit syste- matischen Untersuchungen der energetischen Wech- selwirkungen im Elektronenmikroskop, mit denen er sich im Friihjahr 1945 an der Technischen Hoch- schule Berlin habilitierte. Fiir seine Verdienste erhielt v. Borries 1941 die Silberne Leibnizmedaille der PreuBischen Akademie der Wissenschaften. Als wahrend der Kriegsjahre die Fortfuhrung der elektronenmikroskopischen Arbeiten wegen der fiir dringlicher gehaltenen Entwicklungen und Fabrika- tionen immer aufs neue in Frage gestellt wurde, kampfte er mit erfolgreicher Hartnackigkeit fiir das Bestehenbleiben dieser Arbeitsstiitte, so daB bis Februar 1945 etwa 35 Elektronenmikroskope in wissenschaftlichen und industriellen Forschungs- instituten aufgcstellt werden konnten. Das Kriegsende verschlug ihn mit seiner Familie und einigen Berliner Mitarbeitern in seine west- falische Heimat. Dort legte er die bisher erarbeiteten Kenntnisse und Erfahrungen in dem 1949 im Verlag W. Saenger erschienenen Buch ,,Die Ubermikro- skopie" nieder. In diesem Werk untersucht er ins- besondere die Leistungsgrenzen der Elektronen- mikroskopie, welche durch die energetischen Ver- hiiltnisse in der Elektronenstrahlquelle, am Objekt und auf dem Leuchtschirm sowie auf der photo- graphischen Platte gegeben sind. Fine Kliirung dieser Fragen ist fiir die Elektronenmikroskopie von groBer Bedeutung, weil sic nicht nur die Ent- stehung elektronenoptischer Bilder verstandlich macht und ihre Deutung und Auswertung ermoglicht, sondern weil sie auch erkennen liiBt, auf welchen Wegen man die Mikroskopie mittels Elektronen- strahlen bis zur Sichtbarmachung einzelner Molekiile bzw. Atome vervollkommnen kann. Mit Tatkraft und Umsicht setzte er sich in den schwierigen Nachkriegsjahren fiir die Weiterfuhrung der elektronenmikroskopischen Forschung ein. Auf seine Initiative wurde 1948 die ,,Gesellschaft fiir Ubermikroskopie e. V. zu Diisseldorf" ins Leben gerufen, der die Ministerien fiir Kultus und Wirt- schaft des Landes Nordrhein-Westfalen und ver- schiedene industrielle Unternehmungen angehorten. Diese griindete und betrieb dann das .,Rheinisch- Westfalische Institut fiir Ubermikroskopie", das er als Direktor leitete. Aus diesem Institut, das sowohl der Weiterentwicklung des Elektronen- mikroskops selbst als auch seiner Anwendung auf mannigfachen Gebieten dienen sollte, sind von ihm, seinen Mitarbeitern und den wissenschaftlichen Gasten viele wertvolle Arbeiten verofTentlicht wor- den. Sein personliches Interesse wandte sich ins- besondere der Berechnung und Konstruktion von permanentmagnetischen Elektronenlinsen zu, bei deren Verwendung die bisher erforderlichen Strom- quellen hoher Konstanz fortfallen. Er entwickelte in seinem Institut mit neuen Konstruktionsideen ein leistungsfahiges magnetostatisches Elektronenmi- kroskop. Spiiter wurde von ihm und seinen Mit- arbeitern ein Ultramikrotom neuer Bauart verofTent- licht, das sich inzwischen ebenfalls gut bewiihrt hat. Schon friih hat sich B. v. Borries mit der Frage auseinandergesetzt, auf welchen Wegen eine inten- sive und universelle Auswirkung der Elektronen- mikroskopie in der naturwissenschaftlichen Grund- lagenforschung und in ihren speciellen Anwendungen zu erreichen ware. Hierbei war ofFensichtlich der zu groBe Aufwand fiir die Beschaffung und den Betrieb eines Elektronenmikroskops hinderlich. Er hat daher versucht, rationellere konstruktive Losungen fiir die Geriite zu finden und so zu einer Senkung der Kosten beizutragen. Notwendig war es auch, an mikro- skopischen Forschungen interessierte Wissenschaft- ler verschiedener Fachgebiete zusammenzubringen, um die Probleme der bis in die GroBenordnung der Molekiile und Atome vordringenden neuen Mikro- skopie zu erortern und ihre Einfiihrung und breite Anwendung durch stiindigen Erfahrungsaustausch zu fordern. Ihm ist es an erster Stelle zu danken, daB nach dem Krieg die Deutsche Gesellschaft fiir Elektronenmikroskopie gegriindet wurde, die er — zuerst als verantwortlicher Schriftfiihrer und dann als Geschaftsfuhrender Vorsitzender — umsichtig und verantwortungsbewuBt bis zu seinem Tod geleitet hat. Mit dem Ziel, die Veroffentlichung elektronenmikroskopischer Ergebnisse starker als bisher in einem hierfiir besonders geeigneten Organ zu konzentrieren, hat er sich erfolgreich mit dafiir eingesetzt, daB die Zeitschrift fiir wissenschaftliche Mikroskopie und mikroskopische Technik, die nach dem Krieg — wie viele andere — aus wirtschaftlichen Griinden ihr Erscheinen einstellen muBte, wieder fortgefuhrt und durch einen elektronenmikrosko- pischen Teil erweitert wurde. Auf vielen in- und auslandischen Kongressen hat er die Interessen un- seres Faches vertreten und die internationale Zu- sammenarbeit gefordert. Bei der Grundung der Bodo von Borries International Federation of Electron Microscope Societies 1954 in London wiihlten ihn die Delcgierten der auslandischen Gesellschai'ten in Anerkennung seiner wissenschaftlichcn und organisatorischen Leistungen und Faliigkeilen zum Priisidenten. B. V. BoRRits tuhlte sich in erster Linie als Inge- nieur, und in der Tat entsprangen seine Leistungen der Synthese von wissenschaftlichem, technischem und wirtschaftlichem Denken und SchafTen. Seine Liebe gait ganz besonders dem Konstruiercn. Er wurde 1949 als Honorarproi'essor an die Medi- zinische Akademie zu Diisseldorf und 1953 als ordentlicher Professor auf den neugeschatrenen Lehrstuhl fiir Elektronenoptik und Feinmcchanik der Technischen Hochschule Aachen berufen und konnte so — wenn auch nur noch wenige Jahre — seine reichen Erfahrungen als ein selbst begeisterter und daher mitreiBender Lehrer an die jiingere Generation weitergeben. Seinen Studenten und Mitarbeitern gab er nicht nur vielfiiltige Anregungen, sondern brachte ihnen auch ein warmes menschliches Interesse entgegen und half ihnen in personlichen Schwierigkeiten mit Rat und Tat. B. \. BoRKii s hat die Flektronenmikroskopie seit ihrem Entstchen uber 25 Jahre lang leidenschaftlich uiul unermiidlich gefordert. Seelischen Ausgleich fijr seine dauernde geistigc Anspannung suchte und fand er in oinem harmonischen Familienlcbcn. Mit groBer Liebc hing er an seiner Lebensgefahrtin und an seinen fiinf Kindern, denen er ein verstiind- nisvoller Vater imd Freund war. Er schiitzte und ptlegte die hiiusliche Gcsclligkeit und verbrachte auch auf den elcktroncnmikroskopischen Tagungcn im in- und Ausland mit den Fachgenosscn manchc frohe und beschvvingtc Stunden. Ls erfiilltc liiii dabei mit groBer Genugtuung, daB sich in so viclen Liindern der Kreis clektronenmikroskopisch ar- beitcndcr Kollegen von Jahr zu Jahr sichtlich vergroBcrtc. Wir Freunde und Kollegen bedauern zutiefst den so unerwartct friihen Tod dcs aktivsten Vorkiimpfers der Flektronenmikroskopie, dessen Pcrsonlichkcit und Lebensleistung uns unvergeBlich bleiben werden. Auch seine Schiiler und engcrcn Mitarbeiter werden ihren Lehrer und Forderer in dankbarer Erinnerung behalten. It is my sorrowful duty and privilege to recall to you the untimely passing, after a short illness and serious operation, of the President of our Interna- tional Federation of Electron Microscope Societies. It is sad, sad almost beyond words, that we must commence this meeting under the shadow of the death of the man who would have been presiding here, had he but lived. Professor von Borries had worked hard to prepare the way for this Congress — I believe it is correct to say that the idea of holding Regional Meetings, between the larger World Con- ferences, was his in the first place, and he was very much concerned to make a success of the first gathering of this kind. Could he have been with us this morning, I am sure he would have agreed that in attendance and volume of contributions it has exceeded his highest expectations. Bodo von Borries has been one of the chief founders of electron microscopy in the strictly scien- tific as well as in the organisational sphere, it should be known to even the youngest of our members how he was successful, with Dr. Ernst Ruska, in obtaining in 1932 the first transmission pictures with an electron microscope. From that time on. the collaboration of von Borries and Ruska continued the development first of the electron microscope in the research laboratories of the Technical University of Charlottenburg and then of the original model for the firm of Siemens and Halske. He occupied himself with the applications as well as the design and operation of the instrument, and was especially interested in explaining its principles and potenti- alities to wider scientific and semi-scientific audiences. After 1945 he had a long and hard struggle to get a laboratory established again for independent re- search in the subject. It is a remarkable tribute to his vision and ability that he finally successfuli\ formed and financed, almost entirely b\ his own efforts, the Rheinisch-Westfiilische Institut fur Uber- mikroskopie in Diisseldorf. This must be almost unique of its kind in the world, being, so far as I know, the only independent research centre for electron microscopy, financed on a very wide basis from state, industrial and philanthropic sources, and established and directed by one man. And at the same time he was actively promoting the organi- sation of electron microscopy in Germany and on the international level. Others can speak with better knowledge than myself of the part he played in setting up and carrying on the Deutsche Gesellschaft fur Flektronenmikroskopie, but it is evident that that Society has largely bcL-n inspired by him and without his constant activity on its behalf would not now be the large and influential body that it is. He was its main founder, and at his death was its executive President. I mention almost as an aside that he was Professor vi\ electron microscopy at the Technische Hochschule at Aachen, at the same lime Bodo von Borries as he was directing the Institute in Dusseldorf and fulfilling a number of other functions in educational and scientific bodies. In the international sphere von Borries was also most active, and deeply concerned to establish good scientific relations with electron microscopists in other countries, despite the deep wounds caused by the War. He visited us in Britain in 1948 and ever since that time has worked hard to set up an effective international organisation for electron microscopy, and to ensure that his own country played a full part in it. I need not remind you that he was elected first President of the Joint Commission for Electron Microscopy that was set up at the London Confer- ence in 1954. When that body was in effect vetoed by the International Council of Scientific Unions he was most concerned that all the ground work put into it should not go wasted and that some viable organisation should be established as quickly as possible. He played a major part in the transfor- mation of the Joint Commission into the Interna- tional Federation of Electron Microscope Societies, and was its first President, in this capacity he worked unceasingly for the extension of the Federa- tion's activities, for the success of the Regional Conferences being held for the first time this year, and in preparing for the next World Congress in 1958, which was to be held in his country. At the time of his death he was planning to visit Japan during October, to attend the Regional Congress for Asia and Oceania, at which he was to have given the opening address. By his untimely end. in the midst of full creative activity, the young science of electron microscopy has lost one of its greatest exponents and indeed one of its two original pioneers. Germany has lost the founder and mainspring of its electron micro- scope society and the International Federation its first president and chief advocate. I commend his career to you as an outstanding example of service to science in its fundamental, applied, educational and organisational aspects equally. Let us honour his memory by continuing to advance the aims and ideals he followed, particularly in developing the poten- tialities of electron microscopy in Science and tech- nology, nationally and internationally. We mourn the death, but we salute the memory of Bodo von Borries, pioneer of electron microscopy and first President of our International Federation. I INSTRUMENTATION Observation directe des surfaces metalliques par reflexion C. Fert Labor atoire d'Optiqiie Electroniqiie, Toulouse Au cours des dernieres annees, I'observation directe des surfaces par reflexion d'electrons a donne lieu a diverses recherches, et plusieurs communications de ce congres sont consacrees aux applications de cette methode. Je desire presenter ce probleme du point de vue du physicien, en insistant sur les faits nouveaux depuis le Congres de Londres (1954). Generalites. — Dans un microscope electronique par « reflexion », Tobjet est « eclaire » par un faisceau d'electrons monocinetiques. Les electrons diffuses dans la direction de Tobjectif servent a la formation de I'imagei. L'axe du faisceau qui eclaire I'objet est incline par rapport a Taxe de Tobjectif d'un angle d^ -r 0, (fig. 1). L'objet est eclaire sous un angle 0^ (angle d'eclaire- ment). La surface de Tobjet fait avec l'axe de I'objectif un angle 6. (angle d'observation). Angle 02 et distorsion de I'image : L'image qui se forme sur I'ecran est celle de la projection de la surface de l'objet sur le plan de front conjugue de I'ecran. Soit Mi le grandissement. Si un quadrillage carre de cote a est trace sur ^objet^ l'image de chaque carre est un rectangle de cote Mi a et M^ sin Oo a. On parle de deux grandissements caracte- ristiques Mj et M. = Mi/sin 6.,. On peut dire que le rapport M2/M1 = sin d.^ caracterise une « distorsion » de l'image. Angle 61 et sensibilite au relief : Nous verrons que 6*1 est toujours petit devant (I. ■ I'eclairage est tres rasant. Les asperites de l'objet portent ombre sur sa surfaced Si 0^ est tres petit, les ombres sont tres l=h sin (Oi + gg) sin Oi allongees, la sensibilite au relief tres bonne : une asperite de quelques millimicrons est mise en evi- dence. Cette cause de contraste est preponderante devant toutes les autres, ce qui distingue nettement la microscopic electronique par reflexion d'autres methodes d'observation directe des surfaces (micros- cope Electronique a emission par exemple). Influence de 0^ et d, sur la resolution : Micros- copic electronique par reflexion sous grand angle. — Jusqu'en 1954, la seule technique utilisee etait celle de ['observation rasante* : 0., petit, de I'ordre de 4 a 6°, M1/M2 de I'ordre de 10 a 15; dans ces condi- tions, la distorsion des images est tres genante, mais il etait admis que si 0, etait plus grand la dispersion des vitesses des electrons diffuses devenait importante et la resolution mauvaise^ Canon a electrons y*^ Condenseur Canon a ions Objet Objectif Diaphragme d'objectif Diaphragme de selection Lentille intermediaire (Systeme dispersif) r Projecteur Groupe quadripolaire (anamorphose) Ecran fluorescent Emulsion ^ ^ n r 6^ Binoculaire Fig. 1. Direct observation of solid surfaces by reflection. Apparatus. Nous avons cherche a determiner experimentale- ment I'influence reelle de 62. A notre surprise, I'experience a montre que, si l'objet est tres propre, tres peu corrode et le faisceau incident rasant (0i petit, 1 a 3 par exemple), la resolution ne varie pas, pour une ouverture donnee, si 02 augmente. Elle est, par exemple, voisine de 300 A pour une ouverture 2a =5 10"^ radians, pour la plupart des echantillons examines. ^ Le premier essai demicroscopieelectroniqueparreflexion est du a Ruska (5). 2 On suppose que un cote est oriente perpendiculairement au plan d'incidence. ' La longueur apparente, en projection sur le plan de front conjugue de Tecran, de Fombre d'une asperite de hau- teur Il est * Voir par exemple : Dupouy et Fert, Congres de Londres (1954). La premiere publication sur cette technique est celle de von Borries (1). * Dans le microscope electronique par retlexion, c'est I'aberration chromatique de Tobjectif qui limite la resolution. Observation directe des surfaces metallk/ues Kig. 2. White pearlitic cast-iron etched by cathode sputtering. Experimental conditions: Oj =^ 2°, 02= 23 : ratio ol'theciiarac- teristic magnitkations: 2.5. Objective focal length/^ 3.5 mm. Objective diaphragm: 2 r = 30 microns. Direct magnification: 1800. Exposure time: 15 seconds. This photograph has been obtained by joining together three photographs which were taken with slightly dilTerent focusing, corresponding to a knov\n variation ofthe resistance \\hich is placed in series u ith the objective. Autrement dit : si 0^ reste petit, la dispersion des vitesses des electrons difTuses n'augmente pas sen- siblement avec 0,.' L'emploi de grands angles d'observation (0. 15 ) ^ Pour preciser ce resultat, nous faisons actuellcment une etude experimental directe de cette dispersion en fonction de Oi et 02 par spectrographie des vitesses electroniques; I'element dispersif (en pointille sur la fig. 1) est monte sur le microscope electronique par reflexion, el permet d'obtenir le spectre des vitesses des electrons diffuses par une region choisie de I'objet. - Diverses consequences resultent de Temploi de grandes valeurs deSg • bande de nettcte plus large, sensibiliteau relief plus grande etc... Nous n'insistons pas ici sur ces resultats, qu"il est facile de prevoir (4). change completement Taspect des images. Nous avons principalcmcnt utilise des valeurs de Oi •- Sj = 25 , soit 0., ly. La figure 2 montrc une image obtcnuc dans ces conditions. Le rapport A/, A/.., caractoristiquc de la distorsion des images, est A/, M. 2.5 au lieu de 15 pour 0., - 4 .- Reduction de la distorsion des images par une leniille correctrice. — Pour rcduire encore davantage cette distorsion (quel que soit 0,), nous avons en outre utilise I'artifice suivant. Un systeme a symctrie non axiale est dispose apres Ic projecteur. Son role est d'augmenter le grandisscment dans le plan d"incidence, et de le reduire dans le plan iicrpcndiciilairc (3). 10 C. PERT Fig. 3. Correcting lens operating. The specimen is an eletrolytic deposit of Nicivel on cast-iron. The two parts of this figure represent the same region of the specimen with and without correcting lens. Observation directe des surfaces metallic/ues 11 Fig. 4. Austenitic steel, di-r 6^= 25°: with correcting lens, M^j Mi= 1.3. Nous avons obtenu d'abord ce resultat avec une lentille electronique spherocylindrique. Nous utili- sons maintenant un groupe quadrupolaire magne- tique (fig. 1) convenablement oriente. Si / est le courant d'excitation des bobines de ce systeme, le rapport Mi M, devient^ : Ml Mo 1 1 CO I sin 0, \ + coi (o), coefficient caracteristique du groupe quadrupo- laire). Les figures 3 a 5 ont ete obtenue dans ces conditions, pour 6, = 23° (Fig. 3), 0 + 0 = 25° (Fig. 4), 0 12 (Fig. 5). Cet artifice reste evidemment valable pour des valeurs plus petites de 0.. qui prescntcnt Tavantage de donner des images mieux eclairees, ou de per- mettre la micro diffraction. II facilite toujours bcau- ^ L'expression de la longueur appurcnte d'Line ombre portee prend la forme : l=h sin {(Ji + Oi) 1 -.oji sin Oi I -co i "impression de relief est augmentee. coup Tobservation des images ou rintcrprctation des photographies. Description de I'appareil. — La figure I precise le montage. On notera : Le canon a ions: Ic bombardemcnt ionique de Techantillon evite sa contamination (2). Le diaphragme de selection qui limite la region de I'objet si on veut obtenir soit le diaphragme de diffraction de cettc region (0., petit), soit le spectre des vitesses des electrons diffuses par cette region. Le groupe quadrupolaire corrccteur. L'ecran fluorescent incline qui assure une cor- rection complementaire de la distorsion de Timage, dans Tobservation visuelle. Cet ecran est relalive- ment c'loii;ne de I' emulsion afin d'observer une image a grandissement reduit, done plus eclairee. Nos efforts se poursuivent actuellement atin d'ame- liorer la resolution ct reclairemcnt des images par : (a) Temploi d'une difference de potent iel accele- ratrice plus elevee, afin d'augmenler la hnllance de la source, et, peut-etre, de reduire A V/V. (h) i'emploi de systemes correcteurs de Taberra- tion chromatique. a symetrie non axiale. Conclusion. — La microscopic electronique par reflexion est possible pour une tres large gamme des valeurs de Tangle d"observation, a resolution sensi- 12 V. E. COSSLETT AND P. DUNCUMB Fig. 5. White pearlitic cast-iron. Oj + 0, ' magnification: 1000. 14'', 02= 12', with correcting lens, M^l M.,= 3. Exposure time: 3 seconds. Direct blement constante. Elle presente une sensibilite au relief comparable a celle des methodes optiques in- terferentielles ou par contraste de phase, tout en ayant une limite de resolution dix fois meilleure en- viron, dans les conditions actuelles. Ces caracteristi- ques essentielles, fixeront, sans aucun doute. le domaine propre d'utilisation de la microscopic eiec- tronique par reflexion dans Tobservation directe des surfaces. BiBLIOGRAPHlE 1. VON BoRRiES, B.,Z. Physik 116, 370 (1940). 2. Pert, Ch., Compt. rend. acad. sci. 248, 333 (1954). 3. Pert, Ch. et Marty, B., Compt. rend. acad. scl. 241. 1454 (1955) 4. Pert, Ch., Marty, B. et Laporte, R., Comptes rendus du colloqiie CNRS a Toulouse, Avril 1955. 5. RusKA, E., Z. Pliysik 83, 492 (1933). A Scanning Microscope with either Electron or X-Ray Recording V. E. CossLETT and P. Duncumb Cavendish Laboratory, Cambridge 1 HE purpose of the scanning microscope is to form an image of a surface either by electron scattering or by x-ray emission, and to analyse the elements in a selected volume of about one cubic micron in the surface by the characteristic x-ray lines emitted. A block diagram of the apparatus is shown in lenses into a small spot on the specimen, using a similar technique to the x-ray projection microscope (4, 6), and is scanned over the specimen by the deflec- tion coils. These coils are similar to those used by McMullan and Smith in their scanning electron microscope (5, 7), and give the beam a double deflec- fig. 1. The electron beam is focussed by two magnetic tion balanced so that the beam always goes through A Scanning Microscope with Electron or X-Ray Recording 13 DISPLAT r l— «nPLIFIER g - ELECTRON! GUN MftGMETIC LENSES CRYSTAL SPECTROMETER 'MAY Be INSERTED AMPLIFIER PULSE ANALYSER Fig. 1. Block diagram of the scanning microscope. the lens aperture. The spot size is 0. 1 /< to 1 // and the area of scan about \ mm square at largest. Part of the emitted x-rays are collected through a window in the polepiece gap by a scintillation counter, or alternatively the phosphor of the counter may be pushed into the vacuum to record high energy scat- tered electrons. The amplified signal from the counter modulates the brightness of a cathode ray tube scanned in synchronism with the microscope beam, so that an image is obtained of the variation of scattered electron intensity or of x-ray emission over the surface. The scan may then be stopped and the micro- scope beam accurately positioned on any feature in the specimen by observing the spot on the afterglow of the picture on the display tube. Through another window part of the emitted x-rays pass into a crystal spectrometer for analysis of the emission spectrum from that point, as in the method of microanalysis developed by Castaing (I, 2). Alternatively, the crystal can be removed from the spectrometer so that the x-rays pass straight into a proportional counter, which gives for each quan- tum a pulse of height approximately proportional to its energy. A single channel pulse analyser can be used to pass only pulses corresponding to a given characteristic line and, with the beam scanning, these can modulate the display tube so that the distribution of the element emitting that line is shown up. The second lens is shown in fig. 2. The electrons are focussed on to the specimen which is level with the lower pole face, and x-rays pass through a beryllium window into the phosphor of the scintilla- tion counter just outside the gap. The light generated is transmitted by a perspex rod to the photocathode of the multiplier outside the lens. The electrons scattered from the specimen are deflected by the lens field, but may be collected by removing the window and pushing the phosphor into the polepiece gap. The signal from the photomultiplier with elec- tron recording is several hundred times that with x-ray recording with other conditions unaltered. Through another window in the polepiece gap x-rays pass into a hydrogen filled tube and into the crystal spectrometer or proportional counter. The 1 SPECIMEN MOVEMENTS Fig. 2. Final lens. specimen can be moved laterally and vertically or rotated and is insulated from earth in order to meas- ure the incident electron current. The focal length of the lens is about 4 mm; this is slightly short- ened by a thick ferromagnetic specimen but the spot is not distorted. Fig. 3 shows an area of 1500 mesh/inch silver grid as photographed from the display tube with electron collection in A and x-ray in B. The accelerating volt- age was 25 kV giving an electron penetration in the specimen of about 1 //. The topography of the bars is shown up with electron collection (A) but not with x-ray collection (B) unless the irregularities are more than a few microns in size. With x-ray collec- tion the resolution can be no better than the electron penetration since x-rays are emitted throughout the volume of electron diffusion, but with electron collec- tion the resolution is well below this. Smith reports a resolution of about 200 A with his scanning microscope (7). The particular advantage in using x-rays to form the picture lies in being able to show up the distribu- tion of one element by using a proportional counter and pulse analyser. For example, with a specimen of silver and copper grids it is possible to make either one appear the brighter by selecting the CuK A B Fig. 3. 1500 mesh inch silver grid, bars .^ /( and I \ /i wide. A, electron collection; B, x-ray collection. 14 B. V. BORRIES ti G. LANGNER AND W. SCHEFFELS D Fig. 4. Impurities in beryllium foil, x-ray collection. A, all radiation recorded; B. C, D, characteristic emission from dif- ferent impurities separately selected — manganese, nickel and calcium respectively. or AgL characteristic radiation to form the picture. The distribution of the copper can thus be shown in one picture and the distribution of the silver in the other. Another way of displaying this same result is to regard the two pictures as components of a colour picture and photograph them through different filters in register on the same piece of colour film. In this way the copper was made to appear red and the silver green. In principle one can attach a given colour to a given x-ray wavelength, which may be a useful technique for presenting in one picture the relative positions of several elements. As an example of the identification of inclusions in a surface, fig. 4 shows some impurities in a piece of beryllium foil (3). In A all the emitted x-rays were used to form the picture and the impurities show up as emitting more strongly than the beryl- lium. In 5, C, and D the pulse analyser was set to pass the characteristic radiation from three different types of impurity in turn, which were identified as manganese, nickel and calcium respectively. The in- formation may again be presented on one picture in colour. The energy resolution of the proportional counter is not good enough for the separation of character- istic radiation from elements which differ by less than about four in atomic number. However, it seems that there would be sufficient intensity reflected from a curved crystal in the spectrometer to form the picture, in which case adjacent elements would easily be separable. The instrument, therefore, can show up either the topography of a surface by electron scattering with a resolution of well below I //, or the distribution of an element over a surface by x-ray emission with a resolution of about I //. A selected point in the surface can then be analysed quantitatively by plot- ting the characteristic x-ray emission spectrum by means of a crystal spectrometer. References 1. Castaing, R., Thesis. Paris, 1951. 2. Castaing, R. and Descamps, J., J. pliys. 16, 304 (1955). 3. COSSLETT, V. E. and Duncumb, P., Nature 177, 1172 (1956). 4. CossLETT, V. E. and Nixon, W. C, /. Appl. Pins. 24, 616 (1953). 5. McMuLLAN, D., Proc. Inst. Elec. Engrs., Pt. 1, 100, 245 (1953). 6. Nixon, W. C, Proc. Roy. Soc. A 232, 475 (1955). 7. Smith, K. C. A. and Oatley, C. W., Brit. J. Appl. Phys. 6, 391 (1955). Imaging Elements Operating with Permanent Magnets B. V. BoRRiES t, G. Langner and W. Scheffels Rheinisch-Westfdlisches Iiistitut fiir Vbermikroskopie, Diisseldorf Electron lenses operating with permanent magnets can be employed in electron microscopes, as several authors (I, 8) have shown. The advantage of such lenses is that no source for a highly stabilized lens current is needed. As the main disadvantage during operation must be named the impossibility to switch off the magnetic field. Moreover there must be at least two gaps in a system excited by permanent magnets to avoid stray field. This does not neces- sarily mean, however, that these gaps must act as separate lenses. Permanent magnetic einzel-lenses with variable focal length have been used as con- densor lenses by v. Borries (1954), the magnetic circuit of which has been calculated by v. Borries and Lenz (4) and the properties of which have been investigated experimentally by Langner (6). Also as Imaging Elemenls with Permanent Magnets 15 ^gS ^ ^m: Strahlnchtung 1, Permanent magnet. 2a, 2b, Disploceable polepieces of the intermediate lens. 3, Insert, at the same time fixed pole- piece at the final projector lens. 4, Disploceable polepiece of the fmal projector lens. 5, Control nut. 6, Bevel gear. Fig. 1. Cross section through a two-stage projective system operating with permanent magnets. an electron microscope objective lens a permanent magnetic einzel-lens may be used. On its electron optical properties Lenz (9) has published a paper. Such einzel-lenses and any lens system of two or more gaps operated by a single permanent magnet as described e.g. in a paper by v. Borries and Langner (4) show no image rotation, or, if the specimen im- merges in the field as is the case in strong lenses, the image rotation is at least small and does not vary noticeably when the magnification is changed. The size of such a system depends on the niagne- tomotoric force of the permanent magnet. The opti- mum shape of a permanent magnetic lens system has been calculated by v. Borries and Lenz (4), and it can roughly be said that for the diameter D of such a system, provided the best obtainable Alnico magnets with at least 4.5 10'' Gauss-Oersted are used, the following simple rule holds: Z) = 7 10 - mm Amp ' /, where / is the number of ampere-turns the magnet supplies when in the system. Based on the design of what we think an optimum system for an electron lens system with rotational symmetry three perma- nent magnetic electron microscopes have been com- pleted with / =i 2500 Amp on which papers have been published by v. Borries (1,2). Recently we have built a two-stage projector system of / = BOO^.. VergroOerung lC% P loy. 5% 10 75 — » ! mm ausgenulzler Bereich 20 Fig. 2. Magnification and distortion of a two-stage perma- nent magnetic projector system. 1720 Amp, PU = 60 Amp- Volt for 50 kV, for the use with an immersion objective. The cross section is shown in fig. I . The magnet is outside the vacuum. The intermediate lens is formed by the polepieces 2a and 2/?, the axial displacement of which permits the magnification of this stage to be varied. The properties of such lenses have been described earlier (6, 7). A rather large range of magnifica- tion is achieved at the cost of considerable radial distortion. The final projector has therefore been designed as a lens with a variable gap-w idth which yields less distortion but also less variation of magni- fication. Fig. 2 illustrates magnification and radial distortion versus the displacement of the polepieces. The magnification can be continuously varied in the ratio I :20 while the specimen is under observation. Furthermore we have enlarged the above-men- tioned system and used a magnet with 3250 Amp. With this two-stage microscope a magnification of 2 axial displacemcnl of an iron piece 2 the second gap opens. So the back focal plane of the first gap can he imaged to obtain tiiree-stage difiVaction diagrams. In the first gap we use a magnetic stigmator 3 which is schematically shown below. Two pieces of iron slide within slotted pieces, so the a/imuth and the streiigtli of the stigmator can be controlled separatel>. The projector consists of a polepiece unit 4 which is also axially displaceable and permits a variation of the magnification of 1:11. New in respect to for- mer instruments of that kind is the separate control of the magniticalion in the projector and in the objective. Owing to this it is e.g. possible to image 16 B. V. BORRIES t, G. LANGNER AND W. SCHEFFELS ijjjj' "S*. J / = 3250 Amp., (J = 60 kV. Continuous variation of magnifi- cation between the three stage diffraction patterns and Amax = 28 000. Fig. 3. Cross section of a permanent magnetic microscope system. three-Stage diffraction patterns with different magni- fications. It is also possible to use the whole system as a diffraction condensor, and also to use it as an elec- tron shadow microscope. If crystalline matter is put into the caustic, dynamic diffraction patterns can be obtained which resemble Kikuchi lines. Such patterns from MgO crystals are shown in fig. 4. Another problem in the field of permanent mag- netic lenses which we have investigated is the so-called "reflexion microscopy" which is really a dark field microscopy of surfaces hit by the electron beam in grazing incidence. In order to avoid immer- sion the focal length must be considerably greater than half the gap width in order to give space for the axial movement of the specimen. On the other hand the chromatic aberration constant C< should be as small as possible because the mostly inelastically scattered electrons which are used to image have suffered considerable energy losses AE. If we put for a moment C- «^/ (/- focal length), and d be the diameter of the aperture stop, we obtain for the resolving power d> / C eU* cU* Fig. 4. Kikuchi lines of MgO crystals obtained in the caustic of a strong permanent magnetic lens. when using permanent magnetic lenses with the rela- tivistic accelerating voltage U*. The image brightness goes with the 2nd power of a, so using half the focal length one should get four times the brightness in the image at the same resolution and magnification. The requirement of non-immersion objective lenses sets a limit to the lens excitation. We have tried an objective lens with h = 1 mm bore, .v = 0.8 mm gap-width,/ 1.45 mm, C = 1-35 mm which needs / - 1450 Amp at 60 kV. In a two-stage permanent magnetic system this demands the same low excita- tion also for the projector which then is likely to cause distortion. This is a difficulty which arises when using permanent magnetic microscopes for "reflexion microscopy". This difficulty can be over- come by using two-gap objective lenses, over which the effective ampere turns of the magnet are dis- tributed so that the objective lens gets a suitable low excitation and the series gap takes the rest and is used as an intermediate lens. The latter, however, may also cause some distortion. We leave now the permanent magnetic lens systems with rotational symmetric lenses, and turn to ele- ments with less symmetry. For "reflexion micros- copy" normally the condensor is tilted. It is also possible to deflect the beam so that its angle of in- cidence is suitable for "reflexion microscopy". K. Ito and T. Ito (5) have used deflection coils. We have tried to use small permanent magnets. For small deflection angles [i the following relation holds Magnetostatische Linsenanordnungen mil Dwchanischen Regelglicdern 17 Fig. 5. Reflection units using small permanent magnets. ^ = 2 mo U* BAz)dz. Here c is the charge of an electron, /?/„ its mass, Bx the component of the induction normal to the plane of motion. - measures in the initial direction, the field extends between, r„ and r,. Fig. 5d illustrates the principle of a deflection unit working with small flat AInico magnets. Fig. 5a shows the unit assembled and Fig. 5/) the magnets, the magnetic link and the moimt. With these magnets an angle of deflection of I 1 can be obtained. It may only be mentioned here that two permanent magnets of this type can be arranged in a way that they form a magnetic quadrupole and so have focus- sing properties. Lenses of both positive and negative focal length in the order of 5-6 mm could be realized with the small magnets described above. The very preliininary experiments were carried out with the help of a device shown in fig. 5c, which has two or four degrees of freedom. For many fruitful discussions the authors wish to thank Dr. Lenz. The technical designs of the lens systems lay in the hands of Ing. J. Huppertz. References 1. VON BORRIES, B., Kolloid-Z. 114, 164-167 (1949). 2. — Z. wiss. Mikioskop. 60, 317-358 (1952). 3. VON BoRRiES, B. and Lanuner, G., Comples rendus du colloque CNRS a Toulouse, 4.-8. April 1955, pp. 285-295 (1956). 4. VON BoRRiES, B. and Lenz, F., Optik 13, 264-276 (1956). 5. Ito, K., Ito,T. and Watanabe, M., /. Electron-microscopy (Japan) 2, 10-14 (1954). 6. Langner, G. Optik 12, 554-562 (1955). 7. Langner, G. and Lenz , F., Optik 11, 171-180 (1954). 8. Reisner, L H. and Dornfeld, E. G., /. Appl. Phys. 21, 1131-1139 (1950). 9. Lenz, F., Z. ong. Plus. 8, 492-496 (1956). Uber magnetostatische Linsenanordnungen mit mechanischen Reeeleliedern^ K. MiJLLER Siemens & Halske AG, Wernerwerk fiir Mefitechnik, Berlin-Siemensstadt IM Rahmen einer Entwicklung magnetostatischer Elektronenlinsen haben wir vor allem die Moglich- keiten studiert, die Brennweite solcher Linsen zu regein, wie es zur Scharfstellung des Bildes oder zum Wechsel des AbbildungsmaBstabes notwendig ist. Die Brennweite einer magnetischen Elektronen- linse hangt von drei GroBen ab: von der Strahlspan- nung, der Maximalfeldstiirke im Linsenspalt und von der Halbwertsbreite des Linsenfeldes. Die Anderung einer dieser drei GroBen bewirkt eine Regelung der Brennweite, doch sind nicht alle GroBen in gleicher Weise geeignet: Strahlspannungsiinderungen beein- flussen in meist unerwiinschter Weise das Durch- ^ Eine ausfiihrlichere Veroffentlichung crschcini dem- nachst in Z. wiss. Mikroskopie. 2 — 568204 Electron Microscopy dringungsvermogen der Elektronen. Man wird diese Moglichkeit der Brennweitenregelung deshalb hoch- stens in einem schmalen Bereich anwenden, der zur Feinfokussierung des Bildes geniigt, dagegen fur Grobscharfstellung und VergroBerungswechsel die Anderung von Maximalfeldstarke und Halbv\erts- breite vorziehen. Die Halbwertsbreite des Feldes ist — wenn man von Siittigungserscheinungen absieht — allein eine Funktion der Linsengeometrie. also des Polschuhabstandes und des Bohrungsdurchmes- sers. Sie kann daher durch das Auswechseln von Polschuhen oder durch Verschieben der beiden Pol- schuhe einer Linse gegeneinander veriindert werden. Das Auswechseln schien uns technisch unbequem:die Moglichkeit einer Polschuhverschiebung gefahrdet eine exakte Zentrierung der Linsen. Wir wahlten deshalb den dritten Weg: die Regelung der Feld- 18 K. MULLER ,5iJ: [\\\\\\\\\.\ As\';,'.^\'q' Ps m I I Tn^ t^tv\vjjt^\v ^1 ^ t -^ 5^M ^ A\\V\\\\31 ^^ tssssszzztx: ^ Eis»nschluO Luftspalt wo^ \^ 50 H *^ Drrhwinkel — W 2^ 30° ~*d' SO' Abb. 1. Magnetostatische Linsenanordnung mit Zackenreg- ler. a) Magnetische Anordnung; b) elektrisches Ersatzbild: c) Zackenregler; d) Feldstiirke-Regelkurve. K^ 50 H E \ *°"^ 1 Regelweg i. 20 40 c 60 Abb. 2. Magnetostatische Linsenanordnung mit Innen- scheibe. a) Magnetische Anordnung; b) elektrisches Ersatz- bild; c) Feldstarke-Regelkurvc. Starke durch Anderung der magnetischen Spannung am Linsenspalt. B. v. Borries und Mitarbeiter er- reichen eine solche Anderung z. B. durch wechselnde Aufteilung einer konstanten Spannung auf zwei Linsenspalte (z. B. G. Langner 1955). Es gibt aber noch andere Moglichkeiten, das Linsenfeld bei starrer Polschuhanordnung durch die Bewegung von Eisenteilen zu beeinflussen: Abb. 1 zeigt eine magnetische Linsenanordnung, daneben das elektrische Ersatzschaltbild: Ein Per- manentmagnet moge zwei Linsenspahe (Z.i und L.) gegenpolig erregen. Dem Magneten entspricht eine Stromquelle mit dem inneren Widerstand R^. R^^ und /?i„ entsprechen den magnetischen Widerstan- den der Linsenspalte. Man kann die Spannung an Ri, nun z. B. durch Einfiigen eines Widerstandes R vermindern. Im magnetischen Kreis geschieht das am einfachsten durch einen Zackenregler (Abb. 1 c): Man verwendet zur FluBfLihrung zwischen Magnet und Linse ein Rad mit einer beliebigen Anzahl von Speichen, die so geteilt sind, daB das Rad in einen ruhenden Teil. den ,,Stator", und einen beweglichen Teil, den „Rotor", zerfallt. Bei Drehung des Rotors andert sich die GroBe der Beriihrungsflache zwischen Stator und Rotor und damit der magnetische Wider- stand zwischen Magnet und Linse. Infolgedessen andert sich die Feldkurve im Spalt der unteren Linse (Lo) gemaB der in Abb. 1 d dargestellten Kurve. Die etwas wechselnde Belastung des Magneten verur- sacht eine wechselnde Durchstromung des inneren Widerstandes /?;. Daher liiBt sich eine geringe RCick- wirkung des Regelvorganges auf die zweite Linse des Systems nicht vermeiden. Sie ist gegensinnig, d.h. bei Schwiichung der Linse 2 wird Linse 1 etwas st-irker. Abb. 2 zeigt einen anderen permanentmagneti- schen Kreis und sein elektrisches Ersatzschaltbild. Es sei diesmal eine Beeinflussung der oberen Linse (Li) erwiinscht. Dazu dient eine ringformige Regel- scheibe, die zwischen Magnet und Polschuhhalter axial beweglich ist und einen variablen Teil des Magneten zusatzlich belastet (Innenscheibe). Durch geeignete Dimensionierung kann man erreichen, daB beim Aufwartsbewegen der Scheibe der Weicheisen- kreis zwischen Magnet und Linsenspalt infolge der FluBerhohung in wachsendem MaBe gesattigt wird. Das entspricht einer Veranderung des Widerstandes Ri und bewirkt demzufolge eine Anderung der Feldstiirke in der oberen Linse. Auch diese Regelung erfolgt wegen der wechselnden Durchstromung des Widerstandes R^ mit geringer, diesmal gleichsinniger Rijckwirkung auf die zweite Linse des Systems. Gelegentlich legt man Wert darauf, daB beide Linsen in gleichem MaBe geregelt werden. Dazu eignet sich eine Ringscheibe zwischen dem Magneten und dem AuBenmantel des Systems (AuBenscheibe, Abb. 3). Bei dieser Anordnung ist die Anderung der Durchstromung von /?,, die bei einer Axialbewegung dieser Scheibe auftritt, keine Nebenerscheinung, sondern der bestimmende Regelvorgang. ^ ■.Aj/wvy X\ .\\\\\\yj\\ . '/A ES3' Abb. 3. Magnetostatische Linsenanordnung mit AuBen- scheibe. a) Magnetische Anordnung; b) elektrisches Ersatz- bild; c) Feldstiirke-Regelkurve. Magnet ostatische Linsenanordmmgen niif mechanischen Rcgelglicdern 19 100'/'- 80 60 40 20 0 4-5lufige 3-stufige ■( ' ^i\ Abbildung :^ \ 1 t x/ ^^^^~~ 6 8 Antnebsskala - 10 Skt Abb. 4. Magnetostatische Vierllnsenoptik (vereinfachle Dar- stellung). Wir haben nun eine Vierllnsenoptik zur elektro- Abb. 5. VcrgroBcrungseichkurvc fiir magnetostatische Vier- linsenoptik mit Zackcnrcglcr. und einen Zackenregler gespcist. Bei gcschlossenem Regler erhielten die Mittellinscn die voile Magnct- spannung von etwa 2000 AW und waren so kurz- brennweitig, daB z.wischen ihnen ein recllcs Bild nenoptischen Erprobung der verschiedenen Regel- entstand. Durch OfTncn dcs Rcglers konntc die verfahren gebaut. Abb. 4 zeigt eine — der Uber- sichtlichkeit halber stark vereinfachte — Schnitt- zeichnung dieser Optik: Die beiden iiuBeren Linsen, Objektiv und Projektiv, wurden mit je einem axial angeordneten Rohrmagneten aus Alnico 400 aus- gestattet. Die Systeme wurden so magnetisiert, daB gleichnamige Magnetpole einander gegeniiberstan- den. Die Mittellinsen wurden uber den AuBenmantel magnetische Spannung an den Mittellinsen stetig auf die Hiilfte des ursprunglichen Wcrtes vermindert werden. Die Brennweitcn der Linsen wurden dadurch so vergroBert, daB aus den zwci Ein/cllinsen ein Dublett entstand. Der Zackenregler ermoglichte auf diese Weise eine wirksame Anderung des Abbil- dungsmaBstabes. Eine Eichung desZackenregleran- triebes gab eine schnelle und fiir die meisten Zwecke Abb. 6. Doppelaufnahme desselben Objektfeldes ciner Platin-Aufdampfschicht mit der magnetostatischen Vierlinsen- optik. Elektronischer MaBstab 28000:1. Wicdergabe 240000:1. Strahlspannung 60 kV. Aiifnalimcmalcrial: Pe- rutz-Kontrastplatte. Belichtungszeil 4 sec. 20 S. LEISEGANG UND O. SCHOTT ausreichend genaue Angabe des jeweiligen Abbil- dungsmaBstabes. Abb. 5 zeigt eine derartige Eich- kurve, links das Gebiet vierstufiger Abbildung, rechts das Gebiet dreistufiger Abbildung mit Dublettwir- kung der Mittellinsen. Wir dimension ierten die Linsen so, daB sich ein maximaler AbbildungsmaBstab von 60000:1 ergab. Im Gebiet vierstufiger Abbildung Hess sich die VergroBerung auf weniger als jq herabregeln, ohne daB die Verzeichnung des Endbildes in radialer oder tangentialer Richtung groBer als 6 "„ wurde. Durch den Ubergang zu dreistufiger Abbildung lieB sich der ausnutzbare Regelbereich noch erweitern. Es stellte sich niimlich heraus, daB dreistufig erzeugte (jbersichtsbilder wesentlich weniger verzeichnet wa- ren als ebenso schwach vergroBerte vierstufige Abbil- dungen. Bei diesem groBen Regelumfang war es besonders angenehm, daB diese Regelart keine Bild- drehung verursachte und es durch gute Zentrierung von Linsen, Zackenregler und Elektronenstrahl ge- lang, die Lage des Bildmittelpunktes wahrend des Regelvorganges praktisch unbeweglich in der End- bildschirmmitte zu halten. Die mit der Anderung der Mittellinsenerregung verbundene Defokussierung des Bildes muBte durch andere Regelorgane kompensiert werden. Wir hatten dafur urspriinglich eine Innenscheibe im Objektiv- kreis vorgesehen. Es stellte sich aber heraus, daB der Regelbereich der Scheibe dieser Aufgabe nicht anniihernd entsprach. AuBerdem war ihre Axialbe- wegung infolge des unvermeidbaren mechanischen Spiels mit geringen Querbewegungen verbunden. Die dadurch verursachten Verschiebungen des Bildes storten die Beobachtung gerade bei der Scharfstel- lung. Ein derartiger EinfluB der Scheibenbewegung auf den Elektronenstrahl nimmt ab, wenn die Ent- fernung zwischen Regelkorper und Strahl wiichst. Nun enthielt die Anordnung auBer dieser Innen- scheibe noch zwei weitere Regelscheiben, die sich zwischen den beiden Magneten und dem Mantel des Systems befanden. Diese AuBenscheiben waren vorziiglich geeignet. eine Grobscharfstellung des Bildes vorzunehmen. Die Feinfokussierung mit mechanischen Regelgliedern erwies sich als unbe- quem, weil bei der Hin- und Riickbewegung infolge der Hysterese des magnetischen Materials etwas verschiedene Regelkurven durchlaufen werden. Sehr viel angenehmer ist statt dessen eine Feinregelung der Strahlspannung. Der notwendige Regelumfang von ± 500 V hat nicht den oben erwahnten Nachteil, daB die Durchdringungsfahigkeit der Elektronen merklich geiindert wird. Es ist noch zu erwiihnen, daB es durch gleichzeitige Betatigung von Zackenregler und AuBenscheiben gelang, Beugungsdiagramme durch die nicht ab- schaltbaren Linsen zu fiideln. Allerdings muBte der Objektbereich dabei auf weniger als 30 /.i c begrenzt werden. Wir wollten die besprochene Anordnung gleich- zeitig benutzen, um zu demonstrieren, daB sich auch mit derartig geregelten magnetostatischen Linsen ein recht gutes Auilosungsvermogen erreichen liiBt. Wir legten deshalb groBen Wert auf gute Zentrierung von Linsen und Elektronenstrahl. Bei der schritt- weisen Entwicklung der optischen Eigenschaften war auBerdem der Einbau eines Stigmators besonders wirkungsvoll. Wir erreichten bisher ein Auilosungs- vermogen von 2 m/< (Abb. 6). Diese Grenze wird durch den Farbfehler bestimmt und ist wahrschein- lich auf die objektbedingte Streuung der Elektronen- energien zuriickzufi-ihren. LiTERATUR Langner, G., Optik 12, 554-562 (1955). Der EinfluB der Bestrahlungsbedingungen auf die Objektverschmutzung S. Leisegang und O. Schott Siemens und Halske AG, Wernerwerk fiir Mefitechnik, Berlin-Sicmensstadt Die Objektverschmutzung ist bei der Mikroskopie hochster Auflosung eine sehr storende Erscheinung. Nach den Messungen von Ennos ist die Verschmut- zung stark abhangigvonderTemperaturdesObjektes. Durch verschiedene Bestrahlungsbedingungen kann die Temperatur des Objektes bei gleicher Stromdichte in weiten Grenzen verandert werden. Das hat zur Folge, daB auch die Objektverschmutzung stark von den Bestrahlungsbedingungen abhiingt. Diese Ab- hiingigkeit wird theoretisch und experimentell unter- sucht. Von besonderer Bedeutung ist dabei die Tat- sache, daB bei Bestrahlung der Objektblende selbst eine ganz wesentliche Erhohung der Temperatur des Objektes eintritt. Die Objektverschmutzung wurde in einem El- miskop I als Funktion der Stromdichte bei Zim- mertemperatur (etwa 25 C) auf folgende Weise gemessen: Mit Hilfe des Feinstrahlkondensors wird ein klei- ner Bereich des Testobjektes von 2 /< 0 bestrahlt. Durch Bestrahlung eines so kleinen Bereiches wird die Temperatur des Testobjektes auch bei hoher Stromdichte nicht merklich erhoht [4]. Die Bestrah- lungsstromdichte wird bei bekannter VergroBerung im Endbild mit Faradaykiifig und Elektrometer gemessen. Die Objektdicke d und die Dicke der aufgewachsenen Verschmutzungsschicht c/,, wird durch Messung der Streuabsorption (Objektivaper- Bestrahliingshi'dingungen imd Objektverschmutzung 21 10 A/sek I f \ oos 0,1 015 A/cm^ dZ Bild 1. Objektverschmutzung V als Funktion der Strom dichte j bei Zimmertemperutur. turblende 5Q i-i 0 , Objektivapertur a„ - 9- 10-=*) nach den in [5] angegebenen Formeln bestimmt. Die Mes- sung erfolgt auch hier mit Faradaykiilig und Elek- trometer im Endbild. MeBgroBen sind die im Fara- daykafig durch die unverschmutzte Folie gemessene Stromdichte y und die durch die verschmutzteZone ins Endbild gektngende Stromdichte /,.. Die Dicke der aufgewachsenen Verschmutzungsschicht d^ er- gibt sich dann aus: Jv — =exp i-dyiXe). J (1) Fiir /.^., die mittlere freie WegUinge fijr elastischen StoB, wurde bei 80 kV Strahlspannung der Wert Ap 2000 A eingesetzt. Als Testobjekte wurden etwa 300 A dicke Kohlefolien verwendet. Das Ergebnis dieser Messungen ist in Bild 1 dargestellt. Der in Bild 1 wiedergegebene Verlauf der Objekt- verschmutzung als Funktion der Stromdichte legt eine statistische Deutung der Erscheinung nahe. Die die Verschmutzung erzeugenden Molekiile mogen einen Radius R, ein Molekulargewicht M, einen Partialdruck p und eine durch Adhasion be- dingte Verweilzeit auf der Folie t haben. Die Wahr- scheiniichkeit ^V, daB eines der die Verschmutzung erzeugenden Molekiile bei gegebener Elektronen- zahl // cm- sec getroffen wird. ist gegeben durch W^TiR-TH R'nr- ^ e (2) 7 ^ Stromdichte (A cm-), e -- Elektronenladung (Coul). Die Wahrscheinlichkeit W wird gleich 1 I'iir die Stromdichte y„, die gegeben ist durch 70 = jiR^r (3) Fiir die Objektverschmutzung f^(7)ergibt sich daraus die — die MeBergebnisse gut wiedcrgcbendc — Beziehung J/(i) = J/(oo)[l -exp(-/7o)]. Eine weitere Betrachtung erlaubt. aus Annahmen liber den Molekiilradius die Verweilzeit t und den Partialdruck p der die Verschmutzung erzeugenden Diimple ab/uschiitzen. Aus der kinctischcn Gastheoric [2] folgt zuniichst bei gcgebencm Partialdruck /; die Zahl der StoBe gegen die Folie ( Wand) pro Sekunde und Quadrat- zentimeter A: /4 = 3,5 • 10" °K^- mm Ug"' cm"- sec"' 1 MT (4) Die Verweilzeit der Molekule auf der Folic bei rein elastischem StoB t„ sei dadurch dcfiniert, daB ange- nommen wird: Verschmutzung erzeugende Mole- kiile sollen dann noch von den Elektronen auf der Folic „festgenagelt*' werden, wenn sie beim Zusam- menstoB nicht ueitor als eine Strecke A.v von der Folie entfernt sind. Ausdcr mittlercnGeschwindigkeit r - 14,5- lO'cm -1 l^-'/o \j/]^ (jer Molekule folgt damil fiir cm sec" Ax/z;. (5) Die Zahl der Molekiile /?, die sich pro Sekunde und Quadratzentimeter bei einer Verweilzeit t > t„ auf der Folie belinden, ist dann gegeben durch n (6) Die Objektverschmutzung F(oo), bei der die Strom- dichte der Elektronen so groB ist, daB allc Molekiile festgenagelt werden, ergibt sich aus der einfachen Annahme, daB die Zahl der Molekiile fiir eine cin- molekulare Schicht der Fliicheneinheit /?i gegeben sei durch (7) K'n zu F(cxd) 1R = n. >3 T n R 71 • 10-"' • mm Hg^' cm ' sec^" ^ -- • (8) Zwischen M, R und Dichte q gelte die oft gebrauchte Beziehung [2] R 0.66-1 (9) Damit fiillt aus (8) R und M heraus und mit n ^ 1 g cm'' (schwcres Ol) gilt F(oo),. 10^ cm- sec " mm Hg -1 -^P A.v (10) Fiiry,, und r( • ) crgebcn sich aus Bild I die Werte /o= 0,014 A cm- und Vi ^ ) 8 A sec. Als Ver- schmutzung erzeugende Molekule seien hochmolc- kulare KohlcnwasserstotTc mit A/ ^ 300 und damit nach Gl. (9) R ^ 4,5 A angenommen. 77 S. LEISEGANG UND O. SCHOTT Dann folgt aus Gl. (3) fijr die Verweilzeit t ^ 2- 10"^ sec und mit der groBenordnungsmaBig sicher richtigen Annahme A.y =t 1 A fiir den Partialdruck p ^ 5- \0~^'^ mm Hg. Die Verweilzeit ist also relativ sehr groB, der Partialdruck der die Verschmutzung erzeugenden Diimpfe auBerordentlich klein. Als unsichere Faktoren enthalt Gl. (3) den Wert von R, ein um den Faktor 4 kleineren Radius, der als minimaler Wert gelten kann, fiihrt zu einem um den Faktor 16 groBeren Wert fiir t und damit nach Gl. (10) zu einem um den Faktor 16 kleineren Wert von p. AuBerdem ist p noch proportional zum nur groBenordnungsmaBig bekannten Abstand A.y. Die Objektverschmutzung als Funktion der Tem- peratur wurde von Ennos [3] bei einer Stromdichte von 0,01 A cm- im Bereich von 50 bis 200 C ge- messen. In diesem Bereich lassen sich die Messungen wie- dergeben durch die Beziehung: ViT) = F„exp(-r/r„), T = Temperatur CK), r„ =- const =^ 84°K. (11) Es sei angenommen, daB diese Beziehung im ganzen interessierenden Bereich 300 'K < T < lOOO'K gelte. Die wesentlich temperaturabhangige GroBe in Gl. (3) und (10) ist die Verweilzeit r. Die Objektver- schmutzung als Funktion von Stromdichte und Tem- peratur V{j, r)sollte sich dann darstellen lassen als: t{T) - exp - 7 y T(T) Jo '^z (12) Tz, ^z = Werte von T und t beiZimmertemperatur. FiJry = 0,01 A cm- soil die von Ennos gemessene Beziehung (11) gelten. Bild 2. Objektverschmutzung Tals Funktion von Stromdichte j und Temperatur 7" (nach Formel (12)). + + = Temperaturabhangigkeit bei J -= 0,01 A cm- nach Ennos. O O = Stromdichteabhangigkeit bei T 300°K nach Bild 1. Mit diesen Annahmen kann die Funktion r (T) und damit aus Gleichung (12) die Verschmutzung als Funktion von Stromdichte J und Temperatur T, V (j, T) berechnet werden. Die Ableitung der Formeln ist in Anhang I gege- ben, das Ergebnis der Rechnung zeigt Bild 2. Die Temperaturverteilung auf einer Folic glei- cher Dicke wurde in einer friiheren Arbeit als Funktion der Stromdichte und des bestrahlten Be- reiches berechnet [4]. Bei dieser Rechnung wurde vorausgesetzt, daB sich die Objektblende oder das Kupfernetz, von dem die Objektfolie getragen wird, auf Zimmertemperatur befindet. Diese Bedingung ist nicht immer erfullt. Wird auBer der Objektfolie auch ein groBer Teil des Objekttragers bestrahlt, so kann die Temperatur des Objekttragers wesentlich hoher sein als die Temperatur, die sich bei gleicher Stromdichte bei Bestrahlung der Objektfolie allein ergibt. Auch die Erwarmung der Objekttrager liiBt sich aus den in [4] angegebenen Formeln in einfacher Weise berechnen. Da die Warmeabstrahlung bei den relativ dicken Objekttragern (Dicke etwa 50 //) praktisch keine Rolle spielt, wird die Temperatur proportional zur Stromdichte j. AuBerdem geht die geometrische Gestalt der Objektblende ein: Die Dicke d und der Radius R des inneren Bereichs der Blende mit der Dicke d spielen dabei — neben den Materialkonstanten — die wesentliche Rolle. Die Reichweite der Elektronen ist kleiner als die Dicke der Objekttrager: praktisch die gesamte kinetische Energie der Elektronen wird in Warme umgesetzt. Der Blendenrand, dessen Dicke (etwa 1 mm) sehr groB gegeniiber der Dicke d des inneren Bereiches ist, sei als so gut wiirmeleitend angenommen, daB dort Zimmertemperatur erzwungen wird. Auf die Einzelheiten der Rechnung sei hier nicht naher eingegangen, die Formeln sind in Anhang II zusam- mengestellt. Die Temperatur einer Objektblende bei einer M °C furj-0fllk\m^ _^ \ ' "■ / ^ / / / 100 / / / / ?\-Blen(Je: > Cu-Netz : > ■ = 50\i.; R= 0,3mm i~2S\i: /?= / mm / A / n ^ k 0^1 furP\-B lende ¥ mm 0^ 0,1 WrCn-Nefz 0,5 mm 1 'b- Bild 3. Erwarmung \t einer Objektblende aus Platin der in Bild 5 dargestellten Form oder eines Kupfernetzes von 25 (x Dicke, das mit einem Kiipferring vom Durchmesser 2 R = 2 mm gehalten ist. als Funktion des Radius des bestrahlten Bereiches rs bei einer Stromdichte j = 0,01 A cm^. Die Erwarmung ist proportional zur Stromdichte j (siehe Glei- chung 11,4). Bestrahhrngsbedingungen und Ohjck t vcrschnwtziing 23 Stromdichte von / 0,01 A cm- ist in Bild 3 darge- stellt. Als geometrische Daten sind die Wcrte ciner oft verwendeten Objektblende {d = 50 //, R 300 //, Material Platin, siehe Bild 5) in die Rechnung cin- gesetzt. Die Temperatur wird Funktion vom Radius des bestrahlten Bereiches 1-^. Zwischen Stromdichte y, Radius des bestrahlten Bereiches r^ und Strahl- stromstiirke / besteht die Beziehung: r^-rrv = /• (13) Danach erreicht die Temperatur der Objektblende • — die Temperatur ist in erstcr Niiherung proportio- nal zur Stromdichte — bei Bestrahlung eines Be- reiches von 600 fi 0 bei einer Strahlstromstarke von 28 /rB die Stromdichte y = 0 ist. Tat- siichlich hat die Stromdichte als Funktion von r eine GauB- verteilung j (r) =Jq exp — {rJrB)'-. Die aus Gl. (11,1 ) berechnete Temperatur wird, solange /-s^ Blendenradius ist, merklich groBer als die gemessene Temperatur. Bestrahlungshedingiingen unci Ohjck I vcrschmiitzung 25 Anhang IV wiedergegeben, das Ergebnis in Bild 7 als gestrichelte Kurve eingezeichnct. Die Uberein- stimmung beider Kurven ist fiir //, > 0,4 A cm- recht gut, die Abweichungen fiir y^ < 0,4 A cm- werden in Anhang IV diskutiert. ANHANG T Berechnung der Funk t ion t(T) Annahmen: Fiir V(j\T) gelte die Glcichung (12). Fiir J 0,01 A cm- gelte die Beziehung von Ennos (Gleichung II). Dann folgt fiir die Verschmutzung als Funktion der Temperatur bei einer Stromdichte von 0,01 A cm-: 1/(0,01, T)=V(c^, r^)""^^^ Tz / 0,01 r(T)\ (1,1) Aus dieser Gleichung ist t{T) zu bestimmen. Es ergeben sich einfache Losungen fiir den Fall, daB (0,01 y„) [t{T) t^] sehr groB oder sehr klein gegen eins ist. Es gilt in diesen beiden Fallen: T{T)=r, fur 0,7 >!, F(co, Tz) Tz t( f F(oo, Tz) (1,2) fur 0,7 <1, Als einfache Losung fiir t( T) sei eine Summe dieser beiden Grenzlosungen angenommen: T(r)=Tz K(co, Tz) 0,7 • V<, K(oo, Tz) exp (TjlTf^) (1,3) Aus der Bedingung, daB fiir T = Tz^ 300'K die Verweilzeit t r^ sein soil, kann K„ bestimmt wer- den. Es gilt dann fiir t:{T) die Beziehung: r{T) = 6,2- 10-3 exp (~ 0,12 ^^ •[1 + 1,55 exp (0,006 7)]. (1,4) Durch Einsetzen dieser Werte fur t(D in Gleichung (12) ergibt sich die gesuchte Beziehung fiir V{j, T). Das Ergebnis der Rechnung zeigt Bild 2. ANHANG II Berechnung der Temperatur der Ohjek I blende Fiir die Temperatur auf einer diinnen Scheibe gilt die von v. Borries und G laser [1] aufgestelltc Diffc- rentialgleichung: d^'rio) ^driq) y do'' o do ^ (THQ)-f*) = 0. (11,1) Dabei bedeuten e=o = r Entfernung vom Mittelpunkt der Blende r(o) f' R Radius, bei dem Zimmertemperatur erzwungen wird T(o) Folientemperatur als Funktion von o Zimmertemperatur T' '0 1 + 0,12Qt-J CTt mit Q , der in der Folic in Wiirmc umgcsetzte Teil des Energievcrlustcs eines Elektrons (V), bei dickcn Folien glcich der Beschleuni- gungsspannung der Elektronen j = Stromdichte im Objekt (A cm'-) C = Strahlungskonstante (cal cm- sec grad') y- SCR^Tt x = Foliendicke (cm), / - Wiirmeleitzahl des Objektes (cal/cm sec grad). In [4] werden zwei Naherungslosungen fiir Glei- chung (11,1) angegeben. Der Geltungsbereich der Losungen ist dadurch gegeben, daB y/4 entweder sehr klein oder sehr groB gegen 1 (f - l)/f * ist. Fiir y/4 = I (f- l)/f* haben beide Losungen das gleiche Ergebnis, das fiir 78 nicht mehr als 10 "„ vom wahren Wert abweicht. Fiir den hier interessicren- den Fall relativ dicker Blenden (xsi^5-10~' cm) mit guter Warmeleitzahl ist die Bedingung y/4 < Kf-l)/f* allgemein erfiillt und y4 I (f - 1) f^ erfiillt, so daB nach [4] fiir die Temperatur auf der Folie die Naherung gilt, bei der die Warmeleitung in erster Naherung, die Warme- abstrahlung voll beriicksichtigt ist. Bei Bestrahlung der ganzen Folie mit gleicher Stromdichtegiltdanach fiir t(o^) die Beziehung: T(C'f) =T + Ci Jo (.h'T-Qp). (1TT,1) Der relative Radius Op der Folie ist hier gegeben durch das Verhiiltnis von /• zum Radius der Blen- denbohrung = Radius der freitragenden Folie Rp, so daB gilt: op = r Rp. Fiir Qp = 1 Oder r = Rp soil t (1) = t^ sein. Daraus folgt aus (111,1) C^- (111,2) und fiir die Temperatur in der Folienmitte, t(0): T(0)=f + Tb (111,3) ^(/yf5) Nach dieser Formel wurden die Kurven der Bilder 4 und 5 berechnet. ANHANG IV Daten zur Berechmmg der Temperatur einer Kohlefolie Die Kohlefolie liegt auf einer Platinblende der in Bild 5 gezeigten Form. Fiir die Erwiirmung der Platinblende gelten die Tabelle 1 angegebenen Daten. Fiir die Kohlefolie wurden folgende Daten einge- setzt: e=2g/cm3 0r = 8,5eV ^=3-10-^ x = 6-10-«cm /? =3.5 •10-'' cm C \-\Q-'-' Daraus folgt: f'= 1+0,2^:7; r" o 8,6-10-' (IV,1) Zur Berechnung der GroBe des bestrahlten Bereiches wird die Gl. (1) benutzt. Die Blendenbohrung wird dadurch beriicksichtigt, daB ein Nutzeffekt H einge- fiihrt wird. Dieser Nutzeffekt ist definiert durch die relative Zahl der Elektronen, die die massive Blende treffen. R H=\- j [exp - (rirBf]rdr. (IV,2) 0 An Stelle von J rB~ wird HjrB~ in Gl. (11,3) eingesetzt und daraus ts bzw. t^ berechnet. Aus Gl. (111,3) folgt dann die Temperatur in der Mitte der Folie. Die Rechnung wird fiir /,, = 40 //A durchgefiihrt. Die folgende Tabelle 2 zeigt den EinfluB der verschiede- nen Faktoren auf die Temperatur in der Mitte der Folie. Tabelle 2. Berechnung der Temperatur einer 600 A dicken Kohlefolie auf einer Platinblende mit 70 fi Blendenbohrung und einem Bereich von 600 /< Durch- messer mit der Dicke 50 // bei 40 ftA Strahlstrom. /B = Temperatur am inneren Rand der Platinblende /q = Temperatur der Folie fiir tB = 30"C /f = Temperatur in der Mitte der Folie nach Gl. (111,3) j rB H IB ^0 tF A/cm- /' O' /O °c °C °C 0,01 350 100 200 36 190 0,1 113 91 750 84 655 0,3 65 75 760 205 660 1 36,5 38 610 420 680 1,5 30 25 530 520 750 In Bild 7 ist das Endergebnis der Rechnung dar- gestellt. Die Abweichungen fiir kleines J (groBe r^) sind verstandlich durch die in die Rechnung nicht voll eingehende GauBverteilung (siehe Anmerkung Seite 24) und durch die wegen der Blendenform (siehe Bild 5) nicht voll erfiillte Forderung tg = tz fiir r = 300 ft. Die recht gute iJbereinstimmung der Kurven fiir y > 0,4 A cm- zeigt noch einmal, welch groBen EinfluB die Temperatur der Platinblende selbst auf die Objekttemperatur und damit auf die Objektver- schmutzung hat. Ohjektkiihlung im Elektronenmikroskop 11 LiTERATUR 1. VON BoRRits, B. iind Glasfr, \V., K'olloicl-Z. 106, 123 (1944). 2. DiJSHMAN, S., Scientific Foundations of Vacuum Tech- nique. New York, 1949. 3. Ennos, a. E., Brit. J. Appl. Pliys. 5, 27 (1954). 4. Leisegang, S.,Zur Erwiirmungelektroncnmikroskopischer Objektc bei kleinem Strahlquerschniti. I'roc. Int. Conf. El. Microscopy. London. 1954. 5. — Elektronenmikroskope. //;: Handbuch der Physik 33. Springer, 1956. Objektkuhlung im Elektronenmikroskop O. ScHOTT und S. Leisegang Siemens & Halske AG., Wenierwerk fiir Mefitechnik, Berlin-Siememstadt Versuche von S. Leisegang (1) mit einer gekiihlten Objektpatrone zeigten, daBsich dieObjektverschmut- zung vermeiden laBt, wenn der Dampfdruck der restlichen Kohlenswasserstoffe im Vakuum des Mi- kroskops in unmittelbarer Umgebung des Objektes durch Ausfrieren bei Temperaturen um 80 C stark vermindert wird. Es wurde eine Objektkuhlungsein- richtung gebaut mit dem Ziel, die verschmutzungs- freie Beobachtung der Objekte im Elektronenmikro- skop zu erreichen, ohne dabei die Leistungsfahigkeit des Gerates zu vermindern, wie das etwa durch eine thermisch oder mechanisch bedingte Objektwande- rung geschehen konnte. Uber den Aufbau der Ob- jektkuhlungseinrichtung und einige damit erzielte Ergebnisse soil hier berichtet werden. Die kiihlbare Objektpatrone ist in Bild 1 darge- stellt. In dem normalen Patronenkonus sitzt ther- misch isoliert ein Kupferzylinder, der beiderseits durch Blenden gegen den ubrigen Mikroskopraum abgeschlossen wird. Die untere Blende aus Platin- Iridium von 4 mm O mit einer Bohrung von 150 fx 0 sitzt im Polschuh und befindet sich auf Zimmer- temperatur, wiihrend das gekiihlte Objekthiitchen in einem Abstand von 0,2 mm bis 0,4 mm iiber der Blende bewegt werden kann. Das Hiitchen hat eine Bohrung von 1 mm ;3 , so daB ein ausreichendes Gesichtsfeld vorhanden ist. Durch die Blenden- anordnung ist die Wahrscheinlichkeit des Ein- r^ Eisen Bronze TrolituI Kupfer (gekiJhlt) Konstantan zum Thermoelement Polschuh Objektblende 70 p.

. 631 und 27, 7.3 (1936). 2. Degenhard, W. und Mollenstedt, G., Optik 14 (im Druck). 3. Mollenstedt, G., Optik 12, 44! (1955). 4. Panzer, S., Optik 7, 290 (1950). 5. — ihicl. 10, 107 (1953). 6. Rang, O. und Weitsch, W., Optik 13, 201 (1956). 7. Steigerwald, K. H., Optik 5, 469 (1949). Bolzenkathode als Objekt im Elektronen-Emissionsmikroskop E. B. Bas Institiit fiir techuischc Physik, Eidgcnossische Tcclinischc Hoclisclmle. Ziirich In fruheren Arbeiten [1-4] wurde die vom Verfasser entwickelte Bolzenkathode im Zusammenhang ihrer Anwendung in Elektronenkanonen niiher behandelt. In folgender Arbeit soil gezeigt werden, daB diese Kathodenkonstruktion auch im Elektronen-Emis- sionsmikroskop als Objekt gute Dienste leisten kann. Anhand von Fig. 1 wollen wir zuerst das Prinzip der Bolzenkathode mit einigen Worten kurz cr- lautern. Das als Bolzen bezeichnetc Wolfram-Stiib- chen 1 ist an einem Ende in einer Klemmbacke 2 fest eingeklemmt, wahrend die Stirnfliiche am ande- ren Ende als Emissionsfliiche dient und in unserem Falle den zu untersuchenden StofT 3 enthiiit. Um den Bolzen herum ist konzentrisch eine Wolfram- Wendel 4 angeordnet, welche von den Stromzufiih- rungsstiften 5 und 6 getragen wird. Diese Zufiih- rungsstiftc tragen zugleich einen Strahlungsschutz, bestehend aus zvvei konzentrisch /um Bol/en ange- ordneten Molybdiin-Zylindern 7. Kathodenblende 8 vervollstiindigt die Kathode, wahrend V die erste und die zweite Blende des Immersionsobjektives darstellt. Die Heizung des Bolzens erfolgl so. daB die Wendel durch Stromdurchgang geheizt wird und als Primarkathode dient. Sie erhalt gegeniiber dem Bol- zen eine negative Spannung von einigen hundert Volt, womit die von der Wendel emittierten Elektro- nen auf den Bolzen aufprallen und ihn auf die ge- 38 E. B. BAS Fig. 1. Schnitt durch die Bolzenkatliode als Objekt im Emis- sionsmiicroskop. wiinschte Temperatur aufheizen. In der Heizschal- tung ist dafur Sorge zu tragen, daB infolge des Ruckheizeflfektes vom Bolzen auf die Wendel keine Heizinstabilitat auftritt. Bei niedrigeren Objekttem- peraturen kann die Bombardierung vollig aus- bleiben. Die Heizung des Bolzens erfolgt dann durch die Zustrahlung der Heizwendel. Fur die Anwendung der Bolzenkathode als Objekt im Emissionsmikroskop sind vor allem drei Dinge von Bedeutung: Axiale Ausdehnung des Bolzens, Schwingungen des Bolzens und das magnetische Storfeld der Heizwendel. Die axiale Ausdehnung des Bolzens ist nicht zu vermeiden. Die Emissionsflache der Bolzen- kathode bei der in Fig. 1 dargestellten Konstruk- tion verschiebt sich beim Aufheizen auf 2000"K Objekttemperatur um ca. 0,1 mm und bei lOOO'K betragt die Verschiebung ca. 0,04 mm. Die Ver- schiebung des Objektes bei konstanter Objekttem- peratur infolge Nichterreichen des Beharrungszu- standes ist meistens so gering, daB Belichtungen bis zur 30 sec. Belichtungszeit nicht dadurch gestort werden. Den mechanischen Schwingungen des Bolzens ist groBe Bedeutung beizumessen, da sie die erzielbare Auflosung stark beeinflussen konnen. Um hier einen Uberblick zu bekommen, betrachten wir den idea- lisierten Fall eines einseitig eingespannten elasti- schen Stabes von der Lange / mit einem Massen- punkt m am Ende (Fig. 2). Die Transversal-Koordi- nate dieses Massenpunktes bezeichnen wir mit y* und die laufenden Koordinaten des Stabes mit x, V. Dann laBt sich die Schwingungsgleichung dieses Systems wie in Fig. 2 formulieren. | ist der soge- nannte normale Viskositatskoeffizient [5], / equa- torielles Tragheitsmoment und E der Elastizitiits- modul. Fur die Anwendung dieser Betrachtungen auf die Bolzenkathode wollen wir folgende Vereinbarung treflfen: als die Stablange / soil derjenige Abschnitt gelten, welcher sich nicht auf hoher Temperatur befindet und dadurch seine Elastizitat bewahrt. Als Masse in gilt die Masse des iibrig bleibenden End- stiickes des Bolzens. Nehmen wir als Beispiel fiir G ii m d"-y* d'^y ,d Id" y dr dx dt \dx 0 y =yo ^ 6xp 1 a o . COnt] COS O) t IE ml" /o = 2-ln2 ^ E \^ Fig. 2. Mechanische Schwingungen des Bolzens. unsere in Fig. 1 dargestellte Konstruktion mit: 1 = 1 10-1 cm, Jfi = 6 1 0-- cm, Ib-1 = 6 10-^ cm, m = 3,27 -< lO"'^ g, und fur Wolfram E^ 3,5 x 10'- Dyn/cm'^ | * 10* Dyn- sec/cm-, so erhalten wir: ojo ^23 kHz, /„ ^ 10-* sec. Nehmen wir weiterhin an, dass der Bolzen durch einen StoB auf eine Amplitude von: jo = 10-- cm angeregt wurde, so wird diese Amplitude in 1 m-sec. auf 10^- 10-^ = 10-5 cm oder 0,1 /< absinken. Wir sehen hieraus, daB die Erschutterungsempfindlich- keit bis herab zu der Auflosungsgrenze des Emis- sionsmikroskopes ausreichen diirfte. Die dritte uns hier interessierende Frage betriflft das magnetische Storfeld im Immersionsobjektiv, hervorgerufen durch den Wendel-Heizstrom. Dieses Storfeld kann man in zwei Anteile zerlegen; rota- tionssymetrischer Anteil hervorgerufen durch die Windungen der Wendel und Querfeldanteil, her- vorgerufen durch die Wendelzuleitungen. Die rotationssymmetrische Komponente ist sofort aus der Feldstiirkengleichung einer Zylinderspule anzugeben: m 'w W 11 w 1 1 + i-W /w + 2z 1 + «w /w-2z A/cm (6) Hier bedeutet: /„, = Wendelheizstrom W = Anzahl der Wendelwindungen 1^ = Wendelliinge d^jj = mittlerer Wendeldurchmesser z = laufende Koordinate auf der Achse der Wen- del von der Mitte der Wendel gezahlt. In unserer Anordnung hat die Heizwendel folgende Daten: /jv = 3,5 mm d^fj -= 1,35 mm W - 14 l^ ^^ 2 A. Bolzenkatbode als Ohjekt ini Elektronen-Emissionsmikroskop 39 12 10 Jw- 2A "w = J,5mm w ■ = M If, = < mm — 1 1 1 1 1 1 1 1 I 2 3 A 5 6 7 8 Ztinmm Fig. 3. Axialsymmctrische und transversale Antcile der magnetischen Storteldstiirke der Heizwendel. Damit erhalten wir fiir die axiale Feldstarke die Beziehung: H, ^ 50 1 i^^{^:i j/-t-S) Oersted (7) (r in mm). Diese Beziehung ist in Fig. 3 graphisch dargestellt. Die Querkomponente des magnetischen Wendel- Storfeldes, hervorgerufen durch die Wendel-Zulei- tungen, rechnen wir aus einem vereinfachten Strom- kreis. Nach dem Biot-Savartschen Gesetz laBt sich fiir jede einzelne Stromkreisstrecke fiir den in einem Punkt P erzeugten Feldanteil angeben: , / sin A' , H^ I ;- .,dx. In r (8) «! Die Rechnung fuhrt auf folgende Beziehung: Hy w 4 71 In ^ \z~lw /2 + i^ + VI Ik z-lwl2 A/cm (9) Setzen wir wieder unsere obigen Wendcidatcn und fiir die Liinge des Zuleitungsdrahtes l^ 4 mm ein, so erhalten wir: 30 35 AO 45 50 Abstand Hei/wendelmitte- Emissionsflache in mm 55 Fig. 4. Abliiingigkcii der Heizleistung von der Entfernung der Emissionsfliiche von der Heizwendelmitle. Bolzen 0: 0,6 mm, Objekt :" : 1,5 mm. Hy = 0,5 /' ' {^) [Oe] (10) (z in mm). In Fig. 3 finden wir auch Hy in Abhiingigkcit von z graphisch aufgetragen. Aus konstruktiven Griin- den betriigt der minimale Abstand der Emissions- fliiche von der Wendelmitte ca. 2,8 mm. Fiir diesen Fall erhalten wir fiir die beiden Feldstiirkenkompo- nenten folgende Werte: i/, = 7,3 [Oe], Hy^ 1,63 [Oe]. Der Einnuf3 dieses Storfeldes auf die Abbildungs- giite des Immersionsobjektives ist schwer zu bcur- teilen. Die obige Angabe fiir den Abstand zwischen Wendelmitte und Emissionsfliiche von 2,8 mm ist der konstruktiven Ausfiihrung der Bolzenkathode als Eiektronenquelle fiir die Elcktroncnkanonen ent- nommen. Beim Emissionsmikroskop kann die Emis- sionsfliiche bedeutend weiter von der Heizwendel weggeriickt werdcn, wodurch nach Fig. 3 die Stor- 200 ^ 150 '.no c a c ^ 50 Parameter Temperatur der Cmisslonsflache 2.5 30 35 40 <5 50 Abstand Heizwendelmrtte — [missionsflache in mm 5.5 Fig. 5. Axiale Verschiebung der Emissionsfliiche bci \cr- schiedenen Temperaturen in Abhangigkeit des Abstandes Emissionsflache-Heizwendelmiite. Bolzen 0: 0,6 mm, Ob- jekt 0: 1,5 mm. 40 E. B. BAS Fig. 6. Objektpraparation. feldstiirke der Heizwendel noch weitgehend vermin- dert werden kann. Es erhebt sich nun die Frage. wie sich die Bom- bardierungs-Leislung des Bolzens fiir eine bestimmte Objekttemperatur beim Wegriicken der Emissions- fliiche von der Heizwendel verandert. Diese Frage wurde experimentell gepriift, woraus die Kurven- schar in Fig. 4 resultiert. Die Messungen wurden mit einem W-Objekt durchgefiihrt und die Anord- nung war so getroffen, daB der Abstand zwischen Wendelmitte und Einspannstelle des Bolzens stets 11, 2 mm betrug. Bolzendurchmesser war d^ = 0,6 mm und Objektdurchmesser cIk =1,5 mm. Durch die Verlangerung des hochgeheizten Bolzenendes wird nun auch die axiale Ausdehnung des Bolzens ausge- pragter. Die gemessenen Ausdehnungen sind durch die Kurvenschar in Fig. 5 graphisch dargestellt. tJber den EinfluB des Abstandes Wendelmitte-Emis- sionsflache auf die Abbildungsgiite wurden noch keine systematischen Versuche durchgefiihrt. Wir wenden uns jetzt der Objektpraparation zu. Diese besteht aus drei Stufen: a) Praparation des Wolfram-Bolzens; h) Aufschmelzen des zu unter- suchenden Stoffes auf den Wolfram-Bolzen; c) Pre- paration der Emissionsflache. Der Wolframbolzen wird aus dem handelsiiblichen Wolframdraht hergestellt. Der meist angewendete Durchmesser betragt 0,6 mm. Zuerst werden Draht- stucke von ca. 300 mm Lange in einer Streckvor- richtung im Schutzgas bei ca. 1000 C gestreckt. Aus so gerichtetem Draht werden Stiicke von passender Lange herausgeschnitten und in einem Rekristallisa- tionsofen bei 3000 K rekristallisiert (Fig. 6a). Nun werden die rekristallisierten Bolzen in einen Halter eingesteckt und in eine Kathodenstrahlrohre ein- geschieust. Durch den auf die Stirnflache des Wol- fram-Bolzens konzentrierten Elektronenstrahl wird das Wolfram so weit geschmolzen, bis eine Kugel von gewiinschtem Durchmesser entsteht (Fig. 6b). Es stand uns im Elektronenstrahl eine Leistung von 14 kV < 10 mA = 140 W zur Verfiigung, womit Kugeldurchmesser von ca. 2 mm ohne Schwierigkeit erzielt werden konnen. SchlieBlich wird die Wolfram- kugel bis auf die Halfte abgeschliffen (Fig. 6c). Der nachste Schritt ist das Aufschmelzen des zu Fig. 7. Emissionsbild. untersuchenden Stoffes. Dieser Stoff kanninmassiver Form Oder in Pulverform vorliegen. Liegt er in mas- siver Form vor, so wird ein Stuck von geeigneter GroBe auf die Stirnflache eines wie oben vorbereite- ten Wolframbolzens mit Hilfe von Kollodium ange- klebt (Fig. 6d). Falls der zu untersuchende Stoff in Pulverform vorliegt, wird eine bestimmte Menge von diesem Pulver bzw. Pulvergemisch in einer kleinen PreBform auf die Planflache der Halbkugelkalotte des Bolzens aufgepreBt (Fig. 6e). Es kann ein wenig Kollodium als Bindemittel zugegeben werden. Der Wolframbolzen mit aufgeklebtem bzw. aufge- preBtem Praparat wird wieder in den Elektronen- schmelzofen eingeschleust und durch Elektronen- beschuss auf das Wolfram aufgeschmolzen (Fig. 6/). Da Wolfram einen hohen Schmelzpunkt besitzt und das Aufschmelzen unter Umstanden in Bruchteilen einer Sekunde erfolgt, besteht sehr geringe Wahr- scheinlichkeit fur eine Legierung mit Wolfram. SchlieBlich kommt die Fertigstellung des Objektes. Dafur wird zuerst die Schmelzkugel auf dem Wol- frambolzen auf den gewunschten Durchmesser rund- geschliffen und dann die Emissionsflache plange- schliffen (Fig. 6^). Bei diesem letzten Schritt ist zu beachten, dass die resultierende Dicke des zu unter- suchenden Stoffes auf dem Wolfram ausreichend ist. Die SchluBphase der Praparation bildet die feine Politur der Emissionsflache. Selbstverstandlich konnen in vielen Fallen Bolzen aus voUeni Material des zu untersuchenden Stoffes herausgearbeitet und in die Bolzenkathoden-Anord- nung als Objekt eingebaut werden. Wir haben aller- dings zurzeit diesbezugliche Erfahrungen nur mit hochschmelzenden Metallen Molybdan und Tantal gemacht. Zum SchluB zeigt uns Fig. 7 ein emissionsmikro- skopisches Bild, welches mit den Objekten in Bol- zenkathodenform hergestellt wurde. Richtstrahlwi'ite der kalten Kathode 41 LiTERATUR 1. Bas, E. B., Mitteilungen des Inst. f. techn. Physik a.d. Eidg. Technischen Hochschule Zurich, Nr. I (1950). 2. — G. F. F. Milleihinscn 10, 17 (1954). 3. — Optik 12, 377 (1955). 4. — Z. angew. Phys. 7, 337 (1955). 5. Honda, R. und Konno, S., Phil. Mag. 42, 121 (1921). Richtstrahlwerte der kalten Kathode L. Wegmann und M. Gribi Tiiih, Tciiihcr ii. Co., Ziirich Nachdem die Messungen des Geschwindigkeits- spektrums durch MoUenstedt (2) den theoretischcn Nachweis erbracht haben, daB die kalte Kathode nach Induni zur Speisung von Elektronenmikrosko- pen geeignet ist, konnte von der Anwendungsseite her gezeigt werden. daB es durchaus mogUch ist, das Trub-Tauber-Mikroskop fiir die histologische Forschung einzusetzen, wenn man als Kriterium dafiir die Auflosung der Doppelstrukturen in den Mitochondrien und der Palladeschen Korner be- trachtet. Was neben dem sicheren Vorteil der Ein- fachheit im Betrieb jedoch oft noch als Nachteil empfunden wird, ist die mangelnde HeUigkeit der kahen Kathode bei hoheren VergroBerungen, was bei langerem Arbeiten mit Schnittpriiparaten als ermiidend empfunden wird. Diese Lage hat AnlaB gegeben, die Intensitat des emittierten Elektronenstromes der kalten Kathode genauer zu untersuchen. Einige Ergebnisse sollen hier dargelegt werden. Fig. 1 zeigt das im Triib-Tauber-Mikroskop ver- wendete Beleuchtungssystem (1,3). Der Kathode (K) K B, Fig. 1. Beleuchtungs-Systcm mit kalter Kathode. K = Kathode A = Lochanode Ba = Anodenblende Be = Kondensorblende O = Objektebene a= virtueller kathodenseitiger OfTnungswinkel fi = maximale Beleuciilungs- apertur steht die eng anschlieBcnde Lochanode (A) gegen- iiber. Den AbschluB des Kathodcnraumcs mit er- hohtem Gasdruck bildct die Anodenblende (Bj\). In der magnetischen Kondensorlinse mit der Konden- sorblende ( Bf.) w ird das ElektronenbCindel gesammelt und — bei hellster Beleuchtung des I'riiparates — in die Objektebene (O) fokusiert. Die vorliegenden Ergebnisse wurden erzielt mit einer Lochanode von 8 mm Durchmesser in 9 mm Abstand von der Kathode. Der groBerc Abstand von der Kathode wurde notig, weil die Untersu- chungen auf hohere Spannungen (bis 80 kV) ausge- dehnt wurden. Der Abstand Anode-Anodenblende ist fijr die Entladung nicht kritisch. sobald er ober- halb ca. 30 mm liegt. Zur Messung der Richtstrahlwerte dieser Kathode wurden die Blenden B^ und Be, sowie — bei aus- geschaltetem Kondensor — eine Blende (Bq) in der Objektebene benutzt. Es zeigt sich rasch, daB man zu untcrschiediichen Werten kommt, wenn man die Apertur fiir die verschiedenen Blenden aus deren wirklicher Distanz zur Kathode berechnet. Dies ist aus dem Verlauf der Elektronenbahnen im Bcschleunigungsfeld leicht erkliirlich: nach Fig. 2 , muB der virtuelle, fiir die R xlO 56 kV 0.5 1 1.5 2x10 oc 3 Fig. 2. RichlstrahKscrtc dor kalten Kathode in Amp. /cm- als Funktion des kathodenseitigcn OtVnungswinkels. 42 W. C. NIXON Berechnung der Aperturen maBgebende Emissions- fleck hinter der Kathode liegen. Der Richtstrahlwert i?(a) fur einen bestimmten Offnungswinkel a ergibt sich aus dem durch eine Blende des Durchmessers 1 2 r flieBenden Elektro- nenstrom / zu 71 d^ n a wo a der zu r gehorige Off"nungswinkel und d der Durchmesser des Emissionsfleckes ist. Dieser Durch- messer wird am einfaclisten festgestellt durch elek- tronenoptische Abbildung in die Objektebene (O). Von hier aus kann er durch Rechnung in den Emis- sionsfleck zuriickprojiziert werden, und zwar in den virtuellen Emissionsfleck, dessen Abstand auch der Berechnung von a zugrunde liegt. Dabei wird der virtuelle Emissionsfleck in einem solchen Abstande von der Kathode gewahlt, daB sich fur Messungen mit Blenden in verschiedenen Abstiinden von der Kathode (z. B. Ba und Bq) dieselben Werte fur i?(a) ergeben. Fur die vorHegende Anordnung ergab sich auf diese Weise ein Abstand zwischen wirklichem und virtuellem Emissionsfleck von 100 mm. Bei einem Totalstrom der Emission von 0,5 mA erhielten wir damit die in Fig. 3 eingetragenen Richtstrahlwerte. Die Messungen wurden an einem mehrere Tage eingebrannten und somit iiber lange Zeit stabilen Fleck durchgefuhrt. Es ist auffallend, daB man es bei der Emission der kalten Kathode off'enbar mit einem Hohlstrahl zu tun hat, welcher bei einer Apertur von 1,5 10"^ seine groBte Intensitat ausstrahlt. Der Unterschied im Richtstrahlwert zwischen diesem Off'nungswinkel und den bisher benutzten von 0,3 bis 0,5 10"^ ist ofi'ensichtlich wesentlich groBer als die durch Varia- tion der Beschleunigungsspannung erzielten Unter- schiede. Immerhin kann auch mit Steigerung der Beschleunigungsspannung noch eine Erhohung des Richtstrahlwertes erreicht werden. Da bei der in Fig. 2 gezeigten Anordnung die Flachendichte j des Elektronenstromes im Praparat und damit die Helligkeit im Bild gegeben ist durch und anderseits fi gegeben ist durch die weiteren Abbildungsbedingungen, so kann die grdBte Hellig- keit dadurch erreicht werden, daB a = 1,5 ' 10~^ ge- macht wird. Es ist auf diese Weise gelungen, durch Abanderung der Distanzen zwischen virtueller Kathode und Kon- densor auf den giinstigsten Wert und durch gleich- zeitige Erhohung der Beschleunigungsspannung auf 60 kV im Triib-Tauber-Mikroskop, Typ KM4 eine Helligkeitssteigerung im Endbild von einem Faktor 6 zu erzielen. Die Forderung der Histologen konnte also erfiillt werden, und zwar zeigt die mit der neuen Beleuchtungseinrichtung erzielte Auflosung, daB die Geschwindigkeitsverteilung der Elektronen auch iiber diese Winkel geniigend ist. LJber weitere Experimente mit demZiel, die Richt- strahlwerte fiJr kleinere Offnungswinkel zu erhohen, soil an anderem Orte berichtet werden. LiTERATUR 1 . lNDUNi,G.,Comptes rendus du colloque CNRS a Toulouse 1955, S. 189. 2. MoLLENSTEDT, G. und DuKER, H.,Z. Natiirforsch. 8a, 79 (1953). 3. Wegmann, L., Optik 10, 44 (1953). Summary of the Proceedings of a Symposium on X-Ray Microscopy and Microradiography, Cambridge University, England, August 16—21, 1956 W. C. Nixon Cavendish Laboratory, Cambridge — CO 2— >— cc llm/'(z)=l; lim/'(z)^=0 (8) ist. Dann lautet die Definition der asymptotischen Brennweite v und der asymptotischen Brennpunkts- lagen z,,i und z,.! (der ,, asymptotischen paraxialen KenngroBen") lim 3— » + 0O s(.z) + Z + Zj,2 0; lim [/(z)-(z-z,i)] = 0; (9) und es gilt St - t's (10) KZI Abb. 1. Zur Festlegung der Losungen i(r) und / (r) der paraxialen Bahngleichung und zur Definition der paraxialen asymptotischen KenngroBen. 4 — 568204 Electron Microscopy OD/eHi Abb. 2. Zur Veranschauliciiung der Forderung, daR das virtuclle Zwischenbild in der asymptotischen Brennebene zvi des Projektivs liegen soli. Dann entsteht das reelle Zwischen- bild in der Ilbene r/i. Bei dieser Wahl von s und / gcben die beiden komplexen Zahlen A und B DurchstoBpunkt und DurchstoBrichtung der Einfallsasymptoten durch die asymptotische Brennebene z = z„i des Projektivs an. Wenn man also (und das muB man tun, wenn man das reelle Biid des Objektes in die Brennebene des Projektivs bringen will) Zi z,, wiihlt, also die Ebene des virtuellen Zwischenbildes (d. h. des Bildes. das das Objektiv erzeugen wiarde, wenn das Projcktiv ausgeschaltet ware) mit der asymptotischen Brenn- ebene des Projektivs zusammenfallen liiBt (vergl. Abb. 2), ist Wi = A; Ui B. (11) Fiir die Ausfallsasymptote. d. h. fiir groBe positive z gilt nach (7) Z — Zy9 u (z) = -A + Bv + +00 + '-^^ HOBJOd '"/ s(^)BA$)dl (12) Wirdefinieren die asymptotischen Bildfehlerkoef- fizienten B*, C*, c* ... durch // (z) + A Zn-l Bv lim Z — Zj,2 = 5* «;-//; : 2(F* //*)//j»i//i-!- + (F*-/7*)//^//,'- (D* C*)//,'^//[ + + (C* + ic*)uiu[ + {E* + ic*)u\i(^. (13) Der auf der linken Seite von (13) stehende Aus- druck ist die durch die ProjektivvergroBerung divi- dierte Abweichung der Seidelschen Bahn von ihrer paraxialen Nahcriing in dor weit vom Projcktiv cntfernten Endbildcbene. Man kann durch Vergleich von (12) und (13) unter Beriicksichtigung von (6) in bekannter Weise Integralausdriicke fiir die asvmptotischen Bildfehler- 50 FRIEDRICH LENZ koeffizienten gewinnen. Beispielsweise gilt fiir die asymptotische Verzeichnung E* .[i^^\(iAs't^(iP's^~s't- ■^'ss'^t dz, (14) |[g/,v-^ir')^»- ^ ff -^-fi'\s''s't + ^:^(i'ss'-^t-^fts'^t dz (15) Oder nach Vereinfachung durch einige partielle In- tegrationen E* 8v- + 00 \ii'^Y^(il^''^Y^(^'Y^dz, (16) - \[[i'^\(^")s'dz. (17) Als Beispiel wollen wir fiir den Grenzfall hoher VergroBerung die asymptotische Verzeichnung im Glaserschen Glockenfeld [5] 5 = B. if d. h. /? (^ mit A" ieB^dV \ 2 m V I (18) berechnen und mit der iiblichen Verzeichnung im System ohne Blende vergleichen. Wir erhalten fiir die isotrope asymptotische Verzeichnung mit der Abkijrzung oj = Kl + A^ 2k^ + 3 d'E* 4 or (4 k" + 3) 3 + Tik + ^,, ,o — rr sin- TT CO + -, — cotg.Tfo (19) 2 (4 k- + 3) 4 CO und fiir die anisotrope asymptotische Verzeichnung ^2 ^* ^ ^^ (2 k'^ + 1) - -^-^ sin 2 71 CO, (20) 8(o^ 16 ko/ wahrend die Berechnung der in (2) bzw. (3) einge- henden Verzeichnungskoeffizienten e, E im System ohne Blende d E = - — -o 1 +A sm 4o/ '71 \0J) 71 COtg— + CO 1 4A"^-3 + -- 24 4 A-^ + 3 6 (6 A" - 3) sm k sm — Oi CO + i . 9 TT + - sm^ — 8 CO 71 5k^ — sm' — 24 CO (21) und (2 A- + l)7rA- / 2 -2^ d-e = s^ 1 + A' sm — 8 CO'' k^ - \ . 71 71 sin— COS 4 A CO ca k or 71 sm — cos — 4 CO CO (22) ergibt. In Abb. 3 sind zum Vergleich die asymptoti- schen Verzeichnungskoeffizienten zusammen mit den gewohnlichen Verzeichnungskoeffizienten im System ohne Blende graphisch dargestellt. Die experimentelle Bestimmung der Verzeich- nungskoeffizienten geschieht meist durch Ausmes- sung der Aberration A//., im elektronenoptischen Bild (vergl. Abb. 4; dort ist zur Vereinfachung der Zeichnung nur die radiale Komponente von Amj = Ar gezeichnet). Haben dabei die in die Projektiv- linse einfallenden Elektronenbahnen achsenparallele Einfallsasymptoten, so gilt in unserer Bezeichnungs- weise Ar E* V (/.". (23) Waren dagegen die Elektronenbahnen beim Durchgang durch die Ebene des reellen Zwischen- bildes (der Ebene = - z,i) achsenparallel, so wurde — =Ef a. r (24) ? a'f^^ X / J "~-~-~^ 0 ^ 7 \ 2 \ K- Abb. 3. Die asymptotischen und die ublichen Bildfehler- Icoeffizienten der isotropen und anisotropen Verzeichnung fiir das Glasersche Giockenteld. Asymptotische Bildfebler 51 •u.E'tf'ra: v\ *\ y,- ~k< 3 \ \ Cg^£*v3^£/3 (28) Abb. 4. Abb. 5. Abb. 4. Zur Vcranschaulichung der Bcdeutung des isotropen asymptotischen Verzeichnungskoeffizienten E*. Abb. 5. In dicser Abbildung wird veranschaulicht, in welchem Linsenstiirkcnbereich die Anniihcrimg von £*i- durcb Co// brauciibar ist. In Abb. 5 sind E*v-, Ef- und die in der Literatur hiiufig niiherungsweise anstelle von £*v- verwandte GroBe C^j f fur das Glasersche Glockenfeld als Funktion der Linsenstiirke A- aufgetragen. In der Literatur sind mehrfach [1, 2, 3, 6] Metho- den zur experimentellen Bestimmung der Bildfehler durch Ausmessung verzerrter Schattenbilder be- schrieben worden. Wenn das schattenwerfende Objekt im feldfreien Raum vor oder hinter der Linse liegt, miBt man dabei in Wirklichkeit nichts anderes als die asymptotischen Bildfehler. Am bekanntesten ist die Messung des ,,OfTnungsfehlers" aus der Schat- tenbildverzeichnung. in Wirkiichkeit wird dabei aber nicht der OfTnungsfehlerkoeffizient gemessen, son- dern der Koeffizient der isotropen asymptotischen Verzeichnung. Da aber fiir schwache Linsen wegen + oc JO^'-^^-fi- c. + 00 (25) ( /r^.-) + 00 \i^f^rr-l,i>ir)'>^ + 00 (26) [fd: f^v /'• (27) dz besteht, sind die cntsprechcnden Messungen nur mit einem relativ geringcn Fehlcr behaftet und durchaus zur niihcrungswcisen Bestimmung des OlVnungsfchlcrkocftizionten C'^ gecignet, und zwar umso besser, je schwacher die Linse ist. Die von Liebmann [7] durchgefiihrten numcrischen Berechnungcn der Vcrzeichnungskocffi/icnten bczie- hon sich auf die asymptotische Verzeichnung. Es ist aus diesem Grund nicht verwundcrh'ch, daB ihre Abhiingigkeit von der Linsenstiirkc einen grundsiitz- lich andcrcn Verlauf hat als die, die sich aus den Glaserschen Bildfehlerintcgralcn crgeben wiirde. Die Abhiingigkeit der von Liebmann berechneten Ver- zeichnungskoeffizienten von der Linsenstiirkc ist viel- mehr von demselben Typus wie die Kurvcn fijr E* und ('* in Abb. 3; d. h. E* hat fiir k- --- 0 einen positiven cndlichen Wert, steigt mit zunchmcndcr Linsenstiirkc zuniichst etwas an, durchliiuft bei etwa einem Sechstel der teleskopischen Linsenstiirkc cin Maximum, wechselt etwa bei der halben teleskopi- schen Linsenstiirkc sein Vorzeichcn und geht mit Anniiherung an die teleskopische Linsenstiirkc gegen — oo . Als Beispiel fiir das Zusammenwirken der Bild- fehler zweier Linsen werde die isotrope Verzeichnung eines aus zwei Linsen bcstchcndcn Systems bercch- net. Durch Beschriinkung auf die isotrope Verzeich- nung wird aus (2) und (13) U.y = Vp U^ + Vp £* Ml M J (29) (30) Setzt man Hi aus (29) in (30) ein und vernach- liissigt Glieder von hoherer als 3. Ordnung in u„, so erhiilt man «2 = K ^P "o + iylE* + £o) Vo Vp I'l Wo • (3 0 der auch in Abb. 5 erkennbare Zusammenhang Fiir hohe ObjektivvergroBerungen ( | F,, | ; 1) wird in (31) das Glied, das die Objektivverzeichnung £„ cnthiilt, nur cine vernachiiissigbar kleine Rolle spie- Icn. Die Gesamtverzeichnung des Systems ist in diesem Fall also praktisch nur durch die asympto- tische Verzeichnung des Projektivs bestimmt. Liegt dagegegen in einem System aus mehreren Linsen vor dcm letzten Projektiv noch cine oder mchrere Linsen geringer VergroBerung, so ist auch deren Verzeich- nung (wcnn cs sich um F'rojektivlinsen handelt. deren asymptotische Verzeichnung) zu heriicksichtigen. Literatur 1. VON Ardenne, M.,Z. Physik 117, 602-611 (1941). 2. BoERSCH, H., Z. tech. P/iys. 20, 346-.'?50 (1939). 3. DossE, J.,Z. Physik 117, 722-753 und 118, 375-383 ( 1941). 4. Glaser, W., Grundlagcn der Elcktronenoptik. Springer- Verlag, Wicn, 1952. 5. — Z. Physik 117, 285-315 (\')4l). 6. Heise, F., Opiik 5, 479-489 (1949). 7. Liebmann, G., Pioc. Phys. Soc. B 65, 94-108 (1952). 8. RusKA, E., Arch. Elektrotech. 38, 102-130 (1944). Zur Errechnung elektronenoptischer Feldverteilungen mit geforderten Abbildungseigenschaften K. W. J. PiCHT Institiit fiir f /won't ische Pliysik cler Pad. Hochschule, Potsdam-Giiebtiitzsee In der Elektronenoptik ist es bisher im allgemeinen iiblich, von einer gewiihlten Feldverteilung langs der Symmetrieachse auszugehen — z. B. von der soge- nannten Glockenkurve — und fiir diese zunachst Brennwelte und Lage der Hauptpunkte sowie da- durch gleichzeitig die Lage der Brennpunkte zu berechnen. Aus diesen GroBen findet man auch so- fort die fiir eine gewiinschte VergroBerung der Abbil- dung erforderliche Lage der Objekt- sowie der Bildebene. AnschlieBend lassen sich dann fiir diese Bildebene auch die Abbildungsfehler berechnen. Wiihlt man die Feldverteilung langs der Achse — aus der sich ja die Feldverteilung im Raume berech- nen laBt — so, daB ihre mathematische Darstellung noch einige zunachst in ihrer GroBe nicht festgelegte Parameter enthalt, so werden diese Parameter auch in den Ausdriicken fiir die Abbildungsfehler auftre- ten. Sie konnen dann gegebenenfalls nachtriiglich so gewahlt werden, daB die oder doch einige der fiir den gedachten Anwendungszweck des elektro- nenoptischen Abbildungssystems besonders storen- den Abbildungsfehler moglichst kleine Betriige an- nehmen. Ob dies Verfahren bisher in der Literatur bereits vorgeschlagen oder praktisch tatsiichlich benutzt wurde, entzieht sich meiner Kenntnis. Auf der 6. Tagung der deutschen Gesellschaft fiir Elektronenmikroskopie in Miinster im Miirz 1955 hatte ich zur systematischen Errechnung der Feld- verteilung von Abbildungsfeldern, die bestimmte Abbildungseigenschaften sowohl bez. der VergroBe- rung und des Abstandes zwischen Objekt- und Bildebene als auch mit Bezug auf eine moglichst geringe GroBe der bei dieser Abbiidung auftretenden Abbildungsfehler besitzen, eine andere Berech- nungsmethode vorgeschlagen, die in der Zeitschrift Optik, 12, 433, 1955 veroffentlicht wurde. Bei dieser damals vorgeschiagenen Berechnungs- methode war ich ausgegangen von der 1932 von mir angegebeneni Form der Differentialgleichung paraxialer Elektronenstrahlen P" + 3 /O)' 16 ^-) P-0 fiir elektrische Felder, bzw. 3 /0'(z)\2 1 W"{z)+W{z) 16\ CD-/ -^cP^,^''^^^ 0 In diesen Gleichungen ist -e (mit e > 0) = Ladung der Elektronen, / •- _ P(z) = o(z)l'0-(z) bzw. ^r(z) = H'(z)l/(D"(z) mit H'(z) = ^(r) cos ;^(z) ; /o (r) sin ;/(z) = ^(z)e'^ X = y)-a) O) = 0)n^ " 2 f ImJ 10" (z) m{z) dz = - e 2 m J mi=)dt. {dco/dt = Larmor-Prazession,] 1 m o (D-(z) = (D(z) - t/= :, V", 2 - e O (z) = elektrische Potentialverteilung langs der z- Achse, f93(z) = magnetische Feldstarke 93 langs der z- Achse. Q, ip, z = raumfeste Zylinderkoordinaten (der Bahn der Elektronen), deren z- Achse mit der Symmetrieachse des Feldes zusammenfallt, q,X,z = Bahnkoordinaten der Elektronen in einem sich um die z-Achse (Symmetrieachse des Feldes) mit der Larmor-Prazession drehen- den Zylinderkoordinatensystem. Die Methode der 5erechnung bzw. Errechnung einer geeigneten Feldverteilung, die bei vorgege- benem Abstand zwischen Objektebene und Bildebene und einer hierbei geforderten VergroBerung eine Abbiidung mit moglichst geringen Bildfehlern liefert, besteht darin, eine von mindestens einem, besser von mehreren Parametern Pi abhangende Kurve 1) P =P(z) 'V{z,p^,p..,...pn) =0 als Bahnkurve der paraxialen Elektronenstrahlen so zu wahlen, daB sie die z-Achse in zwei Punkten schneidet, deren Abstand dem Abstand zwischen fiir magnetisch-elektrische Felder. 1 J. Picht, Ann. Physik (5) 15, 926-964(1932). J. Picht, finfiihrung in die r/;eorie der Elektronenoptik (E Th EO). Johann Ambr. Barth, 2. Auflage, 1957, § 14 u. § 20. Errechmmg elektronenoptischer Feldveitcilungcn 53 Objektpunkt (z =7^) und Bildpunkt (z = z,,) ent- spricht, daB also 2) P(z,) ^ 0 und P(r,) 0 und daB sie die r-Achse in diesen Punkten so schneidet, daB 3) mit B - laterale VergroBerung. AuBer diesen drei Forderungen muB das zu wiih- lende P(z) [bzw. W{z)] noch folgende Bedingungen erfiillen: 4) P" U) P(z) ^0 fiir Zg ^z <,zt, . Aus Forderung 2) und Forderung 4^ folgt, daB auch gelten muB: 4 a) P"(zg) = P"(z<,) = 0 mit lim P" (z) 2->2g P(Z) C^O mit C = constans Die drei fiir die angegebenen Teilabschnitte zwi- schen z z„ und z z^ geltenden Bahnfunktionen gehen — wenn man die eingekiammerten -Zeiclwn als giiltig annimmt und noch a-z„ 2zj, - .-rB 2 fl(3B : 1) ■ V^B und lerner: 2jZ c = a(B+l)' h = 7i B B+1 ' Zo ( = Zj + c) = f I 1 waJTlt, also 2 Zb - .T • B 2.T «(Bfl) B-1 IV Az) = B 1+3-B 2a z + + sm a (B \)(z z.) W,{z)- -^a( 5) Die mit diesem P(z) [bzw. lV{z)] berechnete Feldverteilung (I>(z) bzw. (oder: und) ''-^(z) muB praktisch — u. zw. moglichst einfach — realisier- bar, also herstellbar sein. Die vorgeschlagene Berechnungsmethode wurde inzwischen unter meiner Anleitung von einem mei- ner Schiiler, meinem derzeitigen Assistenten Herrn Dipl.-Phys. Horst Hansel, in einer als Doktor- Dissertation eingereichten Arbeit fur ein Magnet- feld auf ein spezielles Beispiel angewandt, iiber das zunachst kurz berichtet sei. Sind Zg und Zf, wieder die langs der Symmetrieachse des Feldes gemessenen Koordinaten der (achsen- senkrechten) Objekt- bzw. Bildebene, so wird in der Arbeit von Herrn Hansel fiir den paraxialen Strahlverlauf zunachst angenommen, daB er mit Bezug auf ein sich mit der Larmor-Prazession um die Systemachse (z-Achse) drehendes Koordinatensy- stem darstellbar ist durch die drei folgenden. ab- schnittsweise geltenden Gleichungen W^iz) = a(z-Zg) mit Zg^z,_^^z^ I a Zgj, W^ (z) = azj^ + - z -f sin ( \ c I mi t Zj ( ^ ) z ( r^i ) Zo ( z^ '■■€), ^3iz)= -^a(z-z,) mit (Zi-rC = )Z2(^)Z(^)Z6 schreibt — an der Stelle z z, bzw. Z = Zi + C = Za stetig ineinander iiber, so daB z = Zi und Z = Zi + C = Za als ,,Nahtstellen" der drei Teilkurven angeschcn werden konnen. Geometrisch besagt dies, daB der Strahlverlauf zwischen Zg und z, einerseits, zwischen z.. und z,, ande- rerseits als geradlinig angenommen wird, (z., - z,) c also gewissermaBcn die eigentliche ..Feldliinge" dar- stellt. Da dies bedeulcn wurdc, daB das magnetische bzw. magnetisch-elektrische Feld an den Stellen z Zj und z Z2 abbricht, dies aber in der Praxis nicht der Fall sein kann, muBte der Strahlverlauf und damit auch die Feldverteilung noch etwas moditi/.iert wer- den. Zu diesem Zwecke wurde der zunachst vorgege- bene Strahlverlauf dahin geandert, daB die oben angegebenen Funktionen Wj{z) mit / 1 h/w. 3 nicht bis Zi (fliry I ) und auch nicht von z-, z, c ab (fury 3) gelten, sondern nur bis zu (bzw. von) diesen benachbarten Werten (ab). Aus diesem Grunde sind oben bei Angabe der Giihigkeitsgren- zen der Teilfunktionen die Glcichheitszeichen be- reits eingeklammert. Bezeichncn wir die ncuc obere Grenze fiir IV^iz) durch Zi ti, die neue untere Grenze fiir H',(z) durch Zi ■• f>, so ist der gewiihlte paraxiale Strahl- verlauf jetzt darstellbar durch 54 K. W. J. PICHT W^{z) = a{z-z,) mit Za<.ZQ angeben, in dem noch x{z)= - {z-z^ c ist. Fur die magnetische Feldstarke findet man dann die drei stetig ineinander ubergehenden Ausdriicke: fur Tl Ei^{z-Z^^O ^ -e y a{z-Zg) + f{z-z^ fiir 0 j:(z z^)^c ) B, (z) = 8mf/ 1^2 (B+ 1)2 sin [ia(B+l) {z-z^)]- -f"iz-z,) sin[ia(B+lKz-Zi)]-ia(B-l)(z-Zi) + -\ a{z-^^-Zg)^f{z-z-^) und fiir c ^{z -z^ ^- £2 71 BAz) 8m U f"(z-z^) (z-Zh)+fiz-Zj) B In diesen Ausdrijcken fiir die magnetische Feld- starke auf der Symmetrieachse des Feldes bedeuten wie iiblich m die Masse, e (mit e > 0) die Ladung des Elektrons, U h ('" 'e)vl, ferner Zg = Lageko- ordinate des Achsenschnittpunktes der Objektebene, Zb entsprechend der Bildebene, B = paraxialer Abbil- dungsmaBstab, d. h. die gewunschte VergroBerung. a ist eine Konstante, die prinzipiell noch frei verfiig- bar ist, um durch ihre Variation die Abbildungs- giite zu beeinflussen. Ihre geometrische Bedeutung folgt aus IVi(z) = a(z Zg), namUch, daB a = dem Tangens des Neigungswinkels der den Berechnungen zugrunde gelegten Elektronenbahn in ihrem objekt- seitigen Schnittpunkt mit der Achse, also an der Stelle der Objektebene ist. Zi bezeichnet die z-Koor- dinate der Stelle der Symmetrieachse, an der — in gewissem Sinne — das magnetische Feld zu wirken beginnt oder als vorhanden zu gelten hat. /(z) ist die ,,Korrektionsfunktion", die das abrupte Ab- brechen des Feldes an den Stellen z = Zi und z = Zi + c verhindert. c bezeichnet daher — in gewissem Sinne — die ,,Lange" des magnetischen Feldes. Zur Beeinflussung des Korrekturzustandes dieses Feldes kann man — he i fester Vorgabe des Abstandes {Zf, " Zg) zwischen Objekt- und Bildebene sowie ferner: der Vergrofierung B — folgende GroBen prinzipiell variieren: 1) a = tang der Neigung des zur Berechnung benutzten Elektronenstrahls im Ach- senschnittpunkt der Objektebene, 2) c = „Feldlange", 3) Zi = Stelle des „Feldanfangs". Tn manchen praktischen Fallen wird es moglich sein, auch noch 4) z^ - Zg = Abstand der Bildebene von der Objek- tebene zum Zwecke der Bildfehlerkorrektion zu variieren. In der genannten Arbeit von Herrn Hansel sind derartige Fehlerberechnungen durchgefiihrt und der EinfluB einer Variation der freien Parameter auf die Bildfehler diskutiert. Doch auf diese Diskussion sei hier nicht niiher eingegangen. Wohl aber wollen wir hier noch kurz an Hand eines Beispiels zeigen, daB es nicht unbedingt notig ist, von einer Elektronenbahn auszugehen, die aus mehreren abschnittsweise geltenden Gleichungen darzustellen ist. In der oben genannten Annalen-Arbeit aus dem Jahre 1932 hatte ich darauf hingewiesen, daB eine Losung der fur elektrische Felder geltenden Differen- tialgleichung 3 /(D'\- 16\0~/ fur die Funktion O 2p 2p Lentilles electroniques magnetiques 2p 55 P = l/z-{Ci-/p {±2pAV3yz) 2p + C,'Nj,(±2pAV3yz)} ist. Geht man nun umgekehrt von dieser Bahnglei- chung P P(r) aus und modifiziert sie so, daB sic einem vorgegebenen Abstand z,^~Zg zwischen (par- axialer) Bildebene und (paraxialer) Objektebene sowie einer gewunschten VergroBerung B angepaBt ist, so wird man — wie wir hier ohne Bevveis mitteilen^ — zu folgenden Gleichungen gefiihrt: 2p rJ±2pA*l/a-^ + 2p C*-N, (..... /.i;f^)^o. 2p Jp ±2pA- a Zb Zb-Z, 9 I + 2p + C*nA ±2pA* Zb-C a Zb 0, 2p ';>-! 2/7/4' f + C' + Zb •A^P-i( 2p ±2pA* y a Zb-C •■ip Jp-x{^ ±2pA* a Zb Zb~Zg + 2p C*-N, '-{'"^■\'-m) mr exp — Dp. \3pA* Diese Gleichungen enthaltcn die 7 (bzw. 8) Para- meter i!;> a (bzw. «, A*), C*, /;, c\ Zy, (r^ Zg) - h, B ?■, Zg und von denen die bciden letzten, niimlich Z(, B!'p durch die Aufgabenstellung vorgcgeben sind. 3 weitere GroBcn sind dann durch das Bestchen der 3 Gleichungen bestimmt, so daB noch 2(3)der GroBen — z. B.«,coderzg,/4*oderr, C* oder /4*,/7, r^usw. — frei wiihlbar sind und so bestimmt werden konnen, daB die Abbildungsfehler moglichst gering werden. Das zugehorige Feld, das die den Betrachtungen zugrundegelegte Elektronenbahn liefert, ist dann bestimmt I. im rein elektrischen Fall durch 2p O (z) = C exp 8 3 3 A- iz-c) m + -2e Vq 2. im elektrisch-magnetischen Fall durch m 0(z) = — -[(^(z))^ + ^;^] 2e V^{z)^ t- met) W" w 3 /(IV 1 6 \ O" )- mit \V" W P P Setzt man speziell p = \, also /? - 1 « 0, so muB man, um fiir B/- einen moglichst groficn negativen Wert zu erhalten, fiir C* einen moglichst kleinen (positiven) Wert wiihlen. Lentilles electroniques magnetiques Regies de leur construction et expressions universelles de leurs caracteristiques electro-opliques P. DURANDEAU Laboratoire d'Optique Electronique La construction rationnelle des lentilles electroni- ques magnetiques demande la connaissance : {a) des regies qui decoulent de Tetude de leur aimantation: {h) des expressions de leurs caracteristiques electro- optiques. Circuit magnetique — forme de la topographic dii champ. — II est essentiel de rassembler les amperes- tours d'excitation d'une lentille magnetique selon un champ presque uniquement localise dans ia region de Tentrefer : ce resultat n'est pas obtenu avec un circuit magnetique mal etudie (fig. I). On aboutirait a cette concentration si la pcrmeabilite magnetique ctait infinie mais, en pratique, il suffit que Tinduc- tion soit inferieure a 12.000 gauss. 56 P, DURANDEAU CxcitotiOn nl ,0000 ctirtp t ,.,, CntrrfrK 5 w i mm Fig. I. Pour un circuit magnetique mal etudie un champ non negligeabic cxiste dans le canal des pieces poiaires. Ce champ doit etre elimine pour plusieurs raisons (2). La forme a donner (fig. 2) au circuit magnetique (2) pour qu'il en soit ainsi, tient compte de la varia- tion du flux le long des noyaux poiaires, et du circuit (2, II). C'est rinduction dans le nez des pieces poiaires qui conditionne ensuite Tetalement de la topogra- phic du champ. (a) Si Texcitation est inferieure a 1000 S amperes- tours {S, entrefer en millimetres) I'experience montre (2) que, quel que soit le diametre des pieces poiaires, la topographic fournie par la lentille est celle qui correspond a une permeabilite magnetique infinie (fig. 3), cas pour lequel le calcul est aise (I, 6, 7, 9). Dans ce cas la topographic du champ est assimi- lable a celle que donnerait un solenoi'de de meme longueur S que I'entrefer (3) de rayon egal aux deux tiers du canal et evidemment de meme excita- tion que la lentille (fig. 4). La valeur du champ maximum est alors : ^i , N ^ \ \ N WW \ ;^^ \ f ' ^ J r Fig. 2. Forme du circuit magnetique qui permet d'obtcnir le rassemblement des amperes-tours d'excitation de la len- tille comme si la permeabilite du fer etait infinie : («) forme tronconique des noyaux poiaires : angle de 1 0 a 1 5 " . (/>) Epais- seur £>! au moins egale a /•/2. (c) Epaisseur e.2, telle que Taire de la section de I'enveloppe exterieure soit du moins egale a .t/-. \ V/ N\^ '^ r* — ' -^^ 1 ^Vx^ ^ Fig. 3. Ces pieces poiaires percees d'un canal de meme dia- metre D et espacees d'un meme entrefer S donnent meme topographic pour une meme excitation inferieure a 1000 5 ampere-tours (5, en millimetres). Cette topographie est celle qui correspond a une permeabilite magnetique infinie. AnNI 'M- K^- + 0.45Z)' Le principal interet de cette assimilation est de faire apparaitre un parametre geometrique carac- teristique de la lentille : L = VS" + 0,45 D^. L'expression de la topographie qui resulte de cette assimilation est generalisable a une lentille dissymetrique (fig. 5). ib) Si Texcitation est superieure a 1000 5" tout se passe comme si 5 et D etaient multiplies (fig. 6) par le facteur : /;; NI-WOOS T0005 ni est un coefficient numerique qui depend de la forme des faces poiaires. II est minimum et proche Fig. 4. Pour une permeabilite infinie et en pratique pour une excitation inferieure a 1000 S amperes-tours, la topographie obtenue est assimilable a celle d'un solenoide de longueur 5, de diametre - ., D et d'excitation NI. En tout point de I'axe rinduction a pour expression B iTlNI (cos a + cos /3). LentiUes electroniques mai^tu'tiques 57 Fig. 5. Generalisation a line lenlille dissymetrique dc Texpres- sion de la topographic qui resulte de {'assimilation de la lentillc symetrique a un solenoTde. Lc paramctre caractcristi- quc est; y > + 0,45Z)i + d\. de 0,15 dans les conditions donnees par la figure 7 (2). (c) La figure 7 donne egalement la condition que doit satisfaire la distance a partir de laquelle il est possible d"elargir le canal perce dans les pieces polaires, sans affecter la forme de la topographic (2). Expressions universelles dcs caracteristiqucs electro- optiqiics. — Les caracteristiqucs electro-optiques des lentilles ont ete calculees par differents auteurs (7, 8, 9). Les resultats de ces calculs sont donnes par des faisceaux de courbes assez mal commodes a consulter, aussi a-t-on cherche une representation unique d'utilisation aisee (10). Dans cet ordre d'idees nous avons propose des expressions universelles utilisant la notion de lentilles dans des etats d'excita- tion correspondants (3, 4). 8"lf»""' oinirnjjtf duchamp ma'' <*"""' Pi fcrr po/a I re f I ^r doi^'j 4 000 * 000 ttcilalion (IT) Fig. 6. Pour les excitations inferieures a 1000 S amperes- tours j'etaicmcnt de la topographic est invariahle. Pour les excitations superieures a 1000 5 il varie lineairemenl en fonc- tion de Texcitation. Fig. 7. Pour .\7 1000 .S' la topographic est de largeur mini- mum si : {a) a compris entre 55 et 65*^, {h) e prochc de £)/4, (c) 5i> Nil 500. Par aillcurs si ,\7 1000 .V il faut H>\/S' + D'; si Nr> 1000 S il faut r Ni-\ooosi Ces expressions sont fournies par le cas type du champ uniforme limite a rintervallc entre deux plans (fig. 8). Soit N i„ Texcitation pour laquelle la lentillc uti- lisee en projecteur fournit la longueur focale mini- mum /,„. Les caracteristiqucs electro-optiques prin- cipales ont alors les expressions suivantes, NI etant Texcitation de la lentillc. d")/,, longueur focale de la lentillc utilisee en projecteur 4 sin 2,03 sin 2,03 2,03 radians = 116. NI, NL Fig. 8. Trajectoire d'un electron dans un champ uniforme B limite a rintervallc entre les plans P„ et Z',. to J 8 Wo V* 58 P, DURANDEAU (2°) /„, longueur focale de la lentille utilisee en objectif ; (a) si NI<0,11 NIo, Jm Jrn sin 2,03 sin 2,03 -— — (/>) si NIOJINIq, /o sin 2,03 Nh Jm NI n7. (3°) z„, distance du foyer F„ au plan median (a) si NI<0J7 ^/q, z„ 1/1 A^/\ ^ = cotg 2.03 —- ; fm 0,55 2 ^ NlJ (h) si NI-OJl TV/q, ^n 1 /Jt 1 1^ /,„ 0,55 1 2 A^ 2 (4") C coefficient d'aberration chromatique (a) si NI<0J1 7V/o, Cc-^/ol 2,03 sin 2,03 ^— - cos 2,03 NI. NI n7. 0 (/;) si A'/. 0,77 A^/„, Q. = 0,785/o. Pour le champ uniforme limite a Tintervalle entre deux plans : N!„ 10.01 V* et/,„ = 0.55 L. Ces expressions sent vakihlcs avec line approxima- tion inferieure a 5% dans tons les cas, a condition de donner a Nf„ et a /,„ les valeurs qui correspondent a la topographic fournic par la lentille donncc. Dans les cas usuels (0,5 < Dj S<2) on a : ^'/o=13,5l/V^ l/7^ /„,^ 0,5 KS2 + 0,451)2 La figure 9 donne, pour ces cas usuels, la valeur des rapports des caracteristiques electro-optiques au parametre \ S- + 0,45 D^. Pour la commodite prati- que il est preferable d'utiliser ce parametre plutot que/,,,. Z.i — H 1 L-. Ji'*0.t,i Di ; >. !■ ■ L 2 2.io L 5. SO . L ;1\ 4. Ic \ L 1 S ■\\ J. C^ \ 1 L ;^ yy 0 5 \ ^^*'^~ _.^^ \-^^ ^^\ '••'y- 5 0 c .1 1 ,5 ni Fig. 9. Variations du rapport des caracteristiques electro- optiques fondamentales au parametre de la lentille pour les cas usuels (0,5 < Z)/5< 2). La courbe de variation du coefficient d'aberration de sphericite C, est celle qui resulte de la reduction de la valeur de ce rapport calcule pour les cas usuels et non la valeur donnee pour le cas type du champ uniforme (5). En introduisant dans les expressions ci-dessus, le facteur d-elargissement 1 i m\,(Nl - 1000 5)/ 1000 5] faisant intervenir le degre de saturation des faces polaires on montre que la longueur focale /j est minimum pour un entrefer S calculable en fonction de D et de m (5). Conclusion. — Le probleme de la construction des lentilles electroniques magnetiques semble ainsi re- solu dans ses lignes essentielles pour toutes les ten- sions d'acceleration. BiBLIOGRAPHlE 1. DuRAND, E., Ann. phys. 12*^ serie. 10, Nov. Dec. 1954. 2. DuRANDEAU, P., J. plivs. icicliiini 1956, 17, 18 A. 3. — ibid. 1956, 17, 33 S. 4. — Conipt. rend. acad. sci. lAl, 1710 (1956). 5. — These a paraitre. 6. Hesse, M. B., Proc. Phys. Soc. B 63, 386 (1950). 7. Lenz, F., Z. angew. Phys. 2, 448 (1950). 8. — Optikl, 243 (1950). 9. LiEBMANN, G., Proc. Phys. Soc. B 68, 737 (1955). 10. LiEBMANN, G. and Grad, E. M., Proc. Phys. Soc. B 64, 956(1951). 11. MULVEY, T., Proc. Phys. Soc. B 64, 441 (1953). Ill ELECTRON-SPECIMEN INTERACTION Uber die Entstehung des Kontrastes im elektronenmikroskopischen Bild B. V. BORRIES t UND F. LeNZ Rheinisch- Westfdlisches Inst it ut fiir Obermikroskopie, Di'isseUhif, mid Inst it tit fiir Elektionenoptik unci Feinmechanik der Technischen Hochschule Aachen Ziisanvnenfassiini,'. — Es wild iiber neuere Fortschritte in den Vorsteliungen iiber die eiei is the electron energy in electron volts, and 0 the scatter angle. It will he noted that the cross section is independent of atomic number. in the present experiments, the angular scatter measurements were made directly in the electron microscope. The arrangement was even simpler than that of Biberman et al. [2], who had to remove the projector polepieces and certain screens. The microscope was unmodified, except that the objec- tive was worked at long focal length (1 cm), and a small screening aperture immediately above the object was employed. The condenser lenses were switched off. so that a relatively low beam intensity was employed. This reduced the contamination rate of the specimen, and facilitated the timing of the exposures. The angular scattering distribution of the electrons after passing through the film under test was reproduced in the back focal plane of the objective lens. The projector lenses were used to project an enlarged image of this distribution onto the photographic plate. The photographic plates used (ilford Thin Film Half Tone) were known to have a linear density cur- rent density response over the range 0.2 to 1.0 blackening density, from the work of Digby, Firth and Hercock [4]. A series of exposures (covering the range 1-60 sec) enabled the whole scatter pattern to be recorded at densities falling between these limits. The incident electron intensity was found by allowing the beam to pass through a hole in the specimen, and by photographing the patch of illu- mination at a known instrumental magnification, it was then unnecessary to know the characteristics of the photographic material, as the scattered inten- sity could be related directly to the incident intensity in terms of the blackening density on the photo- graphic plate. The scatter patterns were analysed w ith a microphotometcr. In order to minimise the effect of contamination, a small screening aperture was placed immediately above the object, thus restricting the illumination to the area under observation. It was thus possible to search for a clean and undamaged portion of film without building up a layer of contamination on the specimen. The area of film from which the scatter was recorded was defined by an intermediate aper- ture of known size corresponding to an area of 5 sq. micron at the object. The first experiments were carried out with thin carbon films of known thickness [1]. By measuring the scatter intensity before and after a timed expo- sure to the electron beam, it was possible to calibrate the contamination rale under the conditions under which the film was being examined. This was found to agree very well with the figures given by Ennos a — .568204 Electron Mirroscopi/ ^6 M. E. MAINE AND A. W. AGAR ^1* xlO 300 1' 2 SO o 200 ISO -lOO so a o Id 2 ANGLE e 2x10 3xlO Fig. 1. Differential scattering cross section for carbon tilms of differing thickness. [5] and amounted to about 3 A min. Thus, except for the thinnest films, the scatter pattern could be recorded before the film thickness had changed appreciably. The angular distribution of scattered intensity as obtained from the blackening of the photographic plates was converted into the differential scattering cross section by dividing by the number of atoms per sq. cm of the scattering film, i.e. assuming single scattering conditions, which are the basis of the scattering theory. The additional scatter due to a thicker film should be exactly balanced by the dividing factor which includes the film thickness provided that the scat- tering mechanism remains unchanged. A series of carbon films of different thickness were analysed in this way, and the calculated differential scattering cross-section plotted in figure I . The dotted curve shows the theoretical curve, based on Lenz's expressions for elastic and inelastic scatter, the approximate expression (I) being used for calcula- tion of the inelastic cross section. It will be seen that, while the results agree within a factor of 2 with the theoretically predicted ones, the curves do not lie on top of one another, but show a definite gradation with the film thickness employed. The difference between the curves is too great to be ascribed to faulty measurement of film thickness or incorrect contamination rate (the results for the thinner films have been corrected for the measured contamina- tion rate). The "reference thickness'" or "transparency thick- ness" defined by von Borries as the thickness of film in which every incident electron is, on the average, scattered once elastically, is about 450 A for carbon, it might be expected that carbon films thinner than this would give similar results for scattering cross section. Since they do not, it suggests that, at each thickness, there is a rather higher pro- portion of electrons scattered outside the effective aperture than is allowed for in the theory. Since the inelastic cross section was expected to be xlO 300 1 2SO 1 .---' ( 1 ) Theorciicol 'Elastic + Inelastic) (Atler Ltnz} ( 2j Silver 60A. ' Correciid tor Carbon Film ) ■o|-D 200 IV'^ (3) Carbon 55 A. ( Corrislriciun!^ der gestreiilen Elcktroncn. — Fur die Registrierung der Streuelektronen wurden photo- graphische Flatten benutzt. Die photographische Methode hat gegeniiber der direkten Strommessung mit einem Faraday-Kiitig den Vorteil, daB sie eine genauere Ausmessung der Stromdichten gestattet, als es mit dem Faraday-Kiifig moglich ist, da dieser aus Empfindlichkeitsgrunden eine mehrere Quadratmilli- meter groBc OfTnung haben mUBte. Weitcrhin kon- nen auch sehr kleine Intensitiiten durch lange Belich- tungszeiten noch registriert werden. SchlieBlich wer- den alle Streuwinkel gleichzeitig aufgenommen, wo- durch auch bei Stromiindcrungen wiihrend der Be- lichtung die relative Stromdichtevertcilung die glei- engster Strahlquerschnitt Obiektblende a ("+/') Blendenradius Abb. 4: Zur Bcrechnung des Radius der Primarstrahlspur im Abstand a von der Blende. 70 G, KEMPF UND F, LENZ che bleibt. Ein Nachteil besteht darin, daB die Stromdichten auf dem Umweg iiber die Platten- schwarzung erhalten werden, wodurch eine zusatz- liche Fehlermoglichkeit entsteht. Als Aufnahmematerial wurde die Perutz-Silber- eosinplatte wegen ihrer giinstigen Empfindlichkeit verwendet. Ihre Gradationskurve hat eine mittlere Steilheit; einerseits ist sie flach genug, daB bei einem Belichtungsintervall von einer Zehnerpotenz die Schwarzungen in einem bequem auszumessenden Bereich lagen, andererseits aber auch steil genug, um kleine Belichtungsunterschiede noch geniigend deut- lich wiederzugeben. Die Aufnahme der Gradationskurve geschah durch stufenweise Belichtung einer Platte im Elek- tronenmikroskop. Durch einen verstellbaren An- schlag fiJr die Transportkurbel fiir die Plattenkasset- ten wurde der Kassettendeckel wahrend der Bestrah- lung stufenweise geoffnet und so auf der Platte Streifen bei der gleichen Bestrahlungsintensitat niit verschiedenen Zeiten belichtet. Die gleichmiiBige Stromdichteverteilung in der Registrierebene wurde durch Starke VergroBerung des Strahlquerschnitts mittels der Elektronenlinsen hergestellt. Die Platten wurden dann zu je 12 Stuck in einem Plattenkorb in jeweils frisch angesetztem Agfa-Methol-Hydrochinon in der Verdiinnung 1 : 10 bei 20 C 10 Minuten lang unter standiger Bewegung entwickelt. Die Tempera- tur schwankte nicht mehr als ±0,1 C. Die jeweils gleiche Entwicklerkonzentration wurde durch ge- naues Abmessen des konzentrierten Entwicklers und des Wassers zur Verdiinnung eingehalten. Das absolute MaB der Stromdichte, mit der die Stufenkeile belichtet wurden, wurde nicht ermittelt, da bei den Streuexperimenten der Primarstrom nicht gemessen werden konnte, sodaB nur die relative Ver- teilung der Streuintensitaten auf die verschiedenen Winkel bestimmt werden konnte. Darum kam es auch bei der Gradationskurve nur auf die Kenntnis des relativen MaBes der Bestrahlung an. Fiir die Streuaufnahmen wurden eine groBere Menge der Platten derjenigen Emuisionsnummer. auf die die Stufenkeile fiir die Gradationsbestim- mung aufgenommen worden waren, aufbewahrt. Spiiter stellte sich heraus, daB wahrend der Lagerung die Gradation der Platten mit der Zeit erheblich llacher und der Schleier starker wurde. Es wurden deshalb fur die Streuaufnahmen immer frische Plat- ten verwendet und bei der Entwicklung zusammen mit diesen jeweils einige neu aufgenommene Stufen- keile mitentwickelt. Ein Nachteil dabei war, daB je- weils nur wenige Stufenkeile neben den Streuaufnah- men in einem Plattenkorb Platz hatten und die GroBe des Entwicklungstanks im Hinblick auf einen nicht zu groBen Entwicklerverbrauch nur als aus- reichend fiir einen Plattenkorb gewiihlt war. Es konnte daher aus den wenigen, meist etwas vonein- ander abweichenden Schwarzungen der Stufenkeile die Gradationskurve nur entsprechend weniger genau ermittelt werden. Aufnahme der Streuverteilung. — Die Streuintensi- taten in dem zu untersuchenden Winkelbereichgehen uber etwa vier Zehnerpotenzen. Da auf einer Platte nur ein Intensitatsumfang von wenig mehr als einer Zehnerpotenz mit geniigend groBer Genauigkeit aus- gemessen werden konnte, wurde der gesamte Win- kelbereich in vier Tntervalle aufgeteilt, die mit verschiedener ProjektivvergroBerung aufgenommen wurden. Die Belichtungszeiten wurden so gewahlt, daB der giinstigste Schwiirzungsbereich {S =0,1 bis 1 ) auf der Platte in dem betreffenden Winkelintervall lag. Meist wurden bei derselben ProjektivvergroBe- rung zwei Aufnahmen mit verschiedenen Belich- tungszeiten gemacht, damit die Einzelkurven sich iiberlappten. Das gute Zusammenpassen der Einzel- kurven ist ein Kriterium einerseits fiir die Konstanz des Strahlstroms, andererseits fiir die Richtigkeit der verwendeten Gradationskurve. Das Auseinander- klatTen in immer der gleichen Richtung bei den Einzelkurven der ersten ausgewerteten Streuauf- nahme war der AnlaB zur Nachprufung der Grada- tionskurve und fiihrte zur Entdeckung der Grada- tionsvertlachung bei den lange aufbewahrten Platten. Die gleiche Streuaufnahme mit Hilfe der berichtigten Gradationskurve ausgewertet lieferte die sehr gut zusammenpassenden Kurven der Abb. 5. Die Be- lichtung erfolgte nach einer Stoppuhr durch Auf- und Zuklappen des den Leuchtschirm tragenden Deckels der Plattenkamera. Bei den angewandten Belichtungszeiten zwischen 2 und 30 Sekunden konnte die Belichtung mit geniigender Zeitgenauig- keit mit der Hand vorgenommen werden. Die genaue Kenntnis der Belichtungszeit und der Projektivver- groBerung war fiir die Umrechnung der gemessenen Plattenschwarzungen in die relativ zu den unver- groBerten Aufnahmen richtigen Stromdichten wich- tig. Die absoluten Streustromdichten sind bei einer bestimmten Folic proportional zum Primarstrahl- strom. Eine Kenntnis der Absolutwerte der Streu- intensitiit ist also nur dann von Wert, wenn auch der Primiirstrahlstrom gemessen werden kann. Bei dem verwendeten Elektronenmikroskop war dies aber nicht moglich, well bei eingeschaltetem Strahl die Streufolie nicht gegen eine leere Blende ausgewech- selt werden konnte. um den durch Streuung nicht geschwiichten Strahl aufnehmen zu konnen, und weil nach dem Aus- und Wiedereinschalten der Kathodenheizung der Strahlstrom im allgemeinen einen anderen Wert annimmt (die GroBenordnung des Strahlstroms war 10"'^ A). Die an die Streukur- ven geschriebenen Absolutwerte der Stromdichte im Verhiiltnis zum Primarstrahlstrom wurden der Theo- rie von Lenz [7] entnommen, was insofern berechtigt erschien. als die Form der Streukurven sehr gut mit der nach der genannten Theorie berechneten Kur- venform iibereinstimmte. Jeweils unmittelbar vor der betreffenden Einzelauf- nahme mit einem bestimmten Projektivpolschuh wurde die ProjektivvergroBerung bestimmt. Streuung von 70-kV-Elcktronen an Kohlenstoff 11 Wr \ -la - t s 2 10 5 2 S 2 n s • • < • • « « \ » a 0 • • •(1 a • e 2 nl 5 2 1 e p-OOJ *» « V 1 -i." ro' 2 5 W' 2 5 W' 2 5 '0 ' /O' ^ 5 W' 2 to s 2 , i6 \ 1 c • ■» • « • • /D 1 5 • 2 It tS 5 2 iS 5 2 m « \ P'0.2 e e > j^ tO'r Abb. 5. 2 4 2 4 ra Abb. 6. I0-' 2 5 to-' 10-' 2 P'QSS JO-' 2 Abb. 7. 5 to- to- Abb. 5-7: Kurven der Streuverteilung von Kohlenstoff verschiedener Massendickc. Die durch unterschiedliche Zeichen ( 0) dargestellten MeBpunkte vvurden von verschiedencn Flatten der betieffenden Aufnahmeseric gewonnen. A',, ist die Anzahl der die Folic trefienden Elektronen, dN die Anzahl der in das Raumwinkelelement dil gestrcuten Elcktronen. Diskiission der Experimentalergebnisse. — Die Ab- bildungen 5 bis 7 zeigen die Kohlenstoff-Streuver- teilung bei verschiedenen Foliendicken. Etwa bei 1) = 10 - tritt der erste Beugungsring auf. Das ent- spricht einem Atomabstand von 4 bis 5 A. Infolge des logarithmischen AbszissenmaBstabes liegen die folgenden Beugungsringe dicht hintcr dem ersten Lind lassen die Streuverteilung im Bereich Si ■ 10"- nicht mehr gut erkennen. Dagegen sind die Streu- kurven im Winkelbereich zwischen 10^ und 10 - ungestort und die MeBpunkte liegen hier recht genau auf einer Geraden. deren Steigung von der Massen- dicke abhiingt. Die Massendicke yl wird zweck- miiBig in Einheiten der Aufhellungsdicke -//' gemes- sen; fur Kohlenstoff betragt deren Wert nach Lenz [7] etwa y/' = 2,51 10 10 ViX+eUjlm^c^) g (\ +e UjmQ c')'~ V cm- fur U 70 kV also etwa yl' = 1,46 10-= gcm-2. (2) (3) In der Theorie [5] wird fur p -'Z'2 (p = yl/yl' = Verhiiltnis von Massendicke zur Aufhellungsdicke) der Ausdruck fiir die Streuelektronendichte in zwei Summanden aufgespaltcn, von denen ciner zu 'd ^ proportional ist und fiir If 10 - groB gegen den anderen Summanden ist, so daB er nur allein beriicksichtigt zu werdcn braucht. Wegen der reinen Potenzabhiingigkeit von /> ist die theoretischc Streukurve im doppelt logarithmischen Koordina- tensystem ebenfalls eine Gerade. Ihre Steigung be- triigt 4p Z 2, ist also eine lineare Funktion von/?. In Abb. 8 ist diese Funktion fiir Z 6 dargcstellt und dazu die Werte, die sich aus den experimentellcn Kurven ergeben. Wie man sieht, liegen die MeBwerte bei den Folien mit glatter Oberfliiche in der Nahe der theoretischen Werte. Die Punkte, die zu den Folien mit weniger ebener Oberilache gehoren, liegen hoher. Es ist anzunehmen, daB die unebenc Ober- flache ihre Ursache darin hat, daB durch zu schnelle Verdunstung des Losungsmittels Blasen in der Folie enstanden sind, wodurch dann die Massendicke ge- ringer ist, als es sich aus der Dickenmcssung ergab. Die hoher liegenden Werte der Abb. 8 konncn hier- durch erkliirt werden. Bei Winkcln 'V 3 10 ' werden die Strctikiirvcn • c 2 t !5 0 O ^N* « 1.7 X te . o J5 K O V \ 1000 2000 3000 4000 5000/ — d 0. 1 Q? QJ 0,< P5 t — P Abb. 8: Abhangigkeit der Steigung der Strcukurven der Abbiidungen 5 bis 7 und zwei weiterer von der Massendicke /; bzvv. Foiiendicke J nach Theorie ( ) und Experimentcn mit den Folien mit ebener ( ) und unebener( ) Oberfiache. 72 G. KEMPF UND F. LENZ nach 0 min nach 2 mm nach 6 mm 10- 5 2 I 10- 5 10- 5 2 10- 5 10- 5 2 10 5 \ \, \ ±^ 5 10-' 2 \ ^ k \ \ ■J 1 — ^ 5 10-' 2 \ X \ \ ; _ ^ 0 T 2 10-' 2 \ \ J L nach 16 mm 10- 10-' 2 Abb. 9: Streukurven nach verschiedenen Bestrahlungszeiten einer 1400 A dicken Folie. Steigung der Kurven: a) 1,5, b) 1,71, c) 1,76, d) 1,8, e) 1,8, f) 1,8. flacher. Diese Tatsache folgt auch aus der Theorie fiir die Einfachstreuung (vergl. Abb. 6 aus [7]); aus der in [7] durchgefuhrten Theorie der Mehrfachstreu- ung geht sie nicht hervor. Das liegt daran, daB dort fiir den unelastischen differentiellen Streuquerschnitt auf eine Beriicksichtigung von dessen endlichen Grenzwert fur fZ-^O zur Vereinfachung der Rech- nung verzichtet wurde, was sich bei der Streuver- teilung von 70-kV-Elektronen fur Streuwinkel & < \0'* bemerkbar macht. Bei Abbildung 6 kann man bei *^ < 1,5 >■ 10"* den iJbergang der Streustromdichte in den ungestreuten Anted des Primiirstrahls beobachten. Eine mogliche Fehlerquelle bei der Bestimmung der Massendicke ist das Aufwachsen einer Koh- lenstoffschicht bei der Bestrahlung im stets etwas kohlenwasserstoffhaltigen Vakuuni des Elektronen- mikroskops. Abbildung 9 zeigt, daB nach 1 6 Minuten langer Bestrahlung die massendickenabhangige Stel- es OS — » t {minj Abb. 10: Anderung der Steigung der Kurven der Abbildung 9 in Abhangigkeit von der Bestrahlungszeit. Da die Belich- tungszeit jeweils 30 Sekunden betrug, liegen die MeBpunkte 15 Sekunden hinter der auf Abbildung 12 in vollen Minuten angegebenen Zeit, die den Beginn der Belichtung angibt. gung der Kurven im Winkelbereich zwischen 10"* und 10~^ keine merkliche Anderung erfahrt, daB bei den vorliegenden Versuchsbedingungen eine merk- liche Zunahme der Massendicke durch Aufwachsen einer Kohlenstoffschicht also nur sehr langsam er- folgt. Die Abnahme der Massendicke zu Beginn der Bestrahlung ist durch das Abdampfen der Nicht- Kohlenstoff-Atomgruppen verursacht (vergl. Abb. 10), wie eben bereits erwahnt wurde. Fiir die Themenstellung, sein standiges Interesse und zahlieiche Diskussionen sind wir unserem kurz nach Fertigstellung der vorliegenden Arbeit verstorbenen Lehrer, Herrn Prof. Dr.-Ing. Bodo von Borries zu tiefem Dank verpflichtet. Bei der Durchfiihrung der Experi- mente hat uns ferner Herr Dipl.-Phys. W. Scheffels manchen wertvoUen Hinweis gegeben. LiTERATUR 1. BiBERMANN, L. M., Izvest. Akad. Nauk SSSR, Ser. Fiz., 15,429 (1951). 2. BiBERMANN, L. M., Wtorow, E. N., Kowner, I. A., SsuscHKiN, N. G., und Jaworsku, B. M., Coinpt. rend. acad. sci. URSS 69, 519 (1949). 3. BOERSCH, H.,Z. Natiirforsch. 2a, 615 (1947). 4. VON BoRRiES, B.,Z. Naturforsch. 4a, 51 (1943). 5. Brockes, a., Vortrag gehalten auf der 6. Tagung der Deutschen Gesellschaft fiir Elektronenmikroskopie in Miinster 1955. 6. KoNiG, H.,Z. Physik 129, 483 (1951). 7. Lenz, F.,Z. Naturforsch. 9a, 185 (1954). 8. ToLANSKY, S., Multiple-Beam Interferometry of Surfaces and Films. Oxford, 1948. 9. Weber, K. und von Fragstein, C, Optik 11,511 (1954). Experimentelle Untersuchungen zum Kontrast cUinner Schichten im Elektronenmikroskop W. LlPPERT Max-Planck-Institut fiir Biophysik, Frankfurt a. M. Die elektronenmikroskopische Durchlassigkcil diin- ner Schichten ist sowohl von dor theoretischen als auch von der praktischen Seite von gewissem In- teresse. Neuere theoretische Untersuchungen von Lenz (3) haben iiltere Ergebnisse (1,2) modifiziert, und es erscheint zweckmiiBig einmal zu priifen, inwie- wcit diese Theorien die Verhiihnisse bei praktisch vorkommenden Schichten wiedergeben. Zum an- deren besteht immer wieder einmal der Wunsch, aus der eiektronenmikroskopischen Durchliissigkeit Riickschliasse auf die Schicht- bzw. Massendicke zu Ziehen. Das Ziel der Untersuchungen war also festzustel- len, ob bei einigermaBen detinierten Schichten eine eindeutige Beziehung zwischen Schichtdicke und elektronenmikroskopischer Durchliissigkeit, die na- tiirlich noch von der Objektivapertur und der Strahlspannung abhiingt, besteht. Ferner sollte ge- priift werden, ob die MeBresultate den theoretischen Berechnungen folgen. Wenn auch die Zielsetzung der Untersuchungen sehr klar war, darf doch nicht verschwiegen werden, daB sowohl bei der experi- mentellen Durchfiihrung, als auch bei dem Vergleich der Messungen mit der Theorie eine Reihe von prin- zipiellen Schwierigkeiten zu diskutieren war, auf die hier aber leider nicht niiher eingegangen werden kann. Es ist aber z. B. offensichtlich, daB man keine strenge Ubereinstimmung erwarten darf, da die hier in Frage kommenden theoretischen Ansatze die Struktur der Materie nicht beriicksichtigen, sondern sich strenggenommen nur auf einzelne Atome be- ziehen. Die MeBmethode war kurz folgende: An einem mit Zwischenhnse versehenen Siemens-Mikroskop wurde die photographische Platte durch einen Auffanger fiir Elektronen ersetzt. Die Objektivaper- tur wurde — vor allem zum Vergleich mit der Theorie — ungefiihr in dem Bereich zwischen I 10- und 9 X 10^- veriindert. Die benutzten Spannungen wa- ren 50 und 1 10 kV. Die Zwischenhnse gestattete die einwandfreie Justierung des Eiektronenstrahls und die durch die Kontrastblende eingestellte Objektiv- apertur zu kontrollieren. Bezijglich der untersuchten Schichten bestand die Absicht, amorphc und kristalline und auch leicht- und schweratomige Schichten zu untersuchen. Ge- nauere Messungen wurden an Kohle-, Aluminium-, Palladium- und Wolframoxydschichten durchge- fiihrt. Die Kohleschichten wurden dabci nach dem Konigschen Verfahren in einer Giimmentladung in Benzoldampf hergestellt. Die ubrigen Schichten wa- ren Aufdampfschichten. Der Auffanger fijr die Elek- tronen war so dimensionicrt, daB bei kristaliinen Schichten iiber eine groBe Zahl von Kristallitcn gc- mitteit wurde. Bei dem Vergleich mil der Theorie haben wir uns bei den relativ groBen Aperturen darauf beschriinkt, die elastische Streuung zu beriicksichtigen. Benutzt man die Wcntzclsche Streuformel (1927), so liiBt sich die eiektronenoptische Durchliissigkeit in allgc- meiner Form als Funktion einer zuniichst noch un- bestimmten Dicke — wir haben dazu die Aufhcl- lungsdicke gcwiihlt — und eincs ebenfalls zuniichst noch unbestimmten Winkels angeben. Mathematisch ergibt sich folgender Ausdruck: D{y,p) = exp 71 p\ 1 + ~yHY>(iy) 'D, hy 7 d{y do) Dabei ist }' die durch die Kontrastblende bedingte Ob- jektivapertur p die relative Massendicke der Schicht. ausge- drtickt in Vielfachen der von v. Borries (1951) eingefiihrten Aufhellungsdickc do die Winkclkonstante der Wentzelschen Streuformel /i, H'l^ die Besselsche bzw. Hankelsche Z\linder- funktion. Die Ausvvertung der Formel zeigt Bild 1: Die Kurven zeigen den Verlauf der eiektronenmikro- skopischen Durchliissigkeit als Funktion der Aper- tur fiir konstantc Dicke. Um diese Kurven mit den Experimenten in Beziehung setzen zu konnen. ist es notwendig, numcrische Werie fiir Aufhellungsdicke und Winkclkonstante anzunchmen. Um einen Uber- Abb. I. 1 Ickironcnmikroskopischc Durciiiassigkeit als Funk- tion tier ApcrUir tiir konstantc Dicke. 74 W. LIPPERT Tabelle I. Aiifhellungsdickcn in tig cm /^.„2 C Al Pd W 50 kV 9,1 7,3 5,4 4,9 (v. Borries) 21,4 17,4 12,6 11,5 (Leisegang) 11 10.5 10 10 (Lenz) 110 kV 17,2 13,8 10,2 9,2 (v. Borries) 40,8 32,6 24,1 21,9 (Leisegang) 20 19,5 19 19 (Lenz) blick iiber die in der Literatur benutzten Werte zu geben, haben wir die fUr die durchgemessenen Sub- stanzen in Frage kommenden Aufhellungsdicken nach drei verschiedenen Arbeiten, der Arbeit von V. Borries (1), der von Leisegang (2) und der von Lenz (3) berechnet. Wiihrend v. Borries und Leisegang fiir die Auf- hellungsdicken Formeln benutzen, die aus der Moliereschen Theorie abgeleitet sind und von der Ordnungszahl abhangen, berechnet Lenz mit Hilfe von Hartreefeldern die Aufhellungsdicken nur fur einige Substanzen. Die in der Tabelle nach Lenz angegebenen Werte fiir Al, Pd und W sind inter- poliert. Eine solche Interpolation ist nicht ohne weiteres zu rechtfertigen. Wir haben es trotzdem ge- tan, urn die Lenzschen Rechnungen mit fur die Diskussion benutzen zu konnen. Fur die Berechnung von <)„ wird meistens eine Formel der Art Zi In- a benutzt. X ist dabei die Wellenlange, Z die Ordnungs- zahl. a bedeutet bei einigen Autoren die Thomas- Fermi-Lange fiir WasserstoflF, bei anderen den Bohrsche Wasserstoffradius. Da der Unterschied bei- der Werte nur 11,4 ','„ betriigt, scheint zum Zvvecke des Verglcichs mit den vorliegenden Messungen eine niihere Diskussion nicht notwendig. Wir be- nutzen als Vergleichswert fur f\, denjenigen Wert, der sich durch Benutzung des Bohrschen Radius ergibt. Resultate: Es sollen zunachst die MeBresultate fiir Kohle, und zwar fiir die beiden Spannungen 50 und 110 kV diskutiert werden: Auf der Abszisse der Abb. la und lb ist dabei die Apertur, auf der Tabelle 2. Winkelkonstanten fiir die unicrsuchtcn Substanzen. Al Pd W 50 kV 110 kV 0,029 0,019 0,038 0,025 0,057 0,038 0,067 0,044 Ordinate die „Durchlassigkeit" angegeben. Die aus der Theorie erhaltenen Kurven sind ausgezogen, die mit Kreuzen gekennzeichneten MeBpunkte durch gestrichelte Geraden verbunden. Die an die Kurven geschriebenen Zahlen geben die Massendicke in //g cm- an. Als theoretischc Werte haben wir hier die Zahlen nach Lenz benutzt. Man sieht, daB die MeBpunkte einigermaBen die von der Theorie ver- langte Abhiingigkeit von der Apertur wiedergeben. Abweichungen lassen sich meistens durch Ringe im Beugungsdiagramm erkliiren. Keine tJbereinstim- mung besteht aber zwischen den gemessenen und den nach Lenz errechneten Massendicken. Die expe- rimentellen Massendicken sind bei 50 kV etwa 1,7- bis l,8mal, bei 110 kV etwa 1,8- bis l,9mal so groB wie die errechneten. Ahnlich sind die Resuhate bei Aluminium (Abb. 2c): Die Aperturabhiingigkeit wird, eventuell unter Berucksichtigung der UnregelmiiBigkeiten im Beu- gungsspektrum, einigermaBen befriedigend wieder- gegeben, die experimentelle Massendicke ist aber etwa um den Faktor 1,4 bis 1,5 groBer als die aus der Theorie interpolierten Werte. Mg-F.,, von dem wir auch einige Proben vermessen haben, verlangt etwa das l,6fache der errechneten Massendicke. Bei den schweratomigeren Substanzen sieht es bei Pd (Abb. 2d) zunachst so aus, als ob eine relativ befriedigende Ubereinstimmung zwischen Experi- ment und Rechnung sowohl bezuglich der Winkel- abhiingigkeit als auch bezuglich der Massendicke vorliegt. Doch ist auch hier schon das angedeutet, was bei Wolframoxyd (Abb. 2e) deutlich sichtbar wird: Der Verlauf der errechneten Kurven wiirde sich besser dem Experiment anpassen, wenn der Ver- lauf steiler, d. h. die Winkelkonstante kleiner ware. Die Vergleichskurven fiir Wolframoxyd beziehen sich auf reines Wolfram. Der Gewichtsanteil des SauerstofTs betriigt bei dem in die Verdampfungs- apparatur gebrachten WO.j nur etwa 20 "„. Da bei der Verdampfung eine Reduktion stattfindet, diirfte bei der eigentlichen Schicht der Sauerstoffgewichts- anteil noch kleiner sein. Auch wenn man z. B. ver- sucht, die Wolframoxydschicht als aufeinanderlie- gende Schicht von W und O aufzufassen, bleibt diese charakteristische Abweichung im Prinzip beste- hen. Auch bei Wismut- und Platinschichten zeigte sich ein ahnlicher Effekt. Die Winkelkonstante sollte bei beiden Substanzen etwa um 40 "o kleiner sein. Die Messungen haben als wichtiges Resultat ge- zeigt, daB bei den bisher genannten Substanzen eine eindeutige Beziehung zwischen elektronenmikro- skopischer Durchliissigkeit und Schichtdicke, die i. a. recht gut durch eine t'-Funktion wiedergegeben wird. besteht. Abb. 3 zeigt als Beispiel hierfiir die Abhiin- gigkeit der Durchlassigkeit von der Schichtdicke fiir Kohle bei 50 kV und einer Apertur von 0,03. Aller- dings gibt es auch Schichten, fiir die diese einfachen und reproduzierbaren Verhiiltnisse nicht zutrefFen. Dies ist der Fall bei Substanzen, bei denen die Kontrast diinner Schichtcn im Elcktionenmikroskop 75 100 Kohle. 50 KV ___— — ^^^ — —' ' ^ — D ^^^^^^^ ^<^^^^'^ ^^-— 5.5 ^<^^ ^ J 50 — \^ 1 1 1- 1 1 1 — 1 100 001 0.02 003 0.04 0,05 0.06 0.07 0 08 ;• Kohle, 110 KV ^^-^-HT" _^^r-^ . ^^^^5^^^ ^^,^=.^:S— - ^^^ /y>:^'' ^^55?-=*'''''^^ ^___„,„-— ^^"-. ^^..--''"''''^ / .^^■^^'^ ^^„--- "^ py ^^-^ / / 20^' 8^ y ■^ ^^ 422 y ^^ ' y^ y^ y^ y^ yT^ ^ ^ 40,^/ y t=- 1 1— 1 \ H 1 1 1 0.01 0.02 0.03 004 005 006 0.07 0.08 la lb uo Al. 50 KV - u ^-—-""'^ — " " _J^ — "^^ — "- " " ^ --^ ** ^'""'^'^ — ' "^ -^ 3?^ ^^^-^^^ " .j!>^C^' ^ ^^^^J^^^^"^^^ ^ ' sn- je" ^.-'"■^'''^ -^ ** --^ '^(C — ^.^-^"^^^ "^ tr-^"""''^ -^ *''^ -^ '^ ^-^"""'''^ ^- "^ lii^^ltv'-/ C^ -^^^ - " ^^^^^ "^ vi.T^'^-y^ *" ^^^-"^^^ ^^^"""^^^ ^ ^-^"^^^^ _y__-— — 3^ -' ^^--""^ "^ ^---^'^ i2 __ — — -^ 1 p 1 1 1 1 1 1 ' 0,01 0,02 0 03 0.04 0.05 0.06 0 07 0 08 100 Pd 50 KV ^^__^ u 5 —- -— "^ ^ -.y^^^^^^^ — >fr^ — j<^ ^-— — "^ — ^ -— • "^ 10 ^ -"^^ ^ €^'\^^^^ 20 ^ ^ ^ -— ^ — - — »- 22 •*'^ -- " ^ _- — " 40 — 1 1 — — 1 \ 1 1 1 1 0.01 0 02 0 03 0 04 0.05 0 06 0 07 0 08 ,• Ic Id Abb. 2 a-e. Die Durchlassigkeit als Funktion der Apertur bei verschiedenen Massendicken in //g/cm- angegeben. Aus- gezogene Kurven aus der Theorie eriialten; gestrichelte Kiir- ven durch die MeBpunkte gezogen. Massendicke im Sublichtmikroskopischen sehr un- gleichmaBig ist wie z. B. bei Gold und Silber. Das Elektronenmikroskop liiBt sich also bei Sub- stanzen mit einigermaBen gleichmaBiger Schicht- dicke nach Eichung zur Dickenmessung benutzen. Der MeBbereich laBt sich dabei durch Verandcriing der Objektivapertur in gewissen Grenzen verandern. Leider laBt sich aber die Abhiingigkeit der Durch- I'l <^<^' lUU D Wolfram oxyd, 50 KV 50' 10 20 igiSSS^: 40 -T — '"'^ i"^ 1 1 1 1 1 ' 0 03 005 0 07 le Abb. 3. Die Abhangigkeil der Durciiiassigkeit von der Schichtdicke fur Kohle bei 50 kV, Apertur 0,03. liissigkeit von der Schichtdicke nicht ohne weiteres aus dem hier benutzten cint'achcn Ansatz und den in der Literatur bekannt gewordenen Berechnungen entnehmen. Die Abweichungen hegen bei den leicht- atomigen Elementen in der Richtung. daB die Lcnz- schen Werte I'iir die Aufhellungsdicke zu kiein sind, man muB vielmehr fiir die Auliielkingsdicke Zahicn einsetzen, die in der Niihe der Leisegangschcn Werte liegen. Bei den schweratomigen Elementen lassen sich die Abv\eichungen in erster Linic so deuten, daB man fiir die Winkelkonstante gegeniiber dem theoretischen Wert einen kleineren Wert cinzusetzen hat, um /u einer besseren Angleichung der Mes- sungen an die Theorie zu kommen. Literatur 1. VON BoRRiES, B.,Z. Naturforsch. 6a, 51 (1951). 2. Leisegang,S.,Z. Physik 132, 183 (1952). 3. Lenz, F.,Z. Naturforsch. 9a, 185 (1954). 4. Wentzel, G., Z. Physik 40, 590 (1927). Zur Veranderung des Streuvermogens eines Festkorpers gegentiber mittelschnellen Elektronen infolge lonisation und Anregung W. SCHEFFELS Rheinisch- Westfdlisches Institiit fiir Vbennikroskopie, Diissekhif Die friiher beschriebenen Versuche mit diinnen Fo- lien im Schattenbildstrahlengang (7, 8, 9) wurden fortgefiihrt in dem Bestreben, quantitative Ergeb- nisse zu erhalten iiber die Anderung des Streuver- mogens durch die Bestrahlungsstromdichte. Zu- nachst soil kurz der Strahlengang in dem benutzten permanentmagnetischen Elektronenmikroskop nach von Borries (2) geschildert werden (Abb. 1). Der etwa 4 10"^ cm im Durchmesser groBe engste Strahlqucrschnitt des Strahlerzeugers (a) wird mit der Projektivlinse (h) und weiter mit der Objektiv- linse (c) auf beispielsweise 6 10 '^ cm verkleinert. So ensteht hinter dem Objektiv eine scharfe Kaustik, die lediglich iiber den Durchmesser des verkleinerten Bildes der Strahlquelle verschmiert ist. Befindet sich eine diinne Folic zwischen Objektivlinse und hinterer Brennebene, so schneidet sic die Kaustik, und in- folge der erhohten Stromdichte in den Schnittlinien (Kaustikstern und Kaustikmantel) lassen sich diese mittcls KohlenstofTauftragung auf die Folic auf- zeichnen. Im Schattenbild kann man diese aufwachsenden Kohlelinien gleichzeitig beobachten (Abb. 2a). Die achsennahen Strahlen projizieren den Kaustikstern in die Mitte des Bildes, aber die Strahlen, die in der Folienebene zum Aufbau des Kohlesterns beitragen, projizieren ihn ebenfalls weiter auBen im Bild. Der Kaustikmantel wird scharf nach ganz auBen proji- ziert. Eine Theorie dieser Schattenprojektion von F. Lenz (4) ermoglicht eine quantitative Deutung dieser Aufnahmen. Wartet man mit der Aufnahme des Schattenbildes aber nicht bis sich die Kohle aufgetragen hat, oder bewegt man gar die Folic wahrend der Aufnahme mit der Objektverstellung in der Folienebene, so erhiilt man ebenfalls eine Projektion von Linien, wie Leuchischirm in Abb. 2h auf einer Kollodiumfolie. Die mittelste Figur wurde friiher als ,,Streustern" beschrieben. Im Gegensatz zu der Kohleauftragung ist die Anderung des Kontrastes der Kollodiumfolie vollig reversibel und hat seine Ursache nur in der erhohten Strom- dichte in der Kaustik und nicht in einer Anderung der Massendicke der Folic. Bemerkenswert sind die hellen Siiume an der Seite der dunklen Linien. Sic lassen sich als Beugungssaume erklaren, wenn man annimmt, dass in der Objektfolie an dieser Stelle eine sprunghafte Anderung der elektronenoptischen Brechungsindex oder des Absorptionskoeffizienten stattfindet. Wie man auf der gerechneten Darstellung (Abb. 3) der Stromdichte im Kaustikquerschnitt auf einer Geraden durch die Taille des Sterns sieht, fiillt die Stromdichte sowohl im Stern wie im Mantel nach auBen viel schrofTer ab als nach innen. Bei einer idealen Kaustik d. h. bei Verwendung einer punkt- formigen Elektronenquelle, wiirde die Stromdichte unendlich werden, aber da diese iiber das verkleinerte Bild der Strahlquelle verschmiert ist, erhalt man durch Mitteilung endliche Maxima in Stern und Mantel (beispielsweise 5 10"- A cm- in der Sterntaille). Man kann annehmen, daB infolge der stark unter- schiedlichen Stromdichte auf der Fclieauchdielonisa- tion und Anregung der Objektatome und damit das mittlere innere Potential in der Folic ortlich stark ver- schieden ist. Rechnet man fiir eine Abschiitzung bei einem scharfen Sprung zwischen zwei konstanten Werten des inneren Potentials die Intensitaten der in der Schattenprojektion enstehenden Fresnelschen Beugungslinien aus und mittelt diese Intensitatsvcr- teilung iiber den Durchmesser des Bildes der Strahl- quelle, so erhalt man — als Phasenkontrast — eine Intensitatsverteilung, wie sie bei den in Abb. lb gezeigten dunklen Linien mit hellem Saum vorliegt. Vergleicht man diese gerechnete Verteilung der a "1 Abb. 1. Registrierung der Kaustikquerschnitte. Abb. la. Schattenbild eines Kaustii:5.J ro'cm d) nur axenparallele Strohlen Abb. 5. Stromdichten auf dem Kaustikquerschnitt in Ab- hangigkeit von dessen axialer Lage — Z = z/C^. y\ = Strom- dichte im Kaustikquerschnitt, /,,- Stromdichte in der Mitte der Linse. inneren Potentials, wie sie hier durch Tonisation und Anregung der Objektatome hervorgerufen wird, ist also verbunden mit einer Anderung der Streuquer- schnitte. Der Zusammenhang zwischen der Anderung des inneren Potentials und der Stromdichte auf der Folic liiBt sich qualitativ in folgcnder Weise bc- schrciben (5): Die Bcrechnung des inneren Potentials aus 0 nach H. Bethe (1) gilt nur. wenn der Fest- korper aus einander gleichen, kugelsymmctrischcn Atomen besteht, was hier nicht der Fall ist, da in der Folic infolge unelastischer Strcuung ein Teil der Atome ionisiert oder angercgt ist. Wenn der Fest- korper aber im Ganzen neutral ist, so kann man ihn auffassen als cine lineare Uberlagcrung des mittlcren Potentials der A' im Grundzustand bcfindlichen Atome mit der durch die AA^ angeregten bzw. ionisierten Atome verursachten Storung. Dann er- gibt sich, daB die Anderung des mittlcren inneren Potentials nicht zu A A^ Wsondcrn zu (A A^W)i pro- portional ist. Wenn also beispielsweise jedes tau- sendste Atom durch unelastischen StoB ionisiert ist, steigt das inncre Potential nicht um ein Promille, sondern um 10 "^'o. Fine solchc Zunahme kann aber schon zu erkennbarem Kontrast fuhren. Diese dritte Wurzel bewirkt aber auch, daB cine sehr starke Erhohung der Stromdichte im Objekt iiber den normalen Wert hinaus das inncre Potential und damit den Kontrast nur um einen geringen Betrag iiber den Wert erhohen kann, wie er bei der normalen Stromdichte bereits gegeben ist. Herrn Prof. Dr. B. v. Borries sowie Herrn Dr. F. Lenz danke ich fur wertvolle Anregungen und Diskussionen zur vorliegenden Untersuchung. LiTERATUR 1. Bethe, H., Ann. Physik, Lpz., 87, 55 (1928). 2. VON Borries, B.,Z. wiss. Mikroskop. 60, 329 (1952). 3. Lenz, F., Z. Naturforsch.9a, 185 (1954). 4. — eingereicht zur ,, First Regional Conference on Electron Microscopy in Asia and Oceania, Tokyo", Oktober 1956. 5. Lenz, F. und Scheffels, W., Z. Nattirforsch. 11a, 656 (1956). 6. Le Poole, J. B., Diskussionsbemerkung zum Vortrag von W. Scheffels, B. v. Borries und F. Lenz, Proceedings of the Conference on Electron Microscopy. London, 1954. 7. Scheffels, W., von Borries, B., und Lenz, F., Rapport Europees Congres Toegepaste Electronenmicroscopie Gent 1954, 293. 8. — Natuiwissenschaften 4\,%'i (\95A). 9. — Proceedings of the Conference on Electron Micros- copy. London, 1954. Contraste de phase et contraste interchromatique Etude comparee des methodes M. LoCQUIN Museum., Pans Le fonctionnement du dispositif comportant un cone creux d'electron tangent au bord d"un dia- phragme a bords amincis se revele different suivant : a) la plus ou moins grande tangence du cone sur les bords du diaphragme; b) Tepaisseur de Tobjet. Dans le cas des objets minces et lorsque le faisceau n'empiete pas sur les bords du diaphragme on a une image dont la structure ressemble exactcment a rimage en contraste de phase d'un microscope photonique, c"est a dire que les contrastes dans certaines limites sont approximativement propor- tionnels a Tepaisseur de Tobjet multipliee par son indice, c"est a dire dans le cas des electrons par les poids atomiques des structures traversees. Dans le cas des objets epais pour un reglage du cone d'eclairage traversant les bords amincis du diaphragme la structure de Timage est totalement differente. Avant de proceder a I'analyse detaillee de cette structure etant donne que les images possedaient un intervalle de contraste sur le negatif trop grand pour etre aisement copiable sur un positif a Taide de photons sans perte des details nous avons mis au point avec MM. Mollinat et Weber un procede de transposition en couleurs qui triple rintervalle des densites copiables et dont voici le principe : Du negatif unique on tire par voie photographique trois positifs correspondant a trois branches de den- sites croissantes du negatif. La premiere tranche est Einfliifi von Temperatiir, Unterlagc ami Bedeck ung auf Praparate 79 obtenue par copie sur une emulsion a grand contraste a Taide d'un temps de pose juste suffisant pour cn- registrer les faibles lumieres du negatif. Les grandes densites sent obtenues dc la meme fagon ma is en travaillant sur un positif intcrmcdiaire. La tranche moyenne est obtenue apres confection d'un masque des grandes lumieres du negatif que Ton superpose a celui-ci et tirage sur emulsion dure de I'image complexe ainsi obtenue. A partir de ces trois cliches on tire trois matrices en gelatine qui sont impregnees des trois couleurs fondamentales et dechargees successivement apres reperage sur papier. On obtient ainsi une transposition en couleurs des contrastes de I'image. Suivant I'ordred'impregna- tion des trois matrices on peut obtenir six images physiquement equivalentes mais donnant physiolo- giquement a I'oeil de I'observateur une impression fort differente. Seules deux images sont aisement interpretables, I'une dans le cas des fortes densites de I'objet, I'autre dans le cas des faibles densites. L'examen de ces images montre immediatement que les contrastes ne sont pas comme dans le cas du contraste de phase proportionnels aux epaisseurs optiques de I'objet. Les contrastes semblent parfaitement indepsndants de I'epaisseur de I'objet et lies uniquement aux poids atomiques des structures traversees. Pour en etre certains nous avons effectue des microincinerations par bombardement electronique dans le corps du microscope lui-meme des fibres observees et pris des images successives jusqu'au squelette mineral final. Ainsi on a pu mettre en evidence que les parties de fibre dans lesquelles se trouvaient concentres les atonies de calcium etaicnt ceiies qui presentaient le plus grand contraste et que ce contraste etait prati- quement independant dc I'epaisseur dc la fibre ou de ses details. Nous avons baptise ce phenomene nouveau du nom de contraste interchromatique rappelant ainsi que le contraste nait cntre des faisceaux d'electrons ayant subi d'abord au niveau de I'objet puis au niveau du diaphragme des pertes chromatiqucs ap- proximativement equivalentes; le diaphragme objec- tif etant suppose homogenc et I'objet hetcrogene les pertes chromatiqucs subies a la traversce de I'objet different statistiquement suivant les points de celui-ci des pertes chromatiqucs moyennes en quelque sorte compensatrices subies a la traversee du bord mince du diaphragme. La compensation parfaite ne pourra se faire que pour les quelques points de I'objet dont les poids atomiques seront en principe egaux aux poids atomi- ques des constituants du diaphragme. L'augmentation et la diflferenciation des contrastes aura done pour origine la plus au moins grande compensation des pertes chromatiqucs produites au passage a travers le diaphragme. II en resulte un contraste proportionnel aux poids atomiques des structures traversees et pratiquement independant de I'epaisseur de I'objet. Ceci n'est vrai en toute rigueur que pour une ouverture du systeme optique electronique non finie. Comme I'ouverture des objectifs est petite il faut superposer a ce phenomene I'efTet de la diaphragma- tion et les contrastes deviennent alors lies aux poids molcculaires plus qu'aux poids atomiques de Tobjet. Der EinfluB von Temperatur, Unteiiage und Bedeckung auf die Veninderung elektronenmikroskopischer Priiparate K.-J. Hanszen Physikalisch- Technische Buiidesaiistall, Biuuiischweig Jede optische Abbildung ist mit einem EingriflT der abbildenden Strahlung in das untersuchte Objekt verbunden. Je kleiner die abzubildende Einzelheit um so groBer die Wirkung des EingrifTs auf sic. Dieses grundlegende Gesetz laBt erkennen, wie sehr das Problem der Objektschiiden im Elektronenmi- kroskop mit wachsendcr VergroBerung an Tragweite gewinnt, da die unverfiilschte Wiedergabe von Struk- tureinzelheiten, die auf Grund des apparativerreich- baren Auflosungsvermogens noch moglich sein sollte, durch diese wesentlich behindert wcrden kann. Die primiire Folge der Wechselwirkung zwischen den einfallenden Elektronen und dem Objekt beste- hen in Anregung, lonisation und anderen quanten- haften Veranderungen der getrofFenen Molckiile, die sekundiire Folge in einer pauschalen Erwiirmung des ganzen Objckts. Durch vergleichende Hetrachtung der im Elcktronenstrahl und der tei Temperaturbc- handlung hervorgerufenen Veranderungen in organi- schen Priiparaten konnten Konig und Mitarb. (5, 6) die thermischen Schiiden von denjenigen trennen, die auf spezifisch elektronische Wcciiscluirkungen zu- riick/ufuhren sind. Ahcr auch die rein temperaturbedingten Schiiden konnen im Elektronenmikroskop wesentlich andcrs verlaufen als die thermischen Umvvandlungs\or- giinge in iciiwn Substan/en (3, 4), da die Praparate im Elcktronenstrahl einerseits mit der Triigerfolie, andererseits mit den sicli auf ihnen niederschlagen- den .,Kohle"-Bedeckungen reagieren konnen. Aus 80 K.-J. HANSZEN B.«t-<4.~ia ^'St%^v ' Fig. 1. Locherbildungen in einer SiO-Tragerfolie durch Reaktion mit Kupferoxyd (obenj und mit Eisen (unten) bei hohcn Bestrahlungsintensitaten. diesem Grunde kommt z. B. der Temperaturbestim- mung im Elektronenmikroskop durch Beobachtung von Schmelzprozessen und anderen thermischen Um- wandlungen nur ein beschrankter Anwendungsbe- reich zu. Ein Beispiel fiir eine Reaktion mit der UnterJage stellt die in Fig. I a wiedergegebene Zerstorung einer SiO-Tragerfolie durch ein Kupfer(I)Oxydpraparat dar, die bei hoheren Bestrahlungsintensitaten ein- tritt. Wiihrend der Reduktion des Oxyds im Elek- tronenstrahl reagiert dieses unter Locherbildung mit der Unterlage. Da iihnliche Erscheinungen auch nach Tempern im Vakuumofen, hier allerdings erst iiber 900'C, d. h. nach vollstandiger Reduktion des Oxyds zu Kupfer beobachtet wurden, liegt es nahe, auch im Elektronenmikroskop einen hauptsachlich durch dicTemperaturerhohungbedingten Reaktions- ablauf anzunehmen, der jedoch durch quanten- bedingte Ursachen friiher in Gang kommen kann, als auf Grund der pauschalen Temperaturerhohung des Praparats zu erwarten ist. Ein weiteres Beispiel fiir die Locherbildung in einer SiO-Unterlage durch Reaktion mit dem Praparat zeigt die Fig. 1 b. Hier haben verdampfende Eisenkiigelchen die Tragerfolie auf ihrem Weg angefressen. Von weitaus groBerer Bedeutung ist der EinfluB der erwahnten „K.o\\W-Bedeckiingen auf die Prii- paratveranderungen. Ihre Mitwirkung bei den tem- peraturbedingten Umwandlungen wurde durch Va- kuumtempern von Objekten ermittelt, die im Elek- tronenstrahl oder durch Bedampfen mit einer Kohle- deckschicht versehen waren. Durch Vergleich der nach den Erhitzungen im Ofen beobachteten Ver- anderungen mit den im Elektronenmikroskop durch die Elektronenbestrahlung hervorgerufenen Schaden konnten Ruckschlusse auf den EinfluB der Bedek- Fig. 2. Zerstorung einer 100 A dicken Silberaufdampfschicht auf SiO-Unteriage durch Tempern im Vakuumofen. Linke Spalte: unbedeckte Schichten; oben: nach Tempern auf 200 C; unten: auf 500'C. Rechte Seite: Schichten mit einer im Elektronenmikroskop entstandenen Kohlebedeckung: oben: nach Tempern auf 200 C; unten: auf 500 C. kungen bei den Veranderungen im Elektronenmikro- skop gezogen werden. Fig. 2 zeigt in der linken Spalte 5/7/)t^raufdampf- schichten, die vor der elektronenmikroskopischen Aufnahme einer kurzzeitigen Erhitzung unterworfen waren. Auf Grund von Adsorptionsschichten auf der Unterlage und der Schicht selbst besitzen die Atome an den Oberflachen der Silberpartikel eine so groBe Beweglichkeit, daB sich die Teilchen bei den Erhitzungen auf die angegebenen Temperaturen be- reits vollstandig abgerundet haben. Ganz anders verhalten sich Praparate, die schon vor dem Tempern dem Elektronenstrahl ausgesetzt waren, also mit .,Kohle" iiberdeckt waren (Fig. 2, rechte Spalte). In diesen Fallen vermag die Kohle die morphologische Schichtstruktur noch bei 200'C praktisch vollstandig zu schiitzen. Bei hoheren Tem- peraturen tritt dagegen eine vermehrte Aggregation der Silberschicht zu groben Klumpen ein. Es sieht so aus, als ob jetzt durch die Kohlebedeckung eine weitere Lockerung der Atome an den Oberflachen der Silberpartikel verursacht wird, die sich in einer gesteigerten Beweglichkeit dieser Atome auswirkt. Fijr die Moglichkeit einer solchen Herabsetzung der Bindungsfestigkeit der Oberflachenatome gegenuber dem Kristallgitter spricht auch die Tatsache, daB selbst in den durch die Bedeckung geschutzten Einflufi von Temperatur, Unterlage und Bedeckunf^ auf Piciparate 81 Fig. 3. Zerstorungen einer 100 A dicken Silberaufdampf- schicht durch Bestrahlungen im Elcktronenmikroskop. Unter den gewiihlten Fokussierungsbedingungcn sublimiert die Schicht ohne Aggregation. Schichten an einzelnen Stellen eine Luckenbildung durch Suhliiiuition der Silberpartikel eingetreten ist (s. Fig. 2). Die Vorstellung, daB es sich hier um eine bevorzugte Oberflachenverdampfung des Silbers un- ter der Einwirkung der Umhiillung handelt, wird durch die Tatsache bestatigt, daB Liickenbildungen dieser Art bei einer Zerstorung der Schicht durch Bestrahlung im Elcktronenmikroskop in erhohtem MaBe auftreten; denn in diesem Fall ist die Wirkung der Bedeckung besonders stark, da sie sich wiihrend der ,,Verdampfung'" auf der frisch gebildeten Silber- oberfliiche immer wieder neu niederschlagen kann. So erkliirt es sich, daB Silberschichten im Elektronen- strahl unter geeigneten Umstiinden ohne vorherige Aggregation vollstiindig abdampfen konnen (Fig. 3). Zur Klarung der Ursache fiir die vermehrte Su- blimation der Aufdampfschichten durch die Bedek- kungen soil Fig. 4 herangezogen werden. Sie zeigt ei- nen im Verdampfungsstand mit 200 A Kohle iiber- decktenSilberkeil nach Tempernauf 185 C. In diesem Bild fiillt auf, daB die groBten Zerstorungen nicht im diinnsten Schichtbereich auftreten, sondern bei einer Schichtdicke, die mit der Starke der Kohlebcdeckung korrespondiert. Hieraus glauben vvir, den Vcrdamp- fungsmechanismus nach folgender Vorstellung er- klaren zu konnen: Wahrend die aufgedampfte Kohle die Schicht- strukturen am diinnen Ende des Silberkeils liickenlos Fig. 4. Zerstorungen in einer keilformigcn Silberschicht, die durch BedampfLuig mit einer 200 A dicken Kohleschicht bedeckt wurde, dLircli Tempern auf 185 C. Die Liickenbiidung iriit vornehmlich bei einer Silberdicke von 200 A auf. zudcckt imd konserviert, liegt bei den groBeren Schichtdicken nur eine unvollstiindige, schneehiiub- chenartige Bedeckung vor (Fig. 5). Vor allem an den Rissen in der Silberschicht tritt eine Unterbrechung der Bedeckung auf. Die Kohle wandert aber an die- sen Stellen auf Grund ihrer groBen Oberlliichcnbe- wegHchkeit weit in das geometrische Schattcngebiet hinein und uberzieht dort das Silber mit einer extrem diinnen Absorptionsschicht, die zu der besprochenen Lockerung der Oberflachenatome des Silbers AnIaB geben kann, so daB die beobachtete Sublimation mogiich ist. Bei starkerer Erhitzung losen sich schlieBlich, von diesen Stellen ausgehend, die Hiillen vollstandig auf und geben zu den gefundenen Aggre- gationen AnIaB. Eine Bestiitigung fiir diese Erklarung bietet die Tatsache, daB unter wechselndem Winkel bedampfte Silberschichten nach dem Tempern we- Bedeckung ^Tragerfolie Fig. 5. Querschniu durch eine kohiebedeckte Silberschicht. Bestrahlung Warmcableitung Fiu. 6 a. Warmezuleitung Fig. 6 b. Fig. 6. Schema der Warmeerzeugung und -abieitung. a) im Elektronenstrahl, h) im Vakuumofen. 6 — 568204 Electron Microscopy ^2 K.-J. HANSZEN Fig. 7. Zerstorungen in einer Wismutaufdampfschicht auf SiO-Unterlage durch Vakuumtempern auf 200'C. Oben: Unbedeckte Schicht mit Verschmutzung; unten: Schicht, mit einer im Elektronenmikroskop entstandenen Kohlebe- deckung. niger Lucken aufweisen als nur senkrecht bedampfte Silberschichten. Liickenbildungen der besprochen Art sind von Bryant und Mitarbeiter (1) erstmalig an Wisinut beo- bachtet worden. Sie erklarten diese Erscheinung als eine Verdampfung der Schicht an Stellen herabge- setzter Warmeleitfahigkeit durch mangelnden Kon- takt mit der Unterlage. DaB jedoch diese Erklarung den wesentlichen Kern des Vorgangs nicht triflFt, lehrt die Tatsache, daB die gleichen Schaden auch beim Tempern im Of en auftreten, wo ganz andere Warmeerzeugungs- und -ableitungsbedingungen vor- liegen und nach solchen Uberlegungen der beobach- tete Effekt nicht auftreten diirfte (Fig. 6). Der Dampfdruck des Wismuts Hegt um mehrere Zehnerpotenzen hoher als der von Silber. Aus diesem Umstand erklart es sich. daB bei diesem Metall auch Liickenbildungen durch Tempern unbedeckterSchich- ten, bevorzugt an verunreinigten Stellen, auftraten. DaB auch hier die Verdampfung durch die Kohle- hiille wesentlich gefdrdert werden kann, ist aus Fig. 7 zu entnehmen. Wiihrend beim Tempern von Silber auf niedrigere Temperaturen noch der schiitzende EinfluB der Um- hiillurg vorherrscht, tritt an A^w/T/t^/'Schichten schon unter den gleichen Bedingungen der gegenteilige Effekt ein: Bedeckte Kupferaufdampfschichten auf SiO-Trager weisen bereits bei niedrigeren Tempera- turen Lucken auf, wahrend unbedeckte Schichten noch nicht veriindert sind. Ebenso findet bei hoheren Temperaturen unter der Bedeckung eine vorzeitige Aggregation statt^ (Fig. 8). Bei der Interpretation dieser Befunde ist zu beden- Fig. 8. Zerstorungen in einer Kupferaufdampfschicht auf SiO-Unterlage durch Tempern im Vakuumofen. Oben: Auf 350X; unten: Auf 400-C. Linke Spake; Unbedeckte Schich- ten, im unteren BiJd beginnt SammelkristaUisation; rechte Spake: Schichten, mit einer im Elektronenmikroskop ent- standenen Kohlebedeckung. ken, daB das Kupfer mit seiner starken Affinitat zum Sauerstoff einen Reaktionsmechanismus auf- weist. der beim Silber unbekannt ist. In den Elek- tronenbeugungsdiagrammen der Kupferaufdampf- schichten traten stets neben den Cu-Refiexen dieje- nigen von Cu.O auf. Daraus isl zuschlieBen,daBdie Aufdampfschichten von einer dunnen Oxydulschicht iiberzogen sind, die eine ahnliche Schutzwirkung auf das darunter liegende Kupfermetall ausiibt, wie es oben von den Kohlebedeckungen auf Silber darge- legt wurde. An Hand der Beugungsdiagramme kann man belegen, daB die Zerstorungen der Kupfer- schicht erst dann beginnen, wenn durch die eintre- tende Temperaturerhohung die Oxydulschicht zer- stort ist. Nun findet aber, wie Erdmann-Jesnitzer und Gunther (2) darlegten, das Verschwinden dieses Oxyds nicht erst bei LJberschreiten der Dissozia- tionskurve statt. Vielmehr liegt der Dampfdruck des Oxyduls so hoch, daB bereits bei niedrigeren Tem- peraturen seine Sublimation eintritt. Nach dieser Verdampfung der Deckschicht hat aber der restliche 1 Auf Kohleunterlage ist die Beweglichkeit des Kupfers so groB, daB die Aggregation bereits auftrat, bevor eine Luckenbildung zu beobachten war. Ei?ifiiiP von Tempcratiir, Untcrloffc iiiul Bedeck iini; aitf Piciparate 83 Sauerstoff in der Vakuumapparatur Zutritt zu der reinen Kupferoberflache und kann mit dicser wiedcr zu CUjO reagieren, da wir uns noch oberhalb der Dissoziationsdruckkurve befindcn. Hiermit kann das Spiel von neuen beginncn, bis das ganze Kupi'er vcr- dampft ist. Die Oxydulschicht hat also beziiglich der Verdampfung eine Art katalytische Wirkung. Es darf nicht auBer Acht gelassen werden, daB bei den hier behandelten Mikrostrukturen wescntliche Verschiebungen der Gleichgewichtskurven auftreten konnen, bedingt durch die bekannten Effekte .,klei- ner Tropfchen" usw. Temperimgsversuche an zu CuO durchoxydierten Kupfcraufdampt'schichten, die untcr gleichen Bedingungen vvie die Kupl'ererhit- zungen angestellt wurden, bestiitigten dieses: Die Reduktion zu Oxydul tritt schon bei etwa 270°C, die zu Kupfer kurz unterhalb 400 C ein\ letztere also in dem Temperaturbereich. in dem sich die ersten Zerstorungen an unbedeckten Kupferschichten bc- merkbar machten. Im Gegensatz zu den Verhaltnissen an unbedeckten Schichten hat man es bei Vorhan- densein von Kohlebedeckungen nicht mit echten Gleichgewichtszustiinden, sondern mit reaktions- fiihigen Partnern im gehemmten Gleichgewicht zu tun. Hier kann also bereits vor der besprochenen Sublimation des Oxyduls seine Aufzehrung durch die Kohleeintreten. AusschlieBend vermag die Bedeckung auf Kupfer iihnliche Wirkungen wie auf Silber aus- zuiiben. Dieses steht im Einklang mit unserer Er- fahrung, daB die Liickenbildung und Aggregation bedeckter Schichten bei niedrigeren Temperaturen eintritt als von unbedeckten Schichten. 1 Es est zu beachten, daB die Oxyde mit der SiO-Unterlage in Beriihrung stehen. Veranderungen der Tragerfolie, wie sic anfangs beschrieben wurden. treten bei den hier benutzlen Temperaturen allerdings noch nicht auf. Eine Angabc des Zersetzungsmechanismus nach stochiometrisch festgelegten Umsetzungen, wie sie im vorlicgenden Beispiel versucht wurde, schcint nur in weiiigen Fallen moglich. Wie verwickelt aber auch die Umstiinde im Einzelfall liegcn, der wcsent- liche Vorgang all dieser Erscheinungen scheint in der Lockerung der Oberfliichenatome durch diinnste Adsorptionsschichtcn zu liegcn. Die mitgcteiltcn Ergebnisse beziehcn sich in der Hauptsache auf Praparatveriinderungcn durch Tcm- pern, unter Beriicksichtigung der moglichen Reak- tionen mit Unterlage und Bedeckungen. Es wurde also Thermod>'namik — odcr besscr gcsagt Thermo- statik — elektronenmikroskopischer Priiparate be- trieben. Neben den allgemeinen Bedenkcn, die man bei der Anwendung der Gleichgewichtslehrc auf reale Verhiiltnisse hegen muB, ist besonders zu beachten, daB die Objekte im Gegensatz zu den Ver- suchen im Vakuumofen unter dem Elcktronenstrahl nur noch sehr bedingt thermodynamische Systeme darstellen. Vor allem sind die unter diesen Umstan- den auftretendcn spezifisch clcktronischen EfTckte auf dem oben beschrittcnen Wege nicht erfaBbar. Um ihren EintluB zu beschrcibcn, bedarf es noch zahlreicher Versuchc auf anderer Grundlage. LiTERATUR 1. Bryant, P. J., Rhoads, U. H., und Weber. A. H., J. Appl. Pins. 25, 13,43 (1954). 2. Erdmann-Jesnitzer, F. und Gunther, F.,Z. Me/rt//A«/;(/e 45, 407 (1954). 3. Hanszen, K.-J., Physik. Verhandl. 6, 36 u. 58 (1955). (Kurze Sitzungsbcr.) 4. — Z. Naturforsch. lla, 878 (1956). Z. IDJ 98, 1709 (1956). 5. KoNiG, H.,Z. P/tysik 129,483 (1951). 6. KoNiG, H., Knoch, M., und Brockes, A.,Z. wiss. Mikro- skop. 62, 450 (1955). IV HIGH RESOLUTION ELECTRON MICROSCOPY AND ELECTRON DIFFRACTION Der Durchgang von Elektronenstrahlen durch das Kristallgitter iind seine Folsen filr das elektronenmikroskopische Bild H. NiEHRS Fiitz-Haber-Institut der Max-Planck-Gesellschaft, Berlin-DahJem VoR kurzem sind von Menter (2) die ersten elek- tronenmikroskopischen Abbildungen von Kristall- gitterstrukturen gezeigt worden. Sie mogen insofern noch als recht roh erscheinen, als nicht die Anord- nung der einzelnen Molekiile oder Atome sichtbar wird, sondern nur das streifenformige Bild einer bestimmten Netzebenenschar auftritt. Jedoch scheint der Zeitpunkt nicht fern zu sein, da die Elektronen- mikroskopie systematiscli als Mittel zur Aufklarung von Kristallgitterstrukturen und ihren Storungen eingesetzt wird. Bei diesem Stande der elektronenoptischen Auflo- sung, der in einigen weiteren Vortragen am heutigen Nachmittag zur Sprache kommen wird, ist es wohl auch von Interesse, sich dariiber klar zu werden, welches Bild ein ideales, ungestortes Kristallgitter im Elektronenlicht iiberhaupt zeigen kann; welche Elektronenstwmverteilimg in der beobachteten Stra/il- aiistrittsfldche des Kristalls auftritt, und wie sie von der Bestrahlung und der Objektdicke abhangt. Bei den in Frage kommenden Schichtdicken von bis zu einigen Hundert A ist es wesentlich die elastische Streuung im inhomogenen Potentialfeld des Kristall- gitters, die die Inhomogenitat der Strahldichte in der Austrittsflache hervorruft und das Aussehen des Kristallgitters bestimmt. Daher ist grundsatzlich die dynamische Theorie der Elektroneninterferenzen beru- fen und imstande, iiber die oben gestellten Fragen Auskunft zu geben; denn gerade sie beschreibt das Verhalten von Elektronenstrahlen beim Durchgang durch das periodische Potentialfeld des Kristall- gitters. Auf diese Bedeutung der dynamischen Beu- gungstheorie fijr die Deutung von elektronenmikro- skopischen Kristallstrukturbildern mochte ich hin- weisen und die Aussagen der Theorie in groben Um- rissen darlegen. Von dieser dynamischen Theorie unterscheidet sich die bekanntere geometrische Theorie im wesentlichen durch die Voraussetzung, daB die abgebeugten Strah- len — verglichen mit dem Primarstrahl — sehr schwach sind, so daB einerseits alle Elementarbe- reiche des Kristalls von der gleichen Primarintensitat getroflFen werden, und daB andererseits abgebeugte Strahlen keine nennenswerte weitere Abbeugung er- leiden. Erfiillt ist diese Voraussetzung in sehr diinnen Kristallen oder bei mangelhaft erfijllter Interferenz- bedingung. Infolge ihrer Annahme braucht die geo- metrische Theorie fiir ihrZiel, die Beschreibung des Beugungsbildes, nicht die Vorgdnge im Kristallgitter selbst in Betracht zu ziehen. Die dynamische Theorie geht grundsatzlich anders vor. Sie untersucht zunachst, in welcher Form die Elektronen das Kristallgitter iiberhaupt durchlaufen konnen, d. h. die moglichen sog. Wellenfelder. In einer zweiten Stufe liefert das Studium der Grenz- und Abstrahlungsbedingungen an der Strahlein- trittsflache sodann die Starken der hier abgestrahlten Wellenfelder. In einer dritten Stufe werden schlieB- lich die von der Strahlaustrittsflache abgestrahlten Interferenzstrahlen bestimmt. Die dynamische The- orie betrachtet also wesentlich das dynamische Gleich- gewicht der im Kristall auftretenden Wellenfelder, wenn eine primare Erregung vorliegt. Diese Behand- lung hat den Vorzug, daB sie 1. auch starke Inter- ferenzwirkungen beschreiben kann, und daB sie 2. auch Aussagen iiber das Geschehen //// Kristall- gitter macht. In Ubereinstimmung mit den Beu- gungsbeobachtungen ergibt sich, daB schon fiir Kri- stalldicken von etwa 100 A bei bestimmten Kristall- orientierungen die dynamischen Wechselwirkungen erheblich sein konnen, sodaB die geometrische Theo- rie dann recht unzuliinglich wird, wahrend die dyna- mische Theorie alle experimentellen Ergebnisse der Elektronenbeugung zu erklaren erlaubt. Diese Tat- sache scheint mir zu rechtfertigen, daB sich auch die Elektronenmikroskopie ihrer vertrauensvoU bedient, wenn es sich um Abbildungen des Kristallgitters handelt. Die Grundaussage der dynamischen Theorie ist, daB die Elektronen das Kristallgitter im allgemeinen nicht in Form von ebenen Wellen, sondern nur in Form von Wellengruppen besonderer Struktur T(K, r) = const. v,,^^ . ^-2^'(A'+f>ft.r)^ ft den schon erwahnten Wellenfeldern durchlaufen. Sie konnen gedanklich als Superpositionen ebener Wel- len mit den Wellenvektoren K + b,, aufgefasst werden, worin b;j den reziproken Gittervektor zum Index- tripel // ( ={hjijh,)) bedeutet, d. h. einen Vektorsenk- recht zur Netzebenenschar //, dessen Betrag gleich dem Reziproken des Netzebenenabstands ist. K ist ein fur das betreflFende Wellenfeld charakteristischer Grundvektor, durch den auch die Partialamplituden //„, also die Struktur des Wellenfeldes, eindeutig festgelegt ist, allerdings bis auf den voranstehenden konstanten Faktor, der die Starke des Wellenfeldes beschreibt. (r ist der Ortsvektor im Kristall.) TriflFt ein Primarstrahl einheitlicher Richtung und Energie auf die Grenzflache des Kristalls, so setzt er sich in diesem in einer Linearkombination 3 Dcr Diirchgan},' von Elektronenstnihlen diiich ilas Kiistalli^iller 87 solcher Wellent'elder fort, wobei die verschicdencn Grundvektoreii K; der einzelncn Wcllenfelder sich selbst zwar sehr wenig voneinander und von dcm primiiren Wellenvektor unterscheiden, die Amplitu- dcnstruktur der Wellcnfelder abcr sehr untcrschicd- lich sein kaiin. Die Anzahl dcr in dieser Linearkombi- nation enthaltenen Wcllenfelder ist eng verkniipft mit der Anzahl (A^ 1) dcr Partialwelien je eines Wellenfcldcs. wobei TV zuglcich die Miihiplizitiit dcr Inwrfercnz ausdriickt. Tatsiichlich wiirde im intcr- ferenzfreien Fall A^ 0 nur 1 Wellenfeld bcstehend aiis nur 1 Partialwelle // (000) aiiftreten. Dieser interferenzfreie Fall kommt aber praktisch nie vor, und es sei sogleich bemerkt, dafJ cs besonders bei der Bestimmung der Stromverteilung in der Strahl- austrittstliiche des Kristalls meistens wesentlich auch auf die zahlreichen schwachen Interferenzen an- kommt. Bei einer N-fachen Interferenz treten, von Sonderfiillen abgesehen, als Fortsetzung des Pri- miirstrahls bei Durchstrahlung A^ -i- 1 Wcllenfelder auf, von denen jedes eine unabgebeugte Partialwelle zum Indextripel /; == (000), sowie A^ abgebeugte Par- tialwelien zu Indextripcln /; ^r- (000) umfasst. Die gruppenweise Zusammenfassung der insge- samt (A' ; 1)- Partialwelien im Kristall zu A' + 1 Wellcnfeldern mag zunachst recht willkiirlich und sinnlos erscheinen; sie hat aber einen sehr realcn physikalischen Grund. Die partiellen Losungen der Schrodingcrgleichung im Kristallraum sind eben nicht die einzelncn cbcncn Partialwelien, sondern nur die komplctten Wcllenfelder. Nicht die einzelncn ebenen Partialwelien, wohl aber die einzelncn Wcl- lenfelder konnen im Kristall unabhiingig voneinan- der fortschreiten. Das Wellenfeld ist mithin Triigcr dcr Elektronenstromung, und bei geniigcnd engcm Strahlquerschnitt stellt cs einen Elektroncnstrahl mit eigener Strombahn bestimmtcr Richtung dar. Die Fortschreitungsrichtung eines Elektronenstrahls im Kristall ist daher nicht unmittelbar durch die Wel- lenvektoren K i b^ gcgeben, sondern durch den Stromdichtevektor des komplctten Wellenfeldes. Es ist daher besser, den Elektroncnstrahl im Kristall nicht als Superposition von ebenen WcUen, sondern als eine Elektronenstromung, die mit der Periodizitcit des Gitters modidiert ist, aufzufassen. Die Multiplizitat der Wcllenfelder andcrerseits be- dcutet, daB der Primiirstrahl sich beim Eintritt in den Kristall unter Mehrfachbrechnng, bei A'-fach- Interferenz unter (A'^ + l)-fach-Brechung, fortsetzt. Ebenso erleidet das einzelne Wellenfeld im Kristall beim Auftreffen auf cine Grenznache cine Mchr- fachreflexion in neue Wcllenfelder, die jedoch prak- tisch stets zu vcrnachlassigcn sind. Beim Austritt aus dem Kristall freilich zcrfiillt das Wellenfeld (unter gleichzeitigcr Brechung) in seine Partialwelien, da diese im Vakuum unabluingig voneinander als Inter- fercnzstrahlen weiterlaufen konnen. Erst hier bc- stimmt der (durch Brechung modifizierte) Wellen- vektor K + b;j wieder die Strahlrichtung. Unter den ublichcn cxperimentellcn Bedingungen haben die \on den verschiedenen Wcllenfcldern / umfaBten Partialwelien gleichen Indextripels // nur sehr wenig verschicdene Wellenvektoren K^ f b/j. Zu- sammen liefcrn sie nach Austritt einen aiifgespaltenen Interferen/strahl li, wobei die Starke der Aufspaltung sehr wesentlich auch von der Lage der Austritls- niiche abhiingt. Beim planparallelen Kristall setzen sie sich geradc zu einem iinaiifgespaltenen Interfe- rcnzstrahl zusammen. Die Superposition der Partialwelien zu verschie- denen indextripeln h erzcugl in der Strahlaustritts- lliiche eine charakteristische Elektronenstromvertei- lung, die die periodische Struktur des Kristallgitters widcrspiegelt. Hinsichtlich der Intensitiiten jedoch besteht kcinc einfache Korrelation mit der Potcn- tialvcrtcilung im Kristallgitter. Lcdiglich wenn die Primarstrahlrichtung in eine Synimeirieaclise des Kristalls fiillt, darf man erwarten, daB vor den Ato- men der Austrittsllache sich auch Symmetriezentren (Maxima oder Minima) der Jntensitdt ausbilden. Die Multiplizitat der Wcllenfelder iiuBcrt sich beim plan- parallelen Kristall ferner in einer Abhiingigkeit der Stromverteilung von dcr Kristalldicke. Diese Ab- hiingigkeit ist bei A'-fachcr interferenz darstellbar durch Superposition von A' periodischen F-unktionen der Kristalldicke. Uber die Ergebnissc einer solchcn Berechnung dcr Stromverteilung in der Strahlaus- trittsfliichc des MgO-Kristalls wird andcrnorts he- richtet (3). Bei Durchstrahlung keilfiirmiger Kristallbereiche von Polyedern erscheint diese Dickenabhiingigkeit bereits innerhalh eines clektronenmikroskopischen Bildes, da der Strahlwcg im Kristall dann von Objcktpunkt zu Objektpunkt variiert. Im Falle dcr einfachen Interferenz z. B., bei der im Kristall 2 Wcllenfelder laufen, tritt im Bild die bekannte ein- fach-periodische Streifung parallel zu den brechen- den Kantcn auf. wie sie crstmals von Heidenreich und Kinder (1942 43) beobachtet und \on Kosscl (1943) als ,,Linien glcicher Kristalldicke" gcdeu- tct worden ist. Bei der Aullosung von Kristallgit- terstrukturen im elektroncnmikroskopsichen Bild wi.irdc sich diese Streifung mit ihren Periodenbreiten von etwa 100 A dcr gitterperiodischen Stromvertei- lung uberlagern, wodurch sich unter I'mstiinden ein zicmlich kompliziertes Bild ergibt. Zum AbschluB sei noch ein Hinweis auf die Absorption der Elektroncnstrahlen, genauer die Schwiichung der kohiircnten Biindels, im Kristall gestattet. Eine der Opiik analoge Erweiterung der dynamischen Theoric unter Beriicksichtigung der Absorption liiBt vermuten, daB die verschiedenen. als Folge der Interferenzen auftretenden Wcllenfelder sich nicht nur hinsichtlich der Brechung, sondern auch hinsichtlich der Absorption verschiedcn ver- halten. Dies liiBt sich im einfachsten Falle am Beispiel optimaler Einfachinterferenz anschaulich in folgender Weise verstehen: Die beiden in diesem Falle auftretenden Wcllenfelder enthaltcn je 2 Par- tialwelien und stellen Elektroncnstrahlen dar, die 88 J. W. MENTER quer zu der die Interferenz hervorrufenden Netz- ehenemchar modidiert sind und parallel zu dieser Netzebenenschar fortschreiten. Die Modulation des einen Wellenfeldes ist derart, daB die Stromdichte- maxima, die Schwebungsbauche, gcnaii in den Netz- ebenen, die Stromdichte/;/m//Hrt, Schwebungskno- ten, in der Mitte zwischen den Netzebenen liegen. Genau umgekehrt ist die Modulation des anderen Wellenfelds: Maxima zwischen, M/>7/ma m den Netz- ebenen. Das erste Wellenfeld fuhrt seinen Elektro- nenstrom also hauptsachlich in den Aromehenen, das zweite Wellenfeld hauptsachlich rii7,vf7/('// den Atom- ebenen durch das Kristallgitter. Es ist verstandlich, daB der Elektronenstrom des ersten Wellenfeldes von der Absorption durch die Atome starker be- trofTen wird als der des zweiten Wellenfeldes. Theorie und Beugungsbeobachtungen ( 1 ) zeigen iibereinstim- mend — quantitative Messungen liegen freilich nicht vor — , daB von den beiden bei Einfachinterfe- renz beobachteten Wellenfeldern das eine iibernor- mal stark, das andere iibernormal schwach im Kri- stallgitter absorbiert wird. In der elektronenmikro- skopischen Beobachtung sollte sich dieser Eflfekt in einer iibernormalen Dwchldssigkeit von Kristallen groBerer Dicke bei Interferenzstelhingen kundtun. Vielleicht bietet sich dadurch die Moglichkeit bei hinreichender elektronenmikroskopischer Autlosung, den Strahlweg im Kristall zu demonstrieren oder auch Kristallgitter groBerer Dicke abzubilden. Ich mochte nicht versaumen zu erwahnen, daB die Entwicklung, der hier skizzierten Theorie weitge- hend gefordert worden ist durch analoge Beobach- tungen und (Jberlegungen von v. Laue und von Borrmann auf dem Gebiet der Rontgeninterferenzen. LiTERATUR 1. Altenhein und Moliere, Z. Pliysik (1954). 2. Menter, J. W., Proc. Roy. Soc. A 236, 119 (1956). 3. NiEHRS, H., Optik (1956). The Resolution of Crystal Lattices J. W. Menter Tube Investments Research Laboratories, Hinxton Hall, Essex. The possibility of resolving atoms in the electron microscope has been discussed by a number of wor- kers including Hillier (4), Schiff (8), Boersch (2), Scherzer (7) and Haine (3). The basic assumption of their treatments has been to consider two isolated atoms each forming its own Airy disc pattern in the image. It is assumed that there is no coherence be- tween the electrons scattered from these two point objects and the criterion for resolution is that the intensity dip between the Airy disc patterns shall just be perceptible. If the atoms or molecules are arranged in a regular array, as in a crystal lattice, for example, resulting in definite phase relationships between electrons scattered from neighbouring atoms then the mechanism of image formation is rather different. Strong diffracted beams are formed and it has been pointed out by some of these workers that under these conditions the instrumental resolution required to resolve regular arrays of atoms or mole- cules is not so high as in the case of isolated noncohe- rent objects. The resolving power of the Siemens Elmi- skop I as limited by the diffraction error and spherical aberration alone is 2.8 A. This is worsened in practice by the chromatic error and astigmatism to 7 A. We were thus encouraged to attempt the direct observa- tion of crystal lattices in crystals with relatively large lattice parameters. After the macromolecular crystals of viruses and proteins, the structures of which have been beautifully demonstrated by WyckofT (9) using replica methods, the most likely class of compounds with smaller lattice parameters are organic molecules Fig. 1. Schematic projection of platinum phthalocyanine Fig. 2. Unit cell of platinum phthalocyanine crystal. Circles molecule normal to the phthalocyanine ring (after Robertson represent molecules, the metal atoms being situated at the and Woodward 1940). • C; O CH; O, N centre of the circle. The Resolution of Crystal Lattices 89 Fig. 3. Crystal habit ofcoppcr plitlialocsanine (after Robert- son, 1934). with intermediate molecular weights. Among these the phthalocyanines appeared a promising group to study and most of our work so far has been carried out on copper and platinum phthalocyanine. The metal phthalocyanines have a number of favourable properties. The molecule of platinum phthalocyanine is shown in figure I, and the unit cell of the crystal lattice in figure 2. From these it is apparent that we may idealise the structure into widely spaced planes of heavy metal atoms (^^o,,, 1 1.94 A) embedded in a matrix of the light elements carbon, nitrogen and hydrogen. We would expect to obtain strong diffracted beams from these planes even in very thin crystals since the scattering from the heavy metal atoms swamps that from the organic parts of the unit cell. The crystals grow as long thin ribbons with (001) as the habit plane of the ribbon surface as shown in figure 3, so that crystals sup- ported on a specimen grid will be oriented with (001) perpendicular to the electron beam and (20T) almost Fig. 4. Schematic representation ol"(20T) planes in platinum phthalocyanine in relation to crystal habit. ABCD is a (001) plane. The almost vertical sheets such as ABEF arc (20T) planes. parallel to the beam, since (001)A(20T) =88". The (201) planes are thus in a favourable orientation for diffraction, which is essential in order to form an image of the planes. A schematic representation of the nearly vertical (20T) planes is shown in figure 4. In copper phthalocyanine the corresponding para- meters are (001)A(20T) 80 , f/201 = 9.8 A. Preparation of specimens and method of examina- tion.— The crystals were prepared by sublimation from the powder after the method of Barrett, Dent & Linstead (I). After irechanical breaking down to reduce their size they v\ere suspended in ethyl alcohol, and a drop of the suspension dried down on to a Fig. 6. Single edge dislo- cation in platinum phtha- locyanine crystal ( 1,000.000). Fig. 7. Guide to figure 6 showing exact position of edge dislocation. Fig. 5. Portion of platinum phthalocyanine crystal showing perfect structure of (20T) planes ( < 1,500,000). 90 J. W, MENTER HlWWi^ m Fig. 8. Copper phthalocyanine crystal showing dislocated region associated with change in width of crystal. nitrocellulose supporting film, containing a large number of holes. It was found that the image was seen to the best advantage when not overlaid with the structure of the supporting film. The specimens were examined in the Elmiskop I operated at 80 kV, using the fine focus condenser (aperture 200 //) and a 50 /( objective aperture (unless otherwise stated) and recorded on Ilford Contrasty Lantern plates at a magnification of 77,000 times. Results. Platinum phthalocyanine. — A considerable number of plates revealed the structure shown in fig- ure 5, consisting of a series of parallel linesinthe[010] direction, the spacing of which was 12.0 A, averaged from measurements on 26 crystals with a standard deviation of 0.2 A (see Table 1 ). These lines may be regarded as the image of the projection of (201) planes seen edge on. Bent crystals have been observed in which the bending of the crystal planes follows the external form of the crystal as would be expected. Imperfections are sometimes seen in the structure in the form of edge dislocations. A particularly simple example is shown in figure 6, the exact position of the incomplete plane being clarified by the sketch in figure 7 which has been copied from the micrograph. Copper phthalocyanine. — Similar results have been obtained with copper phthalocyanine, although less frequently, since the probability of finding a crystal in a suitable orientation for diftYaction is lower on account of the smaller value of (001)A(20T). Two values have been obtained for the spacing of the planes. The first 10.30 ±0.3 A averaged from eight measurements differs significantly from the x-ray value of 9.8 A, and a few isolated values of about 13 A have been observed. There are a number of possible reasons for this discrepancy which are dis- cussed more fully elsewhere (5). A particularly good example of a dislocated lattice is shown in figure 8 where a severe disturbance of the lattice is associ- ated with a change in width of the crystal at the point X. Cracks have been observed in crystals in which the crack may be seen to propagate from one plane to its neighbour as it traverses the lattice. Sodium faujasite. — This is the first inorganic mate- rial in which the crystal planes have been resolved. It is a network silicate structure with the composition 2Al,O3-CaO.Na2O.10 SiO,.20 H.p, being cubic with ao = 24.84 A. The (111) spacing is 14.37 A and the mean of measurements from 16 micrographs gives ^,111; =" 14.4 ± 0.2 A (see Table 2). Figure 9 shows a crystal of this compound revealing the (111) planes. Fig. 10 shows a crystal viewed along the [110] axis in which two sets of ( 1 1 1) planes intersecting at 70^ are resolved. Mechanism of image formation. — The crystals being thin, form a cross-grating diffraction spectrum since the third Laue condition for diffraction from Table 1. Spacing of lines in platinum phthalocyanine. (All plate magnifications 77,000 ) Plate no. No. of spaces measured Average distance be- tween lines on plate (mm) Distance between lines in crystal (A) 1508 220 0.0935 12.1 1481 80 0.0914 11.9 1498 50 0.0916 11.9 1517 20 0.0955 12.4 1532 100 0.0926 12.0 1582 80 0.0920 11.9 1518 30 0.0950 12.3 1525 40 0.0950 12.3 1573 30 0.0933 12.1 1670 50 0.0934 12.1 1630 100 0.0917 11.9 1626 40 0.0907 11.8 1625 40 0.0900 11.7 1624 50 0.0910 11.8 1652 116 0.0919 11.9 1400 200 0.0919 11.9 1445 30 0.0917 11.9 1446 20 0.0925 12.0 1449 20 0.0895 11.6 1442 100 0.0935 12.1 1171 20 0.0935 12.1 1196 20 0.0890 11.6 1213 50 0.0932 12.1 1173 100 0.0900 11.7 1204 30 0.0940 12.2 1192 30 0.0927 12.0 The Resolution of Crystal Lattices 91 Fig. 9. Sodium faujasite crystal showing (111) planes. lattice rows in the direction of the beam is highly relaxed. With a 50 /< objective aperture, the spectra contributing to the image from the phthalocyanine crystals are the (20T), (402), 201) and (402). These spectra are assumed to recombine with the zero order beam in the image plane and form an image of the crystal grating in accordance with the simple Abbe theory of image formation by a lens. This has been confirmed by excluding all spectra except the zero order from the image by using a 10 // objective aperture when it is found that the image of the planes disappears. Further confirmation of this mechanism is obtained from the fact that an array of parallel lines rather than a cross grating of dots is seen in the image. The projection of the lattice in the direction of observation should show a series of planes (010) perpendicular to the (20T) planes. The spacing of ^^^l^r- iHM m these planes however is 3.81 A and the first permis- sible spectrum from them, (020), corresponds to a Bragg spacing of 1.905 A. This is excluded from the image by a 50 // aperture so that the image of these planes is not seen. It is doubtful on other grounds whether such a small spacing could be resolved. Firstly, the permis- sible misorientation of the crystal with respect to the electron beam, arising from the relaxation of the third Laue condition becomes smaller as the spacing to be resolved is reduced. Secondly the effect of spherical aberration in distorting the wavefront of a beam diffracted at a wide angle through the lens becomes so severe as to cause the diffracted beam to be no longer able to interfere coherently vsith the zero order beam and form the image. The effect of spherical aberration may already be severe even J. W. MENTER Fig. 10. Sodium faujasitc crystal viewed along [110] showing two sets of ( 1 1 I) planes intersecting at 70 '. with the (402) reflection corresponding to ^ = 5.97 A since a microphotometer trace across the image shows that the intensity distribution follows fairly closely a function of the form cos- 0. This would be expected from interference between two beams, i.e. the zero order and the first order (20T) indicating that although the (402) beam passes through a 50 /( aperture to the image it is making no useful contri- bution to the image of the planes. Resolution of image. — The high resolution appar- ent in the image can be explained in terms of simple lens theory: neglecting effects of chromatic aberra- tion and astigmatism, the phase delay imposed on a ray passing through a lens at an angle -x is given by e = J Cja*, where Cj is the spherical aberration con- stant. The diffracted beams from the crystal lattice may be regarded as plane parallel beams approxi- mately equal in width to the width of the specimen (neglecting the finite divergence of the incident illu- mination). Thus provided a and Cg are small the distortion of the wave front of the narrow diffracted beam in passing through the lens may be very small (7). A simple calculation shows that the difference in phase between the two ends of a wavefront of width Ir is given by Gr-* Ar /\ where /is the focal length of the lens and /■ the distance of the beam from the axis in the lens plane. Inserting values for the (20T) reflection from platinum phthalocyanine, /• 10-3 cm, Cs 0.28 cm,/ 0.3 cm, Ar 10^ cm, we find that the phase difference across the wavefront Ae = 3 > 10 '^ cm, i.e. ^e< /, since A =4 lO"'" cm. Thus the wavefront remains virtually undistorted in passing through the lens and is able to form an image of the (20T) planes by interference with the Ahhihhinii von Kristallf^'iitcrsinik lurcii 93 Table 2. Spacint; of lines in sodiiini faiijcisite. (All plate niagnitications 77,000 ) Distance between lines in eivstal (A) 14.6 14.7 14.15 13.9 14.3 14.55 14.8 14.65 14.4 14.6 14.4 13.8 14.0 14.45 14.55 14.7 zero order beam. Calculations suggest that it should be possible to resolve planes with spacings consider- ably smaller than 10 A providing that the divergence of the illuminating beam is made sufficiently small. Average No. of distance be- Plate no. spaces tween lines measured on plate (mm) 2020 100 0.1125 20 0.1135 2025 24 0.109 20 0.107 40 0.110 1959 80 0.112 70 0.114 30 0.113 1827 30 0.111 1826 20 0.1125 1834 30 0.111 1962 20 0.106 20 0.108 2030 30 0.111 2032 30 0.112 2036 20 0.1135 Should this prove to be true in practice there are many possibilities opened up in the study of crystal structures and their imperfections. The images ob- tained are, of course, very poor Fourier projections and show none of the detail obtained by x-ray analy- sis but they have the essential advantage of revealing the exact location of imperfections, thus permitting the direct study of the behaviour of dislocations under a variety of physical conditions. The idea of these experiments germinated in the course of a discussion with Mr. .1. h. Gordon. I wish to thank Mr. G. A. Bassetl for his careful attention to the per- formance of the microscope, Mr. R. W. Gooding for preparing the platinum phthaiocyanine. Professor R. M. Barrer and Mr. J. A. Gard for pro\iding samples of sodium faujasite. This paper is published by permission of the Chairman of Tube Investments Ltd. References 1. BARRF.rr, P. A., Dent, C. E., and Linstead, R. P., J. Chem. Soc. 1719 (1936). 2. BoERSCH, H.,Z. Naiiirforsch. 2a, 615 (1947). 3. Haine, M. E., AilviUic. Ek'ilronics Eleclrun Phvsics 6, 295 (1954). 4. HiLLiER, J., Pins. Rev. 60, 743 (1941 ). 5. Menter, J. W., Proc. Roy. Soc. A 236, I 19 (1956). 6. R0BERT.SON, J. M., J. Clwm. Soc. 615 (1934). 7. SCHERZER, O., J. Appl. Phys. 20, 20 (1949). 8. SCHIFF, L. I., Pliy.s. Rev. 61, 721 (1942). 9. Wyckoff, R. W. Ci., Lcs techniques rcccntes en micros- copic electroniquc cl corpusciilairc. Coll. Int. du CNRS, 135 (1956). Elektronenmikroskopische Abbildung von Kristallgitterstriiktiiren R. Neider Fiitz-Haber-Institut cler Max-Planck-Gesellschaft, Berlin-Dahlcni Die von der dynamischen Theorie derElektronenin- terferenzen ausgehenden Uberlegungen von Niehrs, welche dieser im vergangenen Jahr in unserem Insti- tut begonnen hatte, fijhrten zu Beginn dieses Jahres zu dem Ergebnis, daB die periodische Struktur des Kristallgitters eine ebensolche Struktur der Strahl- dichte in der Austrittsfliiche hervorruft, und — dies ist das bemerkenswerteste Ergebnis der Niehrsschen Arbeit — daB Atome oder Netzebenen mil vollig ausreichendem Kontrast abgebildet werden konnen. Voraussetzung ist natiirlich, daB die abzubildenden Netzebenenscharen von dem benutzten Hlektronen- mikroskop aufgelost werden, d. h. daB die /ugeho- rigen Interferenzstrahlen bei der Bildentstehung mit- wirken und die Storung diucii Linsenfehler genii- gend klein ist. Da uns das Elmiskop I der Siemens & Halske AG zur Verfijgung stand, suchten wir nach Kristal- len, die Netzebenenabstiinde der GroBenordnung 10 A und fijr die Abbildung geeignete Kristalltracht haben und schlieBlich als Untersuchungsobjekt ge- niigend bcstiindig sind. Hicrbei sticBen wir — ebenso wic Menter — auf die Gruppe der Metallderivate des Phthalocyanins. Wir erhielten Nickel-i'hlhaloc\anin von Drechsler und WoltT. welche in unserem Institut dicse Substanz auch bei ihrcn Untersuchungen im Feldcmissionsmikroskop \er\\endet hatten. Kupfer- Phthaloc\anin erhielten uir von der Badischen .'\ni- lin- und Soda-Fabrik. Wiihrcnd wir mit Elektronenbeugungsuntersu- chungen und ersten Abbildungs\ersuchen beschiiftigt waren, erhielten wirdurch private Mitteilung Kenni- nis von den schcinen elektronenmikroskopischen Aufnahmen, die Menter ebenfalls mit di-ni ..Elmiskop I" von Pt- und Cu-Phthalocyanin erluiUcn hattc. E.xpcrinicniclh' Eri^'civiisse. — Da man schon aus dem Beugungsbild eines Einkristallcs entnehmen kann. ob Netzebenen dieses Kristalles elektronen- mikroskopisch abbildbar sind, untersuchten wir zuerst die Beugungsbilder von Ni- und Cu-Phthalo- 94 R. NEIDER Abb. 1. a. Elektronenbeugung von Ni-Phthalocyanin. Beugungsliinge: 45, Phthalocyanin. Bcugungslange: 45,1 (mm • A). (mm • A), b. Elektronenbeugung von Cu- cyanin, und zwar wegen der Kleinheit der Kristalle mit der Methode der Kleinfeldbeugung mittels Selek- torblcnde. Abb. I a, h zeigt Elektronenbeugungsdiagramme von Ni- und Cu-Phthalocyanin. Durch Vergleich mit der Rontgenstrukturanalyse von Robertson und Woodward (2. 3) wurden die engsten Refiexe in bei- den Diagrammen als 20T bzw. 201 (9,86 A) und die dazu senkrecht liegenden Refiexe als 110 bzw. 110 (4,65 A) identifiziert. Die beiden Verbindungen Ni- und Cu-Phthalocyanin unterscheiden sich in ihrem kristallographischen Aufbau nur sehr wenig. Beim Cu-Phthalocyanin wurde aber auch noch ein ande- res Beugungsbild gefunden: Abb. lb. Abb. la zeigt die Mikrophotographie von zwei Einkristallen aus Cu-Phthalocyanin, von denen das Beugungsbild aufgenommen wurde. Im Diagramm (Abb. Ih) sehen wir einmal die oben schon gezeigten Reflexe 20T bzw. 201 und 110 bzw. iTO von Cu- Phthalocyanin. Zusatzlich tauchen aber in der Nahe von den Reflexen 20T bzw. 20! noch ein paar enger liegende Refiexe auf, die einem Netzebenenabstand von 12,12 A entsprechen und von dem schmalen Kristall stammen, wie durch Abdeckcn des anderen Kristalls festgestellt wurde. Nimmt man an, daB es sich bei dem schmalen Kristall auch um Cu-Phthalo- cyanin und um die gleiche ModiHkation wie bei den anderen Kristallen handelt, so entsprechen diese Reflexe den 001 bzw. OOl-Netzebenen. Das Erschei- nen dieser Reflexe legt den SchluB nahe, daB der schmale Kristall eine andere Tracht als der breite Kristall besitzt. Auch aus einigen anderen elektro- nenmikroskopischen Bildern ziehen wir den SchluB, daB ftir Cu-Phthalocyanin nicht immer die Tracht vorliegt, wie sie Robertson angegeben hat. Auf anderen Beugungsaufnahmen mit den Reflexen 001 bzw. 001 fanden wir in der senkrechten Richtung dazu Reflexe. die einem Netzebenenabstand von 3,78 A entsprachen und mit 310 bzw. 310 indiziert werden konnen. SchlieBlich wurde bei manchen Cu- Phthalocyaninkristallen noch ein drittes von den beiden vorhergehenden Beugungsbildern verschiede- nes Beugungsdiagramm gefunden: Abb. 3. Dieses Diagramm weist die gleichen Reflexe wie das Diagramm mit der Kombination 001 ,310 auf, aber zwischen den verschiedenen Ordnungen von 001 erscheinen noch zusiitzliche Reflexe. Sie entsprechen dem doppelten Netzebenenabstand, niimlich 22,6 A, und sind mit den bekannten kristallographischen Daten des Cu-Phthalocyanins nicht vereinbar. Die Abb. la, b. Elektronenbeugung von Cu-Phthalocyanin mil dem die Beugung erzeugenden Objektbereich. VergroBerung: 39000. Beugungsliinge 43,1 (mm • Aj. Ahbihlimg von Kiislalli,'itterstniktwen 95 Abb. 3. ElektroncnbeugLing von Cu-Phihalocyanin. Beu- gungslange: 56,5 (mm • k). zusatzlichen Reflexe verschwinden nach gewisserZeit der Elektronenbestrahlung, so daB nur noch die 001 -Reflexe iibrig bleiben. Die entsprechende Er- scheinung konnten wir im elektronenmikroskopi- schen Biid beobachten: Wahrend zuerst Netzebenen mit einem Abstand von 22,4 A abgebildet wurden, halbierte sich dieser Abstand nach 2-3 Minuten, so daB die Netzebenen nun einen Abstand von 1 1,5 A hatten. Die Erkliirung fiir diese bemerkenswerte Erscheinung steht noch aus. Bei der Abbildung der Netzebenen kommt es in erster Linie darauf an, daB die entsprechenden In- terferen/strahlcn bei der Abbildung mitwirken und nicht durch die Aperturblende abgefangen werden. Eine Aperturblende von 50 // laBt in den Diagram- men von Abb. 1 «, h und Ih jevvcils die I. und 2. Ordnung von 20T bzw. 201 und 00! bzw. 00!" durch. In Abb. 3 werden auf jcdcr Seitc vom Nullstrahl fiinf von lAcu cng liegcnden Rellexen durchgelassen. Zur Kontrolle wurde vor den Aufnahmen jeweils das Bcugungsbild der Kristalle betrachtet. In Abb. Aa, h, c sind 20T-Netzebcncn von Ni- Phthalocyanin mit einem Abstand von 9,8 A, 001- Netzebenen von Cu-i*hthalocyanin mit einem Ab- stand von 11,5 A und Netzebenen des Cu-Phthalo- cyanins mit 22,4 A Abstand zu sehen. Die Kristalle wurden durch Sublimieren an Luft hergestelll und in Alkohol oder Ather suspendiert. Fiir die Abbil- dung wurden 7-Loch-Blenden mit graphiticrten Kol- lodiumfolien, die vorher im Elektronenmikroskop einer intensiven Elektronenbestrahlung ausgesetzt wurden, oder netzartig aul'gerissene Formvarfolien benutzt, so daB die Kristalle an Stellen abgebildet werden konnten, an denen sic iiber ein Loch ragten, und so der Folienuntergrund nicht storte. Aufge- bracht wurden die Kristalle durch Verneblung der Alkohol- oder Athersuspension mit 3 MHz Ultra- schall. Bei den Aufnahmen wurde mit Riicksicht auf mciglichst geringe Erwiirmung meist nur ein kleiner Objektbereich mit dcm Feinstrahlkondensor durch- strahlt. Um bei der Beobachtung und Aufnahm.e un- mittelbar vom Bcugungsbild zum mikroskopischen Bild mit aufgelosten Netzebenen iibergehen zu kon Abb. 4. a. 20T-Netzebenen (9.8 A) von Ni-Phthalocyanin. VcrgroBerung: 2 600000. cicktroncnmikroskopisch: 150000. b. Netzebenen von Cu-Phthalocyanin (11,5 A Abslantl). VcrgroBerung: 2 600000, elekironcnmikroskopisch: 150000. c. Netzebenen von Cu-Phthalocyanin (22.4 A Abstand). VcrgroBerung: 1 300000, elcktronenmikroskopisch: 39000. Abb. 5rt, b. Elektronenbild eines Einkristalls aus Cu-Phthalocyanin. Stereopaar stand: 11,5 A. VergroCerung: 750000, elektronenmikroskopisch: 150000. mit Stereowinkel < 3°. Netzebenenab- Ahhildimg von Kristallgitterstrukturcn 97 nen, vvurde die Schaltung des Elmiskop I abgeandert. Der Kondensor K 1 vvurde aus dem Stromkreis des Mikroskops herausgenommen, durch cine unabhan- gige Spannungsquelle von 120 V gespcist iind mit einem eint'achen Schiebcwiderstand eingestcllt. Es wurde bei moglichst schwachen Intcnsitatcn am Objekt (10"^ A cm- - 10^* //A /<-) gearbeitet, wcil die Kristallage sonst nicht stabil genug war oder die Kristalle gar verdampftcn. Die Aufnahmen vviirdcn auf Kranz-Feinkorn-Platten (15 10 DiN) mit Be- lichtungszeiten von 1 min gemacht. Ahbildimg der Netzebencn hci Ahweichimg vom Bragg-Winkel. — Wir untersuchten weiterhin die Frage, wie weit dcr Winkel zwischcn Netzebene und einfallendem Strahl abweichen darf, ohne dali der Kontrast in der Abbildung der Netzebenen allzu stark abnimmt. Ein Cu-Phthalocyanin-Einkristall, dessen 001 -Netzebenen sich in Interferenzstellung befanden. wie mit Hilfe der Kleinfeldbeugung fest- gestellt wurde. wurde bei 1 SOOOOfacher VergroBerung photographiert. Dann neigten wir das gesamte Prii- parat mit der Stereoeinrichtung urn 3 . Der Kipp- winkel um eine Achse parallel zu den Netzebenen betrug daher keinesfalls mehr als 3 , wahrscheinlich sogar weniger. Nach dieser Neigung wurde der Kristall noch einmal photographiert (Abb. 5a, h). Man erkennt auf dem oberen Bild, daB die Netz- ebenen nur in einem Teil des Kristalles mit hohem Kontrast abgebildet werden. Auf dem unteren Bild hat sich der Bereich, in dem die Netzebenen mit hohem Kontrast abgebildet werden, verlagert. Dar- aus muB geschlossen werden, daB der Kristall ver- bogen Oder geknickt ist. In einem Bereich des Kri- stalls, der sich dem Gebiet guter Abbildung der Netzebenen anschlieBt, erkennt man auf beiden Bil- dern eine Kriimmung von Netzebenen und Unregel- miiBigkeiten im Gitter. An dieser Stelle liegt ofTen- sichtlich der Knick im Kristall. Die Bilder beweisen, daB die Abbildung der Netz- ebenen schon bei einer Abweichung von etwa 3" vom Bragg-Winkel fast ganzlich verloren geht. Men- ter (I), der sich experimentell und theoretisch eben- falls mit dieser Frage beschaftigt hat, kommt zu dem Ergebnis, daB die Intensitiit des Interferenzstrahls 20T erstmals bei 1 ,2' Abweichung vom Bragg-Winkel fiir 600 A Kristalldicke und bei 100 A Kristalldicke erstmals bei 7 Abweichung vcrschwindet. Daraus darf man unserer Meinung nach nicht den SchluB Ziehen, daB die Netzebenen bei einer Neigung von 7" gegen den Primarstrahl noch mit genugend hohem Kontrast abgebildet werden, wenn nur die Kristall- dicke klein genug ist. Unsere Berechnung der Inter- ferenzintensitat bei optimaler Kristalldicke crgibt bei einer Abweichung von 2,5 vom Bragg-Winkel einen Abfall der Intensitat auf weniger als 4 "o des Maxi- malwertes, bei 10 auf 0,1 "„. Da dcr Bragg-Winkel 0,5 ist, muB angenommen werden, daB bei Abbil- dung der Netzebenen diese annahernd parallel zum Primarstrahl liegen. Die Tatsache, daB die sehr dunnen Kristalle ver- bogen oder geknickt soin konnen und so nicht iiber den gesamten Bereich des Kristalles die gleichen Interferenzbedingungcn erfiillt sind, erkliirt auch das gleichzeitige Erscheincn dcr Rcflcxc 20T und 110 in den erstcn beiden Bcugungsbildcrn, die wir sahcn. Bei einem unverbogcnen Kristall vviire dies niimlich gar nicht moglich. Es muB daher angenommen werden, daB die 20T- und 1 10-Netzebenen des Ni- oder Cu-Phthalocyanins nicht glcich/eitig an dcnscl- ben Stcllen des Kristalls rcUckticren, sondern wegen einer Vcrbiegung oder Knickung des Kristalls an verschiedenen Stellen, die aber immer noch im Bereich des zur Beugung gelangenden Objektfeldes liegen. Thcorctisclw Ergchnissc. — Um wcnigstens einen groben qualitativen Vergleich dcr experimentellen Ergebnisse mit der Theorie durchfiihrcn zu konnen, wurden die zu erwartende Dichtevcrteilung des Elek- tronenstroms an der Strahlaustrittslliiche eines Ni- Phthalocyanin-Einkristalls untcr folgenden verein- fachenden Annahmen nach dcr dynamischen Theo- rie berechnet. 1. Absorption und inkoharentc Streuung werden gegeniiber der kohiirenten Streuung vernachliissigt, weil es sich um eine sehr diinne Schichtdicke handelt. 2. An Stelle des Ni-Phthalocyanin-Molekiils werden im Kristallgitter je ein Ni-Atom angenommen. denn obwohl die Kohlenstoff- und StickstofTatomc alle zusammen ein groBeres Streuvermogen gcgenCibcr der Elektronenstrahlung haben als ein einziges Ni- Atom, wirkt sich dieses nicht so stark aus, weil sie auf einen viel groBeren Raum verteilt sind und sich daher wahrscheinlich erst bei Interferenzen hoherer Ordnung bemerkbar machen. 3. Wiihrend im Mikroskop erst durch die Apertur- blende alle Interferenzstrahlen bis auf die 1. und 2. Ordnung von 20T kiinstlich zu Null gemacht werden. setzen wir voraus, daB nur die Inlerferenzstrahlcn D --588 (980) S D-39Z(H76)S D--m(i372) 4 D-0(-f5b8)A — X — ^ iO ^ Abb. 6(/. Intensitatsverteilung der Eleklronenstrahldichtc in dcr Austrittsflachc cincs Ni-Phihalocyanin-Einkristalls fiir verschiedene Kristalldickcn (.theoretisch). i I Abb. 6/?. Experimentelle Kurve, gewonnen durch Photo- meirierung senkrechi zu den 20l-Netzebenen des Ni-Phthalo- cyanins im Negativ. 7 — 568204 Electron Microscopy 98 W. D. RIECKE 20T und 201 auftreten. Dabei werden die entsprechen- den Interferenzstrahlen 2. Ordnung vernachlassigt, die oft sehr schwach sind und, wie experimentell festgestellt werden konnte, fur die Abbildung von Netzebenen nicht unbedingt notwendig sind. Abb. 6 a, h zeigt das Ergebnis der Rechnung fiir eine Einstrahlung parallel zu der Netzebene 20T zu- sammen mit einer experimentellen Kurve, die durch Photometrierung des Elektronenbildes von_ Ni- Phthalocyanin senkrecht zu der Netzebene 201 ge- wonnen wurde. Aus den theoretischen Kurven, die fiir die Inten- sitiitsverteilung verschiedener Kristalldicken gelten und dabei gleichzeitig ein anschauliches Bild der Anderung der Intensitatsverteilung beim Fortschrei- ten durch den Kristall liefern, ersieht man, daB der Kontrast, mit dem die Netzebenen abgebildet wer- den, in dieser Naherung eine periodische Funktion der Kristalldicke ist. Weiter folgt daraus, daB die Netzebenen bzw. Potentialmaxima in diesem speziel- len Fall sich im Bild nicht an den Linien minimaler, sondern genau an denjenigen maximaler Intensitat befinden. SchlieBlich sehen wir als wesentlichstes Ergebnis der Rechnung, daB zwischen zwei Intensitatsmaxima an den Stellen der Netzebenen jeweils noch ein wei- teres schwaches Maximum auftaucht, das von der Uberlagerung von 20T und 201 herriihrt. DaB wir dieses im Experiment nicht sehen, ist kaum durch eine ungeniigende Naherung der Theorie erkliirbar, sondern durch eine ungeniigende elektronenmikro- skopische Auflosung oder Fokussierung. Herrn Prof. Dr. E. Ruska danke ich fiir die Anregung und Ermoglichung dieser Arbeit sowie fiir sein forderndes Interesse. Herrn Dr. H. Niehrs verdanke ich zahlreiche anregende Diskussionen und Hinweise. LiTERATUR 1. Menter, J. W., Proc. Roy. Soc. A 236, 119 (1956). 2. Robertson, J. M., J. Chem. Soc. 1935/1, 615. 3. Robertson, J. M. und Woodward, I., /. Chem. Soc. 1937/1,219. Investigation of High Resolution Electron Diffraction Patterns from Individual Micro-Crystals by Using a Three-Stage Electron Microscope W. D. RiECKE Fritz-Haber-Imtitiit der Max-Planck-Gesellschaft, Berlin-Dahlem Besides the visual examination of individual sub- microscopic crystals in bright and dark field micro- graphs, the observation of their diffraction patterns will be of interest. As is well known, the chemical nature or the lattice structure of the crystal can be determined under favourable experimental condi- tions. Moreover, as will be shown in this paper, diffraction experiments in the submicroscopic region may be carried out by combining micro-diffraction and high resolution diffraction. Apart from the dif- fraction pattern, shape, size, and orientation of the scattering crystal may be visually observed on the final screen. It thus became feasible to examine the fine structures at the undiffracted beam, which had hitherto escaped observation. For an interpretation of the diffraction effects, the defocused diffraction pattern may furnish valuable clues. Electron diffraction patterns of small specimen areas. — Up to now, three methods have been em- ployed to produce electron diffraction patterns of small specimen areas: (i) v. Ardenne (1) used a small stop, 6 n in diameter, which he placed just above the specimen. Thus, the area irradiated by the con- denser lens was limited to this size, (ii) v. Ardenne and co-workers (2), Hillier and Baker (4) as well as Ruska and Wolff (11) produced a considerably de- magnified image of the electron source in the plane of the specimen. They obtained illuminated areas of the order of 1 //. (iii) Boersch (3) placed an aperture of a diameter D in the plane of the first-stage image of the specimen. Thus, all the rays were screened except those emanating from an object area d = D'M in diameter. Here M is the magnifica- tion of the objective lens. The experiments to be carried out required the production of high resolution diffraction patterns of specimen areas about 1 /t in diameter. Moreover, the scattering region had to be identified precisely by visual observation of the final screen or examina- tion of the electron micrograph. Method (iii) seemed to be the best suited for this purpose. The reasons are as follows: With method (i) v. Ardenne (1) used stops of very small diameter, which contaminate rapidly. Apart from this, it is mechanically difficult to obtain well rounded bores less than 5 // in dia- meter. Ruska and Wolff (11) (method ii) employed a double condenser to obtain a demagnified image of the cross-over in the specimen plane. The angle of illumination in this plane is kept low by inserting small-diameter stops into the bore of the second con- denser lens. A diffraction resolution of 500 to 1000 can be obtained with this arrangement. Unfortu- Hii^'li Resolution Electron Diffraction Patterns from Microcrystcils 99 ELECTRON SOURCE APERTURE FOR CONDENSER LENS 1 nPERTURE FOR CONDENSER LENS 2 SPEC/MEN OBJECTIVE LENS OBJECTIVE APERTURE DIFFRftCTION PATTERN FIRST-STAGE IMAGE — OF SPECIMEN SELECTING APERTURE INTERMEDIATE LENS FIRST-STAGE IMAGE OF DIFFRACTION PATTERN SECOND -STAGE IMAGE " OF SPECIMEN. APERTURE PROJECTOR LENS SECOND -STAGE IMAGE OF DIFFRACTION PATTERN FINAL SCREEN AND PLATE BRIGHT-FIELD IMAGE O^ERFOCUSED FOCUSED UNDERFOCUSED DIFFRACTION PATTERN DIFFRACTION PATTERN DIFFRACTION PATTERN ft B C D Fig. 1. Ray paths in high resolution micro-diffraction. nately, the scattering area cannot be located with sufficient accuracy, as, in order to procure the diffraction pattern, the objective lens current is cut off. Due to this, the stray magnetic fields between the second condenser and the specimen are reduced. These had previously caused a slight deflection of the illuminating beam, and thus a displacement of the image of the electron source relative to the undis- turbed position by a distance large against its own dia- meter. Now, the deflection is equally reduced, which also reduces the displacement of the image of the source, and the identity of the scattering area with the selected field of the electron optical image is lost. This effect may only be overcome by a tedious and elaborate adjustment of the microscope, the success of which is not guaranteed. The arrange- ments of V. Ardenne and co-workers (2) and of Hillier and Baker (4) only permit shadow-microscopic ob- servation of the specimen and moderate diffraction resolution (R ^ 150). Experimental arrani^'cnients and teclniic/iie. — We have used an electron microscope type "Elmiskop 1" of Siemens & Halske AG (11). Objective lens and intermediate lens are operated independently. The diameters of the apertures selecting the specimen field, which are situated just above the intermediate lens, were chosen as to permit the examination of areas 2.2 and 0.9 /< in diameter. A diagram of ray paths is given in fig. I. The precision of the repro- duction of the camera length (57.5 cm) in routine operation has been discussed (10). For high resolution diftYaction work, a small angle of illumination is indispensable. This was realized by using the first condenser lens to obtain a 50 times demagnified image of the electron source (fig. 1, B, C, D). The lens was fed by a 220 V battery, as its feeding by the microscope power supply is not pro- vided when the intermediate lens is operated inde- pendently. This was feasible, because the require- ments regarding current stability are not stringent for demagnitication work. With the second ctinden- ser lens operating at a very long focal length, the angle of illumination is 5 10" rad, giving a dif- fraction resolution of 8350 at 80 kV. To obtain elec- tron optical images on the final screen, which are sufficiently bright for visual observation, the angle of illumination was enlarged by increasing the current of the second condenser lens (fig. 1, A). Axial astigmatism of the intermediate lens is detrimental to the diffraction resolution (10). When 100 W. D. RIECKE e G ft Fig. 2. Fine structures in tlie undift'racted beam in the diffrac- tion pattern from individual MgO crystals, (b) Pattern from the crystal shown in (a), and (d) that of the crystal (c). (H.T. = 80 kV.) this lens was used to form an image of the diflFraction pattern, existing in the back focal plane of the objective lens, within the object plane of the projector lens, it had a focal length of 4.8 cm and an astigmatic difference of focal lengths of 0.02 cm. This limits the diffraction resolution to /? 1 120 d, which shall be discussed in detail elsewhere, d is the diameter of the selected specimen area in /(. As to the resolution of fine structure in a reflection, the first-stage image of the scattering crystal practically acts as the aperture for the intermediate lens. The other rays, which pass the selecting aperture, do not contribute to this reflection. Even for a crystal size of I /*, an adequate resolution R ^ 1100 may be expected. Under these conditions, astigmatism and spherical aberration of the objective lens have no detrimental influence on the diffraction resolution. Apart from bright-held images or dark-field images taken with definite reflections, defocused diffraction patterns are useful for the interpretation of the focused ones (fig. 1, B, D). The overfocused dif- fraction pattern may be considered as a pin-hole projection of the second-stage image of the specimen onto the final screen. The "pin-holes" are formed by the reflections in the second stage image of the diffraction pattern (fig. 1 , B)_ At this, the undiffracted beam produces a "bright-field" shadow image, and each reflection a corresponding "dark-field" shadow image. In a similar way, the underfocused diffrac- tion pattern may be regarded as a point-projection image of the second-stage image of the specimen. The projecting rays emanate from the reflections of the first-stage image of the diffraction pattern (fig. I, D). The shadow images in the underfocused diffraction pattern are correctly orientated to the focused one, apart from a slight rotation corres- Fig. 3. Focused (a) and overfocused (b) diffraction pattern as well as micrograph (c) of an individual MgO crystal. (H.T. = 80 kV.) ponding to the defocusing of the intermediate lens. Those of the overfocused pattern are rotated with respect to the focused one by 1 80\ Experimental results. — We have examined the dif- fraction patterns of MgO and ZnO crystals. The specimens were prepared by burning magnesium or zinc ribbon and exposing platinum specimen carriers to the smoke. Although the holes of the carriers were not covered by the usual collodion film, in order to avoid additional scattering, a great number of perfectly grown crystals were found to adhere to the rim of the holes. First, we looked for those types of fine structures that are predicted by the dynamical theory of elec- tron diffraction. In the diffraction patterns, that had been published hitherto, the fine structures in the undiftYacted beam were masked by the well-known extensive spot of intense halation, which is caused by the superposition of the scattered intensities of a great number of crystals. This effect was eliminated with our method by producing diflVaction patterns of only one crystal. In the undiffracted beam of the patterns of individual MgO crystals we have ob- tained fine structures., which are caused by interference double refraction, provided that strong Bragg reflec- tions occur (fig. 2, b). This has been predicted by Moliere and Niehrs (7). At longer exposure times, even more weak spots are found on straight lines drawn through the doublets corresponding to each crystal wedge. A great number of spots is obtained with relatively large crystals some 1000 A in size (fig. 2, d). By using the selecting aperture to screen the first-stage image of the crystal, with the exception of a single wedge-shaped part at an edge of the MgO cube, the figure is reduced to (i) the double- refraction doublet, and (ii) some weak spots on the High Resolution Electron Diffnulion Patterns from Mierocrystals 101 Fig. 4. Fine structures in the dilTraction pattern from indi- vidual ZnO crystals, (b) Undeflected beam in the difl'raction pattern of the crystal (a), (c) Central beam with fine struc- tures, and (d) reflection with subsidiary maxima from other patterns. (.H.T. = 80 kV.) Straight line drawn through the doublet. These spots are deviated to larger angles than the doublet itself. Their intensity is two to three orders of magnitude lower than the intensity of the doublet. As the inter- ference figure observed is frequently not symmetric to the center of the diagram, the effect cannot be interpreted as an extension of the reciprocal lattice point (5) or as Fraunhofer diffraction. In both cases, the crystal size obtained from the micrographs would result in spots of much narrower spacing than that found in the diffraction pattern. In our opinion, the weak spots may be caused by beams belonging to further wave fields disregarded in interference double refraction. Some fine structures in (hkO) reflections from MgO crystals are most probably subsidiary maxima, caused by the extensions of the scattering amplitude around each reciprocal lattice point, which arc nor- mal to the boundary faces of the crystal. The fine structure shown in fig. 5, a, corresponds to a crystal 500 A in cube length. Unfortunately the spherical aberration of the objective lens introduces an uncer- tainty of about 0.1 /< into the exact location of the scattering area when the beams are deflected about larger Bragg angles. Therefore, diffraction patterns of individual crystals of this size cannot be obtained. Nevertheless, the defocused diffraction pattern per- mits an estimation of the upper limit of the crystal size, which is O.I //. The above supposition is thus confirmed indirectly. On defocusing a diffraction pattern, darkfield shadow-images of the scattering crystal regions arc obtained from the strong refiections. In the case of a wedge-shaped part of a crystal, two waves and correspondingly two rays are present for each rellec- tion. Therefore, in the overfocused pattern, two Fig. 5. (a) Subsidiary maxima and (b), (c), (d) ne\s t>pcs of fine structures in diffraction patterns from MgO crystals. In (c) and (d) each spot of a quartet is split up into two spots. close lying shadow images of the wedge are ob- tained for each strong reflection (fig. 3). This may be regarded as a "double" projection of the second- stage image of the wedge, produced from the spots of the corresponding double-refraction doublet, which appears in the second-stage image of the dif- fraction pattern. In the diffraction patterns of individualZnO crys- tals, fine structures in the undifTracted beam have also been observed (fig. 4, a, b, c). Contrary to the subsidiary maxima, which are found at Bragg rctlcc- tions (fig. 4, d) and have already been obtained (8). they are not symmetrical to the center of the pattern. For an interpretation of the effect, it is most prob- able that the dynamical theory of electron diffraction will have to be considered. Finally, some other diffraction effects were ob- served, which have not been interpreted up to now (fig. 5, b, c. d). The splitting up of each spot of a double refraction quartet into two spots occurs rather frequently, and has also been obtained by Molicre (6). References 1. VON Ardenne, M., Kolloid-Z. 108, 195 (1944). 2. VON Ardenne, M., Schiebold, E., and Gunther, F., Z. Pliysil< 119, 352 (1942). .V BorRsni, H., .)/;/;. Pliysik (5) 27, 75 (1936). 4. Hill II R. J. and IUktr. R. I-.. J. Appl. Plivs. 17, 12 (1946). 5. VON Laue, M.. Ann. Physik (5) 26. 55 (1936). 6. MoiiiRi', K., personal communication (1956). 7. MoLiERE, K. and Niehrs. H.,Z. Physik 140, 581 (1955). 8. Rees, A. L. G.andSpiNK. J. A.,/fc/fl C/-v.vr. 3, 316 (1950). 9. RiECKE, W. D., P/n.s. It'///. (2) 6. 20 (1955). 10. RinrKE,W. D. and Rlska, E., Z. h/.v.v. Mikroskop.{\951, in press). 11. Rt_SKA. E. and Wni rr. O., Z. u/.v.v. Mikroskop. 62, 465 (1956). An Electron Microscope Examination of Freshly Prepared Silver Iodide Sols R. Ottewill and R. W. Horne Dept. of Colloid Science and Cavendish Laboratory, University of Cambridge Although silver iodide, which forms a typically hydrophobic sol. has been extensively investigated in the field of colloid chemistry, very little research has been devoted to an examination of this material in the electron microscope. Of the numerous hydro- phobic sols which can be prepared the gold sols are the only ones to have received any considerable attention by this method, primarily as standards of resolution. In connexion with other physicochemical work on silver iodide sols, this material has been examined in a high resolution electron microscope. Numerous small particles of the order of 10 to 25 A have been resolved and particles of below 10 A have been detected. Since the resolution of such particles is of extreme interest in the field of electron microscopy and colloid science and probably con- stitutes the first resolution of colloidal particles ap- proaching atomicdimensions, it is thought worthwhile to submit a preliminary report of the work at this stage, although a more extensive investigation is still in progress. 100 ml of a solution of silver nitrate (1.25 mM) were added with continuous stirring to 100 ml of a solution of potassium iodide (1.38 mM). The silver iodide sol formed under these conditions is a nega- tively charged sol, stabilized by the adsorption of I~ ions. The sol was freed from potassium nitrate formed during the reaction and from excess potassium iodide by continuous electrodialysis against distilled water. During this process the K^ gegenions are removed to give a sol which may be represented as (Agl) I H^ (1). The sol obtained after electrodialysis was diluted with an equal volume of distilled water to give a concentration of ca. 0.3 mmoles of silver iodide, and used at this strength for the preparation of the electron microscope grids. Silver iodide sols are reasonably stable to daylight, but specimens were always stored in the dark, as a precautionary measure. Distilled water was always treated before use with an ion-exchange column to remove any remaining ions, and both reagents were recrystallised from this water. All glass ware was steamed for half an hour before use to remove any adhering grease film. The sol was freshly prepared in order that aging phenomena, a process which includes possible chan- ges in the form of the particles and changes in the size of particles owing to recrystallisation or slow flocculation, should not become important. How- ever, only a small change in the mean diameter of a silver iodide sol after aging is found (2). An examination of silver and silver iodide sols in the electron microscope at instrumental magnifica- tions of 40,000 and 80,000 revealed the presence of a large range of particle sizes from ca. 180 A down to particles which were only just resolved by the microscope in the range of 6-10 A. Only such particles were measured which were clearly resolved from the background structure and capable of being produced on separate electron micrographs. A preliminary analysis of diameters of 300 particles from a 400,000 times enlargement gave a distribu- tion curve giving a mean particle diameter of 25 r; 5 A. The fact that the solubility of silver iodide is very small, 5 10"' mmoles litre compared with 0.3 mmoles litre present in the sol form, suggests that the small particles are colloidal in nature and are not formed during the drying down process by crystallisation of the silver iodide in molecular solu- tion. The shape of the silver and silver iodide particles in the electron micrographs appear to take many different forms. In the present work particles having a cubic face of side of 25 A have been clearly resolved, also hexagonal particles are visible in the 50 A region, whilst the larger particles (125 A upwards) appear to be either approximately spherical or have irregular shapes. The formation of a colloidal particle is almost certainly that of rapid growth from a nucleus, the latter consisting of only a few atoms, and it is possible Fig. 1. Electron micrograph of silver iodide sol. Instrumental magnification 80.000. Final magnification 500,000. Examination of Freshly Prepared Silver Iodide Sols 103 ^^^^^^^^^^^^r^^ -A ' ■" 1M W^M ^^^K^^^ m ^^B ^^^^E ' % If l^l ^^■^K'f ^^^^H ' * •^^^^1 ^^^^' « * ^ ^^^1 ^Kki^m * » ^[^1 ^^^^^^^ 1 v^H ^^^^^■| . ^^^H ^^^^^^^^B ** t- « «^^^B ~v.- • • f i«- ft #fH * ft /OOO 4 r n .^ Fig. 2. Electron micrograph ofsilversol. Inslrumenlal magni- fication 40,000. Final magnification 400,000. that the state observed here is that of a large number of small nuclei which may eventually grow to larger particles. It is of interest to determine whether such particles disappear during the aging process or to some extent remain in "equilibrium" with the larger particles. The irregularity of many of the larger particles may to some extent be attributed to the angle at which the particle is resting on the supporting film and, or alternatively, to irregular growth during formation. A serious limiting factor in determining the size and shape of particles below 10 A is the presence of background eflfects due to supporting films and also optical effects caused by focusing. The relationship between the true crystal structure and the form of particles observed in the electron microscope is not easy to interpret because of the "shadowing" eff"ect which occurs. However, it is of interest that J. J. Trillat and A. Laloeuf (3), in the electron diffraction study of silver iodide smokes, found that the particles present existed in both hexagonal and cubic crystal forms. The unit cell of the hexagonal form had the dimensions, r 7.94 A. a = 4.58 A, and the unit cell of the cubic form had a = 6.47 A. It thus appears possible that colloidal silver iodide exists in both crystal forms, a fact which is being further investigated by micro-diffraction techniques where it is possible to select areas down to 5 /< diameter. The presence of very small particles in the sol shows clearly the fact that colloidal particles may exist in size almost down to the range of atomic dimensions, and indicates that there is no strict boundary between atomic and colU)idal particles. The distribution of particles of very small size also appears to extend much further than had been rec- ognised hitherto using electron microscopes with a resolution of the order of 30 A. Comparison of silver with silver iodide sols strongly suggests that the dense scattering material is probably silver. The mean diameter of the particles in a colloidal solution is an important factor in colloid chemistry, and its accurate evaluation essential for correlation between experiment and theory. The presence of such a magnitude of small particles may mean that hitherto, where the electron microscope has been used for the evaluation of size, the mean value has been too high because the smaller particles have been unresolved. The theory of Verwey and Over- beek (4) predicts that small particles of diameter ca. 20 A should have a very low stability, since they would require very high electric potentials (H'*,,), and have low absolute values of repulsive poten- tial energy. The latter is directly proportional to particle radius. The fact that particles of this size have been shown to exist in such profusion in the present work does indicate, however, that the prepara- tion of very small colloid particles in a stable form is by no means impossible. References 1. DE Bruyn, H. and Troklsira, S. A., Kolloiil-Z. 84 192, (1938). 2. Harmsen, G. J., VAN ScHOOTEN, J., and Overbeek, J. In. C, J. Colloid Sci. 8, 64 (1953). 3. Trillat, J. J. and Laloeuf, A., /. Chim. P/nw. 46, 168 (1949). 4. Verwey, E. J. W. and Onerhlik, J. Th. G., Thoor> of the Stability of Lyophobic Colloids. Elsevier, 1948, 170-178. V SPECIMEN PREPARATION TECHNIQUES IN BIOLOGY AND MEDICINE Problems of Osmium Fixation G. F. Bahr, G. Bloom and U. Friberg Institute for Cell Research and Genetics, and Department of Histology, Karolinska Institutet, Stockholm During fixation and subsequent treatment in va- rious fluids great changes in volume of tissue speci- mens may occur. To our knowledge no measure- ments have been reported on the changes of volume during osmium fixation and following embedding according to the general technique applied for elec- tron microscopical specimens. We have therefore considered it justified to present the results of quan- titative measurements on the changes in volume, weight, and specific weight during this procedure. The experimental method was based on the prin- ciple of Archimedes. For the weighing arrangement an analytical balance with a sensitivity of 0.1 mg was used. Mostly small pieces of guinea pig"s liver weighing 150-200 mg served as test objects. Other tissues were used for the sake of comparison. A measuring group was comprised of at least 5 speci- mens. Fixation, dehydration, and methacrylate infil- tration were carried out at 0 or 24 C, whereas pa- raffin infiltration was performed at 58'"C. Fig. I shows the changes which occur during fixa- tion in 1 % osmium tetroxide in isotonic Tyrode's solution at pH 7.2 over a period of up to 24 hours. A marked and rapid swelling is seen which reaches a maximum of 30 % after 4 hours. Half of the final value is reached after only 15 minutes in the fixative. The weight curve follows closely the volume curve and therefore only the volume curves will be referred to. The specific weight rises during fixation as an expression of the binding of osmium in the tissues. Changes during the continued preparation are disclosed in fig. 1. The swelling that takes place in the fixative is now transformed — during dehydra- tion— into a shrinkage that almost brings the volume back to the original value. During paraffin infiltra- tion further shrinkage is noted, especially in speci- mens that were fixed at 0 C. Methacrylate, on the other hand, causes only a small shrinkage during infiltration, and can in this respect be regarded as a more favourable medium. During polymerization, however, methacrylate reduces its volume by 20 % — just as does paraffin during solidification — and it seems that this shrinkage is more or less completely transferred to the specimens, as far as we can judge from planimetrical observations on embedded liver pieces. Fig. I also shows that for methacrylate treated specimens the fixation temperature is of httle influence on volume. The specific weight that rises during the fixation falls during dehydration and finally rises again during infiltration. In another experiment a comparison was made between various tissues (fig. 2). Certain differences in intermediate and final values are to be noted. The principal pattern is, however, the same for all tissues. The swelling seen during osmium fixation, is in no way unique for this fixative. In a series of formaldehyde concentrations, ranging from 0.5 to 16 % there is a swelling in all solutions, the magni- tude of which is related in inverse proportion to the concentration of the fixative. Maximal swelling is 140 12. 48 24 S6 ETHiNOL I OSMIUM FIXATION I I I EMBEOOINO TYROOE HOURS I 12 24 36 IethanolI • 10 CO 5 o > (/) X 3) m 10 60 n HOURS Fig. 1. Volume changes of pieces of tissues during osmium fixation and embedding. I -OSMIUM FIXATIONjI^^r MET>1ACRYLATt | 2 fTrRooSl Fig. 2. Volume changes of pieces of various tissues during osmium fixation and embedding. Lipids and Osmium Fixation 107 I UJ ^ 51.0 o UJ a. l4S 120 "'lOO -C— ^ J PiRAFFlN r-' ! ^ ■ L MeTHACHYLATE UJ 2 D _|90 0 > 24° C \ MtTMACHTUlTE A A VOLUME \ 60 _. _ SPECIFIC • EIGHT \ PARJtFFIN 40 (/> s 10 rn f- r; ^ r 0 •- m i> JO 10 I 4X P0B»1»I-0EMY0C I JIBS ETH«NOL ■I KREBS-IVMCR eo I EUBEODINO HOURS Fig. 3. Volume changes and variations in specific weight of pieces of tissues during formol fixation and embedding. 120 o 100 > 80 1% OSMIUM TET/IOXIDC IN TYRODE WITH ADDITION OF DEXTSAN (Ox) (AND SUCROSE) 0% 3% Dx Dx Dx 12 % Dx 3 % Dx + _ 0.2 M sucRoa 3 % Dx ♦ OSM SUCROSE 6 % Dx ♦ 0^5M SUCROSE o°c 1 24 HOURS to — > I z > (D Fig. 4. Balancing of the volume changes of pieces of tissues during osmium fixation when adding dextran and sucrose to the osmium tetroxide solution. reached in 3 hours. Thereafter a shrinkage can be observed, the speed of which is not related to the concentration. If formalin fixed specimens are followed further in the preparation procedures it is seen (fig. 3) that they react in a manner very similar to that described for osmium fixed material. Fixation in 4 "o formal- dehyde gives an initial swelling of 18 % after 12 hours. Dehydration and paraffin infiltration reduce the volume with 30 "o each. In methacrylate again, the shrinkage is small. The changes in specific weight are unimportant. The results obtained indicate that the swelling in the fixatives is due not only to the fixing agents themselves but in a high degree to the aqueous media in which they are dissolved. Liver pieces were suspended in Tyrode's solution of various concentra- tions at 0 C. A swelling is noted in all, even those strongly hypertonic. As expected the swelling is most pronounced in the diluted solutions and in these, maximal swelling is also reached in shorter time. Similar results are obtained with Krebs-Ringer media and also with sucrose solutions. The addition of small amounts of gelatin to physiological fluids has been recommended in order to reduce their toxical effects on cells. The addition of 0.25 °(, gelatin to Tyrode's solution reduces the swelling significantly with about 20%. Fig. 4 shows that a still more pronounced effect is noted with dextran. It is possible to use dextran in much higher concentrations than is possible with gelatin without increasing the viscosity to an im- practical point. It is also seen that by combining dextran with sucrose it is possible to completely prevent the swelling. The next question of great interest is if it is possible to reduce the shrinkage during dehydration to some extent. Dehydration in rising concentrations of ethanol brings about a more gradual shrinkage than absolute ethanol, but the end result is but little better. In the literature various substitutes for ethanol as dehydrating agent have been proposed. From our results it is obvious that only methanol appears to give better results than those obtained with ethanol. The Quantitative Assay of Lipids Extracted from Untreated and Os04-fixed Beef Brain G. F. Bahr Institute for Cell Research and Genetics, Karolinska Jnstitutet, Stockholm Although we know today where osmium cannot be expected to be found in a thin section, we really know little about the actual places of osmium reac- tion and deposition (1, 2). There has been much dis- cussion about to what extent osmium is bound by the different constituents of a tissue. Now the main interest is focussed on the question whether the lipids or the proteins are demonstrated by osmium deposits. Our recent analyses show that osmium fixation renders a fraction of about 2 "„ of the total dry weight unextractable in the \ovm of lipids (fig. I). Skin 108 I. R. GIBBONS AND J. R. G. BRADFIELD BRAIN-TISSUE EXTRACTABILITY D Ffesh motenol ■ Os(X - filed moleriol obviously makes an exception in this series, because extractability is considerably increased. An explana- tion would be that here the fixation has opened up the tissue and rendered the lipids accessible for extraction. The 2 "^'o lipids bound by the osmium fixation must consist of unsaturated compounds. In rats fed with a diet containing 5 °o fat, about 70 °o of the tissue fatty acids are unsaturated (3). Mammalian fat has as a rule an iodine number of 65 corresponding to a total capacity to bind 0.47 g Os per gram fat. With regards to the fact that only 70 % of such fat is actually unsaturated, we calculate the osmium up- take by the unsaturated fat fraction to 0.63 g osmium per gram. Ten per cent fat can ordinarily be extracted from tissues, which means that 100 g tissue dry weight has a maximal capacity to bind 4.7 g osmium by its lipids. Our analysis showed that maximal uptake values for liver tissue were about 15 g osmium per 100 g tissue dry weight. Thus the uptake exceeds the theoretical capacity of the tissue lipids by |. In other words, a considerable part of the osmium is bound by substances other than lipids. Direct analysis shows that but little of the binding capacity of the lipids is actually used. Instead of theoretically possible 4.7 g only 1.3 g are taken up by the lipids in 100 g tissue dry weight. That is 8.4% of the total osmium uptake, (fig. 2.) In consequence neither proteins nor lipids are significantly preferred by osmium. Upon a question by Dr. Sjostrand if there might possibly be a preference of certain compounds in the lipid fraction itself, an analysis of fixed and unfixed brain lipids was carried out. (fig. 3.) Technical de- tails of the rather complicated analytical procedure cannot be discussed here and will be published else- where. To demonstrate preferences of the osmium tetroxide fixation towards certain lipids an amount of osmium tetroxide was added corresponding to about 30 % of the theoretical capacity of uptake. Six per cent lipids have been rendered unextractable but again no real significant preference for one type of lipid could be observed. It may thus be said that osmium tetroxide fixation stains reactive compounds in tissues in even pro- portions, without giving significant preference to certain substances. Unfortunately more cannot be said from this study but that the well-known pattern of repeating dense and light bands in thin sections corresponds to accumulations of reacting and un- reacting groups. References 1. Bahr, G. F., Exptl. Cell Research 7, 457 (1954). 2. — ibid. 9, 277 (1955). 3. Deuel, H. J., The Lipids. N.Y. Interscience Publ. (1951). The Fixation of Nuclei in Locust Testis I. R. Gibbons and J. R. G. Bradfield Cavendish Laboratory, Cambridge Previous studies on the problem of fixation, both with light and electron microscopes, have demon- strated the excellence of buffered osmium tetroxide in the preservation of cytoplasmic ultrastructure (4, 5). This work has been confirmed by the highly organ- ized and highly reproducible ultrastructures observed with the electron microscope in the cytoplasm of cells after osmium fixation. However, with regard to the nucleus the situation is quite different. Little apparently organized ultrastructure has been observed in osmium-fixed nuclei either interkinetic or dividing. In view of this fact we have considered it desirable to re-examine the question of the fixation of nuclei. The chromatin distribution within live nuclei of locust primary spermatocytes has been investigated by observing squashes of testis follicles in Belar solution with an ultra-violet microscope. Nuclei in meiotic prophase cells appear to have their chroma- tin arranged in intertwined strands, the thickness of these strands is variable and increases as the cell approaches diakinesis. When, however, osmium fixed nuclei of this material are examined in electron micrographs of thin sections they usually appear to have a rather homogeneous fine-grained appearance, and show little, if any, trace of the inhomogeneity of chromatin existing in live nuclei. In order to re- Fixation of Nuclei in Locust Testis 109 m 'P- ^ If •4iJ*.V Figs. 1-2. Electron micrograph (2) of a thin section of an osmium fixed nucleus in locust testis with an inset of an ultra-violet micrograph (1) of the same nucleus in an adja- cent thicker section. Two clusters of mitochondria (w) are indicated. They possess considerable ultra-violet absorption. solve this discrepancy, we have examined adjacent thick. (| /<) and thin (about 200 A) sections of osmium fixed testis in ultra-violet and electron microscopes respectively. Low magnification micrographs permit the identification of identical nuclei in the two sec- tions. A comparison between two sections of the same nucleus is shown in figs. 1-2. Fig. I is an ultra- violet micrograph of the thick section and shows that the inhomogeneous distribution of chromatin in the live nucleus is at least partly preserved by osmium fixation. Fig. 2 is an electron micrograph c'f the ihin sectiiMi of the same nucleus, it shows the homogenecHis appearance typical of osmium fixed nuclei in this material and little trace of the inhomo- geneities of chromatin existing in the section is dis- cernible. This homogeneous appearance of nuclei in electron micrographs must, therefore, be ascribed not to the failure of osmium fixation to preserve chromatin distribution, but to its failure to produce appreciable contrast between chromatin and nuclear sap. We believe that this lack of contrast is evidence that osmium tetroxide does nt)l react with desoxyri- bonucleic acid (2). The general appearance of nuclei after osmium fixation is shown in fig. 2. The nucleus shows a rather homogeneous fine-grained structure with little trace of organisation. When examined at high magni- fication occasional traces of organised structure are seen appearing as groups. The general appearance of nuclei after formalde- hyde fixation is shown in fig. 3. The distribution of chromatin within the nucleus is easily visible because the relative contrast between chromatin and nuclear sap is higher than in osmium fixed nuclei. The pairing of the chromosomes during prophase is indicated but the chromosomes still show no apparently highly organised structure. As regards contracted chromo- somes, the contrast between chromatin and cyto- plasm is higher than in osmium lixed material but there is a similar apparent lack of organisation. The chromosomes may appear to possess a split down the centre but the significance of this is hard to interpret because of the difficulty of identifying the stage of meiosis from a thin section. One of the classic fixatives used in light micro- scope studies of chromosomes is 45 ",, acetic acid and we have considered it worth while to examine material fixed in this way in the electron micro- scope. In sections which are not too thin it is possible Fig. 3. Electron micrograph of locust testis fixed in 5 % for- maldehyde. Most of the field is occupied by a primary sper- matocyte nucleus in which the pairing of the chromatids is indicated by the arrow. Fig. 4. Electron micrograph of locust testis fixed in 45 °o acetic acid. The pairing of the chromatids is indicated in transverse section (M) and longitudinal section (N;. 110 I. R. GIBBONS AND J. R. G. BRADFIELD Fig. 5. Electron micrograph of a longitudinal section of a locust immature sperm head fixed in 1 % osmium tetroxide. to distinguish such gross characteristics as pairing of chromatids (fig. 4) both in transverse section (M) and longitudinal section (N). When thinner sections are used in order to obtain higher resolution the poor preservation makes interpretation almost impossible. Sperm and spermatid heads of the locust have been shown by birefringence studies and ultra-violet dichroism studies (1) to contain nucleic acid mole- cules orientated along the axis of the head. It is, therefore, interesting to examine this material in the electron microscope after various fixatives in order to obtain some idea of the value of these fixatives in preserving chromatin fine structure. While we have not as yet been able to demonstrate structure in mature sperm heads, we have observed well defined and characteristic structure in the heads of sperma- tids. This structure in the most highly developed form we have been able to observe is shown in longitudinal section in fig. 5 where it appears as a large number of parallel lines of thickness about 70 A which are orientated along the axis of the head. The corresponding transverse section — which is from the same cyst in the follicle and so must represent the same stage of development — is shown in fig. 6 to be a large number (about 270 in this case) of tightly packed polygons. We interpret these sections to mean that the head contains a large number of parallel tubes orientated along its axis (3). At this stage of development the head is shrinking rapidly and at later stages this structure seems to become too tightly packed to be easily visible and the head appears structureless. At earlier stages of develop- ment we have observed that material first seems to aggregate into sheets and that these sheets then appear to wrap around each other to form the tubed structure described above. Fig. 6. Electron micrograph of a transverse section of a locust immature sperm head in the same cyst in the follicle as that shown in fig. 5 and hence at the same stage of de- velopment. We have observed this structure after fixation in 1 "o buff'ered osmium tetroxide and after 5 % buffered formaldehyde there being little obvious difference in the quality of preservation. We have also been able to observe the structure, though less well- preserved, after fixation in 45 °o acetic acid — a classic but brutal chromatin fixative. It must be admitted that the results of the appli- cation of electron microscopy to the study of nuclei have been somewhat disappointing. In particular no further insight has been gained into the processes of mitosis and meiosis which at the level of the light microscope appear so mysterious and dramatic. It is not yet possible to say whether this is due to poor preservation by the existing fixatives, or to the presence of a type of organisation very hard to analyse in thin sections or to an actual lack of orga- nisation within the region of size that can be observed with this technique. Our opinion is that a combina- tion of the first two reasons given above is perhaps the most likely explanation. In this connection we feel it is pertinent to point out the possible use of high magnification stereo-photographs of fairly thin sections (from which removal of embedding material is not necessary) in elucidating the complex fine structure of chromatin. References 1. Caspersson, T., Chromosoma 1, 605 (1940). 2. Gibbons, I. R. and Bradfield, J. R. G., Biochhu. et Bio- pliys. Acta (1957, in press). 3. — /. Biophys. Biochem. Cytol. (1957, in press). 4. Palade, G. E., J. E.xptl. Med. 95, 285 (1952). 5. Porter, K. R. and Kallmann, F., Exptl. Cell Research 4. 127 (1953). Ultra-thin Sections of Avian Tubercle Bacilli in a New Embedding Medium A.M.Glaueri' and E.M.Brihger Slrangcways Rcscarcli Lahuiatoiy (iiul Papworlli llospilul. Camhriclgc The reproduction of the tubercle bacillus presents a particularly interesting subject for study. The mycobacteria may be considered to occupy an intermediate position between the ordinary bacteria, which reproduce by binary tission, and the Nocardia and fungus-like organisms that undergo a more complicated life-cycle involving budding, branching and mycelial development. We have previously studied the development of the avian tubercle bacillus in the phase contrast microscope (2) and were able to show that it can reproduce in more than one way. In some strains the rods of the original inoculum elongate and then divide directly by binary fission; these organisms behave like the ordinary bacteria and such strains have been described as "bacillary" strains. In other strains the original rods grow and branch, without immediate division, to form complex mycelial struc- tures. These mycelia finally break down by a process of simultaneous multiple division to produce a mass of rods similar in appearance to those of the original inoculum. These two types of development of avian tubercle bacilli have also been investigated in preparations of intact organisms in the electron microscope (I). Only a limited amount of information can be obtained from such a study owing to the thickness of the bacilli, but it was obvious that a marked difference in internal structure underlay the differen- ces in the reproductive processes. At the same time an unexpected development of round bodies or "intracellular units'' was observed within the elon- gated rods of the bacillary strains. These bodies have since been examined more closely by the technique of ultra-thin sectioning and the accompanying cyto- plasmic changes have been observed. As other workers with bacteria have also observed, we found that the organisms were frequently dis- torted during the familiar preparative procedures for ultra-thin sectioning. These observations led us to consider the possi- bility of using an alternative embedding medium. Maaloe and Birch-Andersen (5) have already had considerable success with a resin of the epoxy type and we have been experimenting with resins of the same series.- The ones we have been investigating are marketed under the trade name of Araldite. 1 Sir Halley Stewart Research Fellow. - These resins were developed by Messrs. Ciba Ltd. ol Basel and are made in England by Aero Research Ltd. of Duxford, Cambridge. A Standard liquid epoxy resin is used and is made more plastic by the addition of dibutyl phthalate. On the addition of a suitable hardener the resin sets uniformly, without shrinkage, to form a clear, light-gold block. Originally we used an aliphatic polyamine as the hardener, this being the usual type of cold-setting hardener for these resins. Unfortu- nately, as was discovered by Maaloe and Birch- Andersen (5), a mixture of the resin with this hard- ener is difficult to handle and is not ideal for electron microscopy. The mixture is very viscous so that im- pregnation is difficult and also it is not readily soluble in absolute alcohol. On the advice of Dr. Glauert, of Aero Research Ltd., we have experimented with a different hardener with promising results. If the casting resin is mixed with equal quantities of this new hardener, which is a liquid anhydride, the resultant mixture is sufficiently fluid to be handled easily and is readily soluble in absolute alcohol. The hardening process takes a considerable time at normal incubation temperatures, but it can be speeded up as much as required by the addition of an amine accelerator. After various trials we have found the following mixture to be suitable as an embedding medium: — ^^ Araldite'' for iiltra-thiii sections Casting resin M 10.0 ml Hardener 964 B 10.0 ml Dibiit\ I phthalate 1.0 ml Accelerator 964 C 0.4 ml Slight variations of this formula ma\ be found con- venient for difTcrcnt types of specimen. For ease of handling the resin is mixed and the specimens are soaked at 48 C. The fixation o\' the specimens in buffered osmic acid and dehydration in graded alcohols is the same as for methacrylatc embedding. From absolute alcohol the specimens are passed to a 50 50 mixture of alcohol and Aral- dite at 48 C for 1-2 hours, then to two changes of the Araldite mixture at 48 C for 2-3 hours and finally into gelatin capsules with fresh Araldite. The blocks harden in about 30 hours at 48 C. It has been found advisable to incubate the specimens at 48°C, because at other temperatures there is a tendency for soft specimens to rise in the cai-»siiic during the setting process. The resultant blocks have a similar hardness to methacrylatc and thin sections have been cut with ease. Parallel specimens were embedded in metha- crylatc so that a direct comparison could be made. 112 A. M. GLAUERT AND E. M. BRIEGER Fig. 1. Ultra-tliin section of avian tubercle bacilli embedded in methacryiate. The organisms are distorted. Note the complex nuclear apparatus and large cytoplasmic granules. 110,000. If the original inoculum is viewed directly, without sectioning, it is seen to consist mainly of short rods with electron-dense polar bodies and a transparent, structureless cytoplasm. After 24 hours' growth these rods elongate and round bodies are clearly visible within them. These bodies have distinct membranes and it was suggested that they might correspond to the segmentation spores of the Nocardia. Methacryiate. — Sections of the small, transparent rods of the inoculum show that they have large central light areas containing threads and granules, the "nuclear apparatus". The cytoplasm, which occupies a narrow peripheral zone, has a granular structure, the granules ranging in size from 100 to 200 A. There is some evidence of the presence of fine threads and membranes within the cytoplasm but the cells are not sufficiently well preserved for these to be accepted with any certainty. The cells are obviously swollen and distorted and nothing definite can be deduced concerning the nature of the nuclear material. After 24 hours" incubation the rods of the inocu- lum have elongated, the dense threads and granules in the nuclear region have more complex configura- tions and the cytoplasm is denser (fig. I). The cyto- plasm still has a granular structure and in some organisms there is also a scattering of larger granules of 200 to 400 A diameter. A few of the organisms are in process of dividing by binary fission to form two daughter cells, each with its own nuclear appa- ratus. At later stages in the development of the bacilli a very different pattern of internal structure is seen. Rows of round bodies are observed forming inside the bacilli. These units have a finely granular struc- ture of uniform density and are enclosed within a definite limiting membrane. Their formation is accompanied by an increase in the number and size of the larger cytoplasmic granules. The bodies do not appear to contain an organised nucleus and in the early stages of their development the nuclear appara- tus of the bacillus is still visible outside them. We inter- preted these bodies as spores (3), although the later stages of their reproduction will have to be followed before their nature can be clearly established. In structure and development they bear a remarkable similarity to the spores of the ordinary bacteria described by Chapman (4). Aralciite. — The study of similar organisms embed- ded in Araldite is only in its early stages but certain differences of structure have already been observed. As was stated earlier, most of the bacilli appear to be far better preserved. The cell wall is no longer separated from the cytoplasm and there is clear evidence of a cytoplasmic membrane underlying the cell wall (fig. 2). This membrane has proved to be particularly elusive in studies of methacryiate sec- tions. The fine limiting membranes of the developing spores are also clearly seen and appear to be double (fig. 3). There seems to be no doubt that these fine membranes are seriously distorted during methacry- iate embedding. The Use of Gelatin for Eniheddini: Bioloi^ical Objects 13 Fig. 2. Ultra-thin section of avian tubercle bacillus embedded in Araldite. The organism appears smooth and there is a fine cytoplasmic membrane underlying the cell wall. 60,000. Fig. 3. Ultra-thin section of avian tubercle bacillus embedded in Araldite. A spore is developing in the centre of the bacillus and is enclosed in a fine double membrane. 80,000. The so-called vacuole is usually filled with a fine network and the central dense structure has a far more luiiform appearance. Instead of the complex mass of threads and granules that are observed with methacrylate we find a smooth thread-like structure with associated dense granules. These results with Araldite are only preliminary but seem to us to be promising. Some similar tests have been made with mammalian tissues and there are indications that Araldite will be useful in the preparation of hard tissues, such as adult hairs. We would like to thank Dr. R. II. Cilaucrt. of Aero Research Ltd., for his cooperation throughout the course of this work. Some of the electron micrographs were taken in the Cavendish Laboratory, Cambridge, and we would like to thank Dr. V. E. Cosslett and Mr. R. W. Horne for providing electron microscope facilities. References 1. Brifger, E. M., Cossiftt, V. E., and Glaltrt, A. M., J. Gen. Microbiol. 10, 294 (1954). 2. Brifger, E. M. and Glauert, A. M., /. Gen. Microbiol. 7, 287 (1952). 3. — Naliire, 178, 544 (1956). 4. Chapman, G. B., y. Bacteriol. 71, 348 (1956). 5. Maaloe, O. and BiRf h-Andi rsfn. A., Vlth Symp. Soc. Gen. Microbiol. "Bacterial Anatomy", p. 261. (1956). The Use of Gelatin for Embedding Biological Objects in Preparation of Ultrathin Sections for Electron Microscopy V. P. GiLEV Lab. of Electron Microscopy, Acad, of Sciences of t lie USSR, Moscow 1 HE majority of investigators working in the field of electron microscopic cytology and histology use the method of embedding biological objects in methacrylates (3). It was proposed to embed biological objects in polyethylenglycols of high molecular weight which are soluble in water, (1,2, 5). But due to the fact that preparation of ultrathin sections of the objects em- bedded in these substances was difficult, this method did not become wide-spread in electron microscopy. The method we oflFer is based on the use of gelatin as embedding medium. It makes it possible to com- pletely exclude treatment of objects with organic solvents and to obtain sections up to 0.03-0.04 // thick and, perhaps, thinner. A 10 "o water solution of food gelatin is boiled for several minutes together with beaten-up hen egg al- bumen and activated charcoal. Per 300 cm-' of gelatin solution we take the albumen of one hen egg and 3 g of activated charcoal. Then the solution, while it is warm, is filtered first through paper and then through an asbestos bacterial filter. One portion of solution is evaporated in a thermostat at a 8 — 568204 Electron Microscopy temperature of 45 C until there remains half of its content, and the other one until there remain three thirds of the initial volume; in this way we obtain 20 "„ and 40 "„ solutions which arc used for impreg- nation and embedding. We may prepare 20 ",, and 40 °o solutions of refined gelatin in Ringer solution. The solutions should be absolutely transparent. To prevent from rotting some thymol is added. The objects (striated muscle tissue of axolotl, Atnblystoina punctatum) were fixed by osmium tetr- oxide (4) at pH 7.4-7.5 during 20 hours at a temperature of TC. After being washed in pipe- line water or in Ringer solution (1-4 hours), the objects were placed for 4 hours into 20 "o solution of gelatin heated up to the temperature of 37°C and then for 15 hours into a 40 "„ solution. During this period of time the pieces of muscle tissue to the size of 1 0.3 0.3 mm are well impregnated with gelatin. Then the 40 "o solution of gelatin with the objects enclosed in it is poured out on oil cloth and dried slowly at the temperature of 37 C to such a state in which gelatin is not brittle. 114 V. p. GILEV The plate of gelatin with the objects included in it is separated from the oil-cloth and cut for rectangu- lar blocks. The blocks are glued up with 40 "o solu- tion of gelatin to the old methacrylic blocks sharp- ened in the form of a truncated pyramid and placed for several hours into a thermostat (37'C) for com- plete drying. Immediately before cutting in a microtome the blocks are sharpened in such a way that the square of the section will not exceed 0.2 mm-. The blocks are kept in cans with tight caps. Cutting is effected with a glass knife (without fluid). In our work we used a microtome of Danon and Kcllenberger. The long side of the block should be parallel to the sharp edge of the knife. The blocks just dried are cut worse when the weather is dry; in this case it is better to cut them not earlier than in 24 hours after they were exposed to the air. In the case of considerable humidity of the air and blocks, it is recommended to dry them some time before cutting. Position of the knife in cutting gelatin blocks is usually, to some extent, steeper than in cutting methacrylic blocks. The rate of cutting is approximately I section per I sec. It is not difficult to obtain sections 0.03-0.04 /< thick. Usually the sections are formed in the shape of endurable un- disintegrating ribbons. These ribbons are sorted out and individual sections or groups of sections are trans- ferred with the aid of a thin filament into drops of water (or better of 2 % solution of acetic acid) lying on a I.e. Parlodion film that floats on the surface of water (temperature of about 37-40 C) filling a Koch cup. But better results are obtained when the sec- tions are placed into water of room temperature, which is subsequently heated up to 37-40 C. Expanding, the sections in most cases lose their bonds with each other and, therefore, not always a series of sections may be obtained. Then the film under each drop is punctured with a thin needle and the sections sink to the undamaged part of the film. The copper grids are placed on the film in such a way that the centre of the grid is above the section. After that from above a microscope slide is placed. Having put it upside down, we withdraw the film with grids.* After drying the specimens are ready for examination (fig. I ). For more complete removal of gelatin from the sections the ready grids with the specimens are placed for 2-4 hours into warm water (37-40 C) or into a weak solution of acetic acid. The osmium tetroxide that remains in the object after fixation and washing (especially when it is done for a short period of time) interacts with gelatin, after which the latter becomes less transparent, hardly removable from the section and creates a rather strong background that makes some fine de- a ** T^ , ' -.»'■ Fig. 1. Striated muscle tissue of the axo\otl( A mblystoma pitnctotttm). The section has been treated with 4% acetic acid. tails of the structure of the tissues less visible. There- fore, it is recommended that before enclosing the objects in gelatin, fixation should be shorter and washing longer. It can be supposed, however, that the presence of the background obliterating the bound- aries of some structures, is connected with the presence of substances which are otherwise elimi- nated or precipitated, while the specimen is treated with alcohols and methacrylates. In the case of thinner sections it is recom- mended to examine them without removal of em- bedding medium (gelatin). In this case the sections should be expanded in cold water (about - I to ^ rc). The tissues, embedded in gelatin, can be easily cut with a glass knife for sections 1 fi thick for exa- mination in the light microscope. In this case the blocks should not be too dry and, therefore, are kept in open vessels. The sections are glued up to micro- scope slides with albumen, to which some glycerine has been added. The gelatin is dissolved in warm water. References 1. Brandes, C. H., Mikrokosmos 44, N7, 167 (1955). 2. FiRMiNGER, H. I., Stain Techno/. 25, N 3, 121 (1950). 3. Newman, S. B., Borysko, E., and Swerdlow, M., /. Research Nat. Bur. Standards 43, 183 (1949). 4. Palade, G. E., J. E.xptl. Med. 95, 285 (1952). 5. Richards, A. G., Anderson, T. F., and Hance, R. T., Proc. Soc. E.xptl. Biol. Med. 51, N 1, 148 (1942). ^ This method ofmounting sections was elaborated together with V. I. Birjuzova, one of the workers of our laboratory. On the Preparation of Ultrathin Serial Sections by Means of a Watchmaker's Lathe W. NiKLOWITZ liistitiit fiir Mikrohiolof^ic iiiul L.xpcrinu'ntelle Theiapie, Jena, clcr Akademie cicr Wissenschaften zii Berlin Ultrathin sections suitablcfor studying biological objects in the electron microscope arc prepared by means of conventional microtomes or by making use of new constructions of these instruments (for references, see 16, 20). To-day's trend is to employ ultramicrotomcs which have proved most satisfactory— such as the excellent ultramicrotome of Sjostrand. However, since about one year we make use of a watchmaker's lathe in preparing ultrathin sections. This apparatus is characterized by great simplicity both in its con- struction and in its operation. I should like to try to give a short description of the instrument, and to report subsequently some results obtained. The basic principle of this ultramicrotome is a commercial watchmaker's lathe, as it was used al- ready by Giuntini and Edlinger (10). Technical de- tails have already been reported elsewhere ( 16). This ultramicrotome is fitted with a thermal advance of the knife similar to that recently described (6). The diflference is, that with our apparatus the cooling of the metal rod supporting the knife is utilized in forwarding the latter. According to the rotating principle, the object to be cut passes the glass knife (made of Jenaer Gerate-Glas 20) only once during one cutting cycle. This ultramicrotome provides se- ries of ultrathin sections that are relatively uniform and have a mean thickness of about 300 A. The thickness of the sections may be adjusted by means of the speed with which the sections are made. In the meantime several of these microtomes are put into operation, and we may now state that they have given the same satisfactory results. In order to demonstrate the usefulness and effi- ciency of this microtome, some results of our in- vestigations, partly still in course, partly already published, shall now be reported. The first micrographs of ultrathin sections shown were given for comparison only. They represent sections of the kidney of the white mouse, an organ generally chosen as an object for tests, because of the variety of its submicroscopic cellular structure (16). Regarding the problem of the origin of specific granules from the milochondria in Elvlich's ascites carcinoma of the white mouse, the microscopical (phase contrast) and histochemical investigations carried out in our institute by Kieser (12) shall be extended by means of ultrathin sections. The mito- chondria of the tumor cells have the same specific substructure as already described for other tissues (21). The cristae, however, may be arranged both in the longitudinal and the transverse direction of the milochondria. Furthermore, there are to be seen single or several spherical granules, strongly osmio- philic, and enveloped by a common membrane. They are, in early stages of development, of the same magnitude as the mitochondria. These preliminary results, therefore, may suggest that the granules possibly may arise from mitochondria, as is the case in the cloudy swelling of the kidneys (8). Recently it was shown on difTercnt objects that the plastids of the higher plants are formed of sub- microscopic lamellae (summarizing literature see Frey-Wyssling, (7)). Systems of such lamellae are demonstrable also in the chromatophores of the green algae. This fact has been demonstrated by means of a micrograph of a chromatophore of a green algae, the mean thickness of the lamellae being 30 to 40 A. This observation led us to the investigation of the substructure of rv«//(v^/nrfY/(' (blue-green algae) dur- ing the past few years. In this group of organisms, too, we were able to state that the chromatoplasm — the region of the cell containing the assimilative pigments — consists of submicroscopic systems of lamellae (18, 19). Having already furnished proof in a previous paper that systems of lamellae are to be found in Rhodospirillum rulvum, an autotrophic bacterium (17), we are now able to state definitely that /// all ori^anisms containitii; chlorophyll , theassi- milatory pif^ntents are probably attached to submicro- scopic systems of lamellae. Furthermore, three types of granules may be differentiated in cyanophyceae: small, strongly os- miophilic granules, granules of inorganic phosphate, and, as a particular and characteristic group, gran- ules exhibiting lamellar structure in ultrathin sections and having mitochondrial function, as was demon- strated in histochemical investigations. This latter group of granules was provisionally designed as "ferment active granules" (4, 5). Among the representatives of the Nostocaccae there are further distinct elements in the region of the ccntroplasm. a detailed description of which will be published elsewhere (19). For a number of years, we have also been con- cerned, to a certain extent, with the substructure of bacteria (13-15). The results of our recent investiga- tions, comprising primarily the problem of the nu- cleus and of equivalents of the nucleus respectively, will be published elsewhere. According to our obser- vations, the equivalents of the nucleus, both in bac- teria and in cyanophyceae. as seen in electron micro- 16 W. NIKLOWITZ Fig. 1. Electron micrograph of an ultra-thin longitudinal section from Escherichia coli (three hours culture). Magnifi- cation 90,000. scopic investigations, reveal a structure deviating from that of higher organisms. In our opinion the idea drawn from what was seen in uUrathin sections and light-microscopical investigations, i.e. that bac- teria contain true nuclei with chromosomes, is pre- mature, especially when taking into consideration that even the results concerning the fine structure of chromosomes in higher organisms are still rather incomplete. Moreover, comprehensive studies of fixed series of different stages of the development revealed that the image will largely depend upon the fixation. According to Geitler (9), emphasis should not rest on the alternative question nucleus or no nucleus, chromosomes or no chromosomes, but on exactly hitting upon the equivalents of the nucleus. From our investigations the following procedure has proved most favourable for the fixation. The objects — whether prefixed (2, 11) or not — are fixed in a puffered solution of OsO^ (pH 6.8; 7.1; 7.4), added with 0.2 Af saccharose for 10-15 min. at 4'C. This fixation is followed by a further one in formol (1:10) for 4-6 hours. This procedure avoids an unfavourably long expo- sure to OsOj, as well as the addition of NaCl (3). As has earlier been stated ( 1 ), a prolonged fixation with OsOj causes a destruction of the tissues, and the addition or a washing with NaCI entails an enormous loss of substances (22). Fig. I represents a section through Escherichia coli. The substructure of the cytoplasm is granular. In the central bright region filamentous elements with a mean thickness of 80 A may be recognized. From our investigations it became evident that the distribution and the diameter of these filaments will vary with different fixations. This short report at this Conference was intended to demonstrate that our ultramicrotome renders possible to do routine sections which are suitable for carrying out submicroscopic studies, and 1 should like to conclude my arguments citing a sentence of the nestor in the field of investigation of submicro- scopic structures, Frey-Wyssling, who states that just as classical physiology reached its present summit only following the perfection of histology, the pro- gress in physiology of the cell will depend upon shedding more light on the submicroscopic morpho- logy of the cell. References 1. Bahr, G. F., Exptl. Cell Research 9, 277 (1955). 2. Birch-Andersen, A., Maaloe, O., and Sjostrand, F.S., Biochim. et Biophys. Acta 12, 395 (1953). 3. Chapman, C. B. and Hillier, J., /. Bacterial. 66, 362 (1953). 4. Drews, G. and Niklowitz, W., Arch. Mikrobiol. 24, 147 (1956). 5. — ibid. 25, 333 (1956). 6. Ekholm, R. andZELANDER,T., EATJe/vV/fZ/rt 12, 195(1956). 7. Frey-Wyssling, A., Protoplasntatologia 2, A/2 (1955). 8. Gansler, Ft., Vortrag auf der 6. Tagung der Deutschen Gesellschaft fiir Elektronenmikroskopie, Miinster 1955. 9. Geitler, L., Fortschr. Bat. 17, 1 (1955). 10. GiUNTiNi, J. and Edlinger, E., Ann. Inst. Pasteui 86,671 (1954). 1 1. Kellenberger, E. and Ryter, A., Schweiz.Z. Pathol, u. Bakteriol. 18, 1122 (1955). 12. KiESER, H., Arch.Geschwulstforsch. (1957, in press). 13. Knoll, H. Zeifl-Nachrichten 5, 38 (1944). 14. Knoll, H. andZAPF, K.,Zentr. Bakteriol. Parasitenk., Abt. I Orig. 157, 389 (1951). 15. — ibid. 161, 241 (1954). 16. Niklowitz, W., Mikroskopie, Wien, 10, 401 (1955). 17. Niklowitz, W. and Drews, G., Arch. Mikrobiol. 23, 123 (1955). 18. — ibid. 24, 134 (1956). 19. — ibid. (1957. in press). 20. SiTTE, H., Mikroskopie, Wien, 10, 365 (1955). 21. Sjostrand, F. S. and Rhodin, J., E.xpil. Cell Research 4, 426 (1953). 22. Sylven, B., Acta Union Intern, contre le Cancer 7, 708 (1951). How to Prepare Ullralhin Sections of Tissue Cullures V. Dost A L Bi'hiins^-Werke, Marhiiii; a. d. La/in. and the Department of Hyj^'icne, Alhert-Liidwii^s-Univer.sitdt, Freihioi,' i. Br. Owing to improvements in method, tissue cultures have in recent years become of growing importance for the cuUure of viruses, for the quantitative deter- mination of infectiosity, for carrying out neutrah/a- tion tests, and for the morphological study of multi- plication processes. In addition, they are being used for diagnoses and for the production of vaccines. In recent years I have worked on the culture of viruses, especially the poliomyelitis virus and the vaccine virus. The basic materials used were monkey and calf kidneys which I prepared for the examina- tion in the electron microscope. The tissues treated (1,5) in culture bottles grow into a cell outgrowth predominantly in monolayers at the bottom of the container within a few days. Various methods as to how to produce ultrathin sections are already known. D. C. Stuart (4) specified a method in which the embedding, i.e. the poly- merization, takes place directly at the cell attached to the tube. C. G. Harford, A. Hamlin, and E. Parker (2) have the tissue grow on a formvar foil, after- wards embedding the latter. I chose to make a sedi- ment of tissue cultures which I obtained from full- grown, normal, and infected cultures. For preparing the sediment I used such tissues as are normally developed in producing poliomyelitis vaccine. The tissues were partly not infected, partly they displayed a certain state of virus multiplication within the cell. The tissue to be examined by electron microsocpy is processed up to its being embedded in the culture bottle, it is fixed with a solution of I "o phosphate-butTered osmium tetroxide (according to Sjostrand). In order to remove cell detritus, the cell surfaces were previously rinsed with Hanks" solu- tion. The cells were dehydrated with alcohol. The change to methacrylate mixture was via various Fig. 1. An ultrathin section of a monkey kidney tissue culture wiiich had been infected with vaccine-virus. The tissue was fixed 48 hrs. after infection. In the photograph the nuclei of two cells are to be seen which distinctly show the double membrane. Furthermore, the cell border is easily recognizable as a double membrane. In the protoplasma various mitochondria with their septa. All over the hyaloplasma there were lots of small, rotund forms; besides, there were some hyaline osmiophilic areas. Siemens Electron microscope UM 100 c, Magnification -28,000. 118 A. MAAS alcohol-methacrylate mixtures. Since the cell out- growth is predominantly in monolayers, the respective periods during which the tissue has to stay in the various media from fixation to embedding may be comparatively short. Cells were damaged by staying too long in alcohol of a higher percentage, e.g. over night. No shrinkage of cells was noticed by light microscopic examination. The tissue was detached from the bottle in the non-polymerized methacrylate mixture, by means of a rubber-coated, bent glass rod. By centrifuging the cell suspension at about 300-500 rpm for five minutes one gets a compact sediment. When properly fixed, its color is a dirty brown. Sedimentation was already effected in a butyl- methyl-methacrylate mixture (4:1) with 2 % dichlor- benzoylperoxide (3). The methacrylate mixture was not polymerized preliminarily; however, the filled gelatine capsules were evacuated for five minutes before being deposited in the incubator in order to ensure removal of gas pockets, which are easily formed in sediments. Incubation was carried out at 48X for about 30 hrs. The ultrathin sections were prepared with an ultra-microtome (Fernandez-Moran) placed at my disposal by courtesy of Messrs. E. Leitz at Wetzlar. The knife used is a diamond built into a container to receive the sections. The container was filled with an alcoholic solution of 25 %. The sections appeared in the shape of an uninterrupted band and spread out properly on the water surface within 15 minutes. By heating the water it was possible to speed up this process without the quality of the sections deterio- rating. The sections themselves showed grey-white interference colors. The bands of sections were re- ceived by mesh copper grids with a formvar film. The sections (fig. 1) were observed and photo- graphs were taken in a Siemens electron microscope Cm 100 c with stigmator and an aperture of 30 n and an accelerating voltage of 60 kV. The examinations carried out have proved that it is possible to obtain results which can easily be reproduced by a simple method of preparation, such as the preparation of sediments by trimming of the cells from the culture containers. It enables both infected and non-infected tissue cultures to be ex- amined by electron microscopy. These investigations were kindly supported by Drs. Grehn and Walter, Wetzlar. I also wish to extend my thanks to Miss M. Hahn of the Behring-Werke for her cooperation in preparing the sections and taking the electron micrographs. References 1. DuLBECco, R. and Vogt, M., /. Expil. Med. 99, 167 (1954). 2. Harford, C. G., Hamlin, A., and Parker, E., /. E.xptl. Med. 101, 6, 577 (1955). 3. Newmann, S. B., Borysko, E., and Swerdlow, M., /. Research Nat. Bur. Standards 4, 43, 183 (1949). 4. Stuart, D. C, Ann. Report 26 (1954). 5. YouNGNER, J. S., Proc. Soc. E.xptl. Biol. Med. 85, 202 (1954). Eine einfache Vorrichtung ZLim Anspitzen von plexiglaseingebetteten Objekten A. Maas ZentraUaboratorimu fiir angewandte Ubermikroskopie dcr Universitat Bonn L)as vor dem Schneidcn auf dem Mikrotom not- wendige Anspitzen plexiglaseingebetteter Objekte, das unter mikroskopischer Beobachtung von Hand mit einer Rasierklinge vorgenommen wird, ist bei kleiner ObjektgroBe sowie bei groBer Hiirte des Plexiglases mit Schwierigkeiten verbunden. Im fol- genden soil iiber eine Vorrichtung berichtet werden, die es gestattet, plexiglaseingebettete Objekte bis zu einer Dimension von einigen // in die Spitze einer Pyramide zu legen, deren Spitzen- und Fliichenwinkel exakt definiert sind. Das Objekt wird in iiblicher Weise in das halb- kugelformige Ende eines zylindrischen Plexiglas- blockes eingeschlossen. Der Plexiglasblock wird so- dann in eine Haltevorrichtung (Blockhalter) einge- spannt. Die sich beim Anspitzen des Plexiglasblockes ergebende Form des angespitzten Objektes ist mit- bestimmend fiir die GroBe und Giite der spateren Ultra-Diinnschnitte. Durch die GroBe des Spitzen- winkels wird die statische Festigkeit des Objektes beim Schneidevorgang festgelegt. Auf Grund der im Objekt beim Schneiden auftretenden Zug- und Druckspannungen wird somit die Giite des erhalte- nen Ultra-Schnittes beeinfiuBt. Mit Hilfe einer Rasierklinge wird der Plexiglas- block so angespitzt, daB sich das Objekt in der Spitze einer Pyramide befindet. Das Prinzip der Anordnung ist in Abb. 1 im schematischen Grundriss dargestellt. Der Plexiglasblock ist um eine Achse An um definierte Winkelbetriige drehbar und in der Liingsrichtung der Achse verschiebbar angeordnet. Das eingebettete Objekt kann mittels eines spater Anspitzen von plexiglaseingehetteten Ohjekten 119 Vitxijlai-BUick ANSICHT Spiegel ANSICHT VON BfW S€lTe Abb. 1. Prinzip-Skizze. J V,— Abb. 2. Oplik. noch zu besprechenden vereinfachten Kreuztisches in die Drehachse A„ verlagert werden (Justierung J). Das zum Anspitzen des Blockes dienende Messer befindet sich zentral verschiebbar auf einem Schwenk- arm (Verschiebungsrichtung V^), der in horizon- taler Ebene um den in Verliingerung von A„ liegen- den Punkt Z gedreht werden kann (Winkeleinstel- lung ^^^^/). Der so veranderbare und jeweils fest einstellbare Winkel zwischen K.v und der Richtung der Drehachse A„ ergibt den als Winkel zwischen zwei gegeniiberUegenden Fliichen der angespitzten Pyramide definierten Spitzenwinkel 2 rp. Der Winkel 7 ist stufenlos von 0-90 , der Spitzenwinkel der angespitzten Pyramide somit von 0-180 wiihlbar. Der Vorschub (Fg) des eingebetteten Objektes the acridine dyes were studied by means of electron microscopy and polarization optical analysis. Fig. 1 shows chondroitin sulphuric acid in a control sample and hg. 2 when treated with trypa- flavin. Except for myosin, all our components are globular. As demonstrated in fig. 2, the structure changes 132 F, GUBA, G. HAJOSSI-KEREK AND G. ROMHANYI Fig. I. Chondroitin sulphuric acid. Control sample. Fig. 2. Chondroitin sulphuric acid-trypaflavin complex. The reticular structure characteristic of trypaflavin complexes can be seen very well. when mixed with the dye. The dyes always brought about precipitation. The consistency of the precipi- tates was different with every compound; the most compact precipitate was obtained with rivanol. The results show that the change in structure is, in principle, characteristic for the dye, but in addi- tion the chemical composition of the substrate is of importance. These conclusions were supported by our polarized light analysis. We think that the structure which develops is due to a coupling of the macromolecules through the dye molecules. This is supported by the following observation: we solved the newly developed structure by the known ditTusion method with NaOH and ethyl alcohol respectively. On the elimination of either of the components the newly developed structure breaks up. The coupling depends on the steric behaviour both of the polyelectrolyte and the dye molecule. It is obvious that the development of this structure depends on the side groups of the acridine, chiefly of its NH., groups. References 1. DoGSON, K. S., Rose, F. A., and Spencer, B., Biochein. J. 60, 346 (1955). 2. RoMHANYi, G., Congress of the Hungarian Pathological Soc, Debrecen, 1955. VI CELL ULTRASTRUCTURE, GENERAL Elektronenmikroskopische Studien an Leberschnitten von Thyroxin-behandelten Ratten H. ScHULZi, H. Low, L. Ernster und F. S. Sjostrand Aus dem Anatoniischen Inst it lit des Karolinska Institutes und dem Wennei-Giens Institiit der Universitdt Stockholm Die Schilddriisenhormone bewirken eine Labilisie- rung der zelluliiren Multienzymsysteme, wie bioche- mische Ergebnisse von Aebi und Abelin (1 ), Tapley, Cooper und Lehninger ( 1 3) und von Klemperer (6, 7) gezeigt haben. Martius und Hess (8, 9) fanden bei steigenden Thyroxinkonzentrationen eine fortschrei- tende Hemmung der oxydativen Phosphorylierung in der Atmungskette, die bis zu einer vollstiin- digen Entkopplung fiihrt. Eigene Studien an iso- lierten Mitochondrien von thyreotoxischen Ratten- lebern ergaben, daB die Wirkung des Thyroxins wahrscheinlich eher verbunden ist niit strukturellen Veriinderungen der Multienzymsysteme als mit einer spezifischen Hemmung einzelner Enzyme (2, 3). Wir haben unsere elektronenmikroskopischen Untersuchungen durchgefiihrt, um Befunde der Morphologie biochemischen Ergebnissen gegeniiber zu stellen. Eine zusammenfassende Beschreibung der submi- kroskopischen Struktur der normalen Rattenleber wurde bereits von Fawcett (4) durchgefiihrt. Palade und Siekevitz (11) beriicksichtigten vorwiegend die Lebermikrosomen. Das Verhalten der Lebermito- chondrien unter pathologischen Bedingungen wie Hunger und akute Hypoxie untersuchten Gansler und Roullier (5), sowie Molbert und Guerritore (10). Erwachsene Ratten erhielten intraperitoneal eine tagliche Dosis von 0,4 mg DL-Thyroxin (Hoffmann- La Roche) uber 5 Tage; am sechsten Tag wurden die Tiere getotet und Leberstuckchen von 0,5-1 mm Kantenlange lebensfrisch nach Sjostrand (pH 7,2) fixiert. Postmortale Veriinderungen versuchten wir auszuschlieBen, indem wir zu jedem Versuch ein Normaltier unter denselben Bedingungen priiparier- ten, sowie nur Schnitte aus der Oberfliiche der Gewebsstuckchen verwandten. Die Schnitte fertigten wir an mit dem Ultramikrotom nach Sjostrand. Fiir die Aufnahmen benutzten wir das RCA EMU 2c Elektronenmikroskop. Die in den Tabellen und im Text angegebenen MaBe sind das statistische Ergeb- nis von Messungen an verschiedenen Leberzellen von den untersuchten sechs Ratten: iiber 100 Mes- sungen liegen den jeweiligen Zahlen zugrunde. Zum lichtmikroskopischen Nachweis des Glyko- gens mit dem Fiirbeverfahren nach Best wurden einige kleine lebensfrische Leberstuckchen von den Versuchs- und Kontrolltieren zur Vermeidung der „Glykogenflucht" funf Minuten mit einer m 3 NaOH-Losung behandeh (12), danach in absolutem Alkohol fixiert und in Paraffin eingebettet. In unseren elektronenmikroskopischen Studien untersuchten wir besonders die Veriinderungen an den Lebermitochondrien. Nach Behandlung mit Thyroxin zeigen die Mitochondrien eine deutliche Schwellung und eine Aufhellung der Matrix. Die Aufhellung der Matrix beginnt immer in der Mitte eines Mitochondriums und nimmt allmiihlich zur Peripherie zu. In Normaifallen besteht die Matrix aus einer iiber das ganze Mitochondrium gleich- maBig verteilten, feingranulierten Grundsubstanz von miiBiger Osmiophilie (Abb. 1). Bei einer Schwellung ist sie zusammengesetzt aus unregelmaBigen Flecken mit unterschiedlichem, weniger dichtem Kontrast Abb. I: Aiissclinitl aus einer normalen Leberzelle der Ratte. Zahlreiche liings- und quergeschnittene Mitochondrien. Im Cytoplasma auBerdema-Cytomembranen sowie doppeltkon- turierte Membranen, die die Mitochondrien zirkuliir umge- ben. Elektronenoptisch 8600 , Abb. 24000 . ^ Wissenschaftlicher Assistent am Pathologischen Institut der Medizinischen Akademie Diisseldorf; Stipendiat der Stadt Diisseldorf. Leberschnitte von Thyroxin-behandelten Ratten Tabelle 1 . Lc'inge iind Breite der Mitochondrien. 35 Geringste GroBte Breite, n Breite, n Durch- schnittl. Geringste GroBte Durch- schnittl. Breite, /' ^ ^ Lange, Verhaltnis Breite zu Lange Normale Lebermitochon- drien der Ratte 0,136 0,506 0,302 Nach Behandiung mit Thyroxin 0,528 0,566 0,939 0,578 0,801 1,73 2,187 1,03 1,376 3:10 6,6:10 Tabelle 2. Anzahl und Lange der Innenmemhranen in eineni M itochondrien-Ldngsschnitt . Hochste Anzahl der Innen- memhranen Kleinste Anzahl der Innen- memhranen Durchschn. Anzahl der Innen- memhranen GroBte Geringste Durchschn. Liinge einer Liinge einer Liinge der Innen- Innen- Innen- memhran membran memhranen m/< m/< mil Normale Lebermitochon- drien der Ratte . . . Nach Behandiung mit Thyroxin 34 25 19 10 380 189 48 16 128 86 Abb. 2: Ausschniti aus einer Leberzelle der Ratio nach Behandiung mit Thyroxin. Geschwollene Mitochondrien und Vakuolisierung des Cytoplasmas. Elektronenoptisch 8600 , Abb. 24000 . und darin liegenden feinen fiidigen Strukturen (Abb. 2). Es ist mogUch, daB dieses durch eine strukturelle Veriinderungder EiweiBmolekiile bedingt ist. Andere geschwollene Mitochondrien habcn eine unveriinderte Matrix. Die morphologischen Unter- schiedc diirften biochemisch verschiedcnen Labili- tatsstufen der Multienzymsysteme entsprechen. Das wesentlichste Kriterium fur eine Mitochon- drienschwellung im elektronenmikroskopischcn Bild ist eine Breitenzunahme des Mitochondrienkiirpers. Das GroBenverhaltnis von Breite zu Liinge bctriigt in Normalfallen 3:10, nach Thyroxinwirkung 6.6: 10 (Tabelle 1). Die statistische Auswertung der elcktronenopti- schen Befunde ergibt weiterhin, daB die Gesamt- oberfliiche der ,,cristae mitochondriales" nach Be- handiung mit Thyroxin sich gcgeniiber normalen Lebermitochondrien um maximal 30 ",, vermindcrn kann. Auf Schnittbildcrn ist die durchschnittliche Anzahl der Innenmemhranen pro Mitochondrium \cin IS aul' 10 reduziert: die durchschnittliche Lange einer Innenmembran ist um 'j verkiirzt (Tabelle 2). Oft erkennt man auf den Bildcrn nur noch 15 m/< kleine Restc von Innenmembrancn. Geschwollene Mitochondrien sind stets umgeben von intakten AuBcnmembranen. Durch die Volumcnzunahme ist die Gesamtohertliiche der MitochondrienauBenmem- branen um das 3^ fache vergroBert worden. Da die AuBcnmembranen nicht diinner geworden sind, muB eine Neubildimg von Mcmbranmaterial erfolgt sein. Es ist moglich,daBdiedafiJrnotv\endigen Baubestand- teile von den Innenmemhranen herrLihren. FaBt man die anahtischcn Messungen der Membransysteme zusammen (Tabelle 3), so sind nach Behandiung mit 136 H. SCHULZ, H. LOW, L. ERNSTER UND F. S. SJOSTRAND Tabelle 3. Dbersicht der Messimgen an den Mitochondrien-Innen- und Aufienmembranen. AuBenmembranen in A Innenmembranen in A „ ^, . Osmiophile Osmiophobes ^ ^, •. Osmiophile Osmiophobes Gesamtbieite c u- u* t * n Oesamtbreite c u- L^ r ^ ^^ Schichten Intervall Schichten Intervall Mittel Mittel a Mittel Mittel o Mittel a Mittel Nor male Lebermitochondrien der Ratte Tier I Tier II Tier III Wert: 168±3 16 42 ± 4 19 84 + 2 20 118±8 25 53 ±5 34 48 ±4 9 128±6 21 50r 12 8 38 r 5 7 147±7 32 38 + 6 18 42±4 16 147±2 12 46 r 8 15 55 ±3 13 189 + 4 19 63 ±3 21 63 + 7 27 150 A 45 A 60 A 150 A 50 A 50 A Nach Behandliing mit Thyroxin Tier I 174±8 9 44+3 12 78 ±4 10 1 83 r 8 18 57r3 6 69±4 8 Tier II 162±5 13 42+4 6 87±4 10 172 + 6 12 61±4 8 76±6 20 Tier III 170 + 3 4 48- 3 8 71±5 5 189±3 21 48^6 12 65 ±5 16 Wert: 170 A 45 A 80 A 180 A 55 A 70 A Thyroxin die osmiophoben Schichten der Aufien- und Innenmembranen um 20 A verbreitert, was wahrscheinlich auf eine Schwellung zuriickzufUhren ist. Die osmiophilen Schichten der Membransysteme bleiben unverandert. Thyroxin bewirkt auBerdem Veranderungen am Cytoplasma der Leberzellen. Man findet zahlreiche verschieden groBe, iiber die ganze Leberzelle gleich- maBig verteihe Vakuolen, deren Inhalt elektronen- optisch leer ist. Die Membran der Vakuolen ist 40- 65 A dick und von auBen mit homogenen Partikeln bedeckt. Der seitliche Partikelabstand betriigt 100- 210 A, die Partikelbreite 1 20 A. Wir nehmen an, daB die Vakuolen aus den a-Cytomembranen durch eine Starke Schwellung der homogenen Zwischenschichten entstanden sind. Diese Veranderung ist so ausgepragt, daB man von einer weitgehenden Veranderung des Ergastoplasmas sprechen kann (Abb. 2). Auf Grund vergleichender lichtmikroskopischer Untersuchungen ist in den Vakuolen wahrscheinlich kein Glykogen enthalten. Die festgestellten Veranderungen (Abb. 3) an den 1 i ^w. t 50 h 50ft I50fi / ^- 150 ft -► / (,0k -1 hSft-*' % -ISO f) Abb. 3. Schematische Darstellung der Modifikationen an den Mitochondrien und den a-Cytomembranen der Ratten- leber nach Behandlung mit Thyroxin. Links sind die normalen Verhaltnisse dargestellt. Geformte Sekrete der Mitochondrien von Paraiiicciiiiii 137 Mitochondrien und am Ergastoplasma bcruhen auf Schwellungsvorgiingen, die veriirsacht sind durch eine Jnsuffizienz der osmotischen Regulation. Die Schwellung der Mitochondrien, die ,.in vitro" auf Grund zahlreicher biochemischer Studicn nachge- wiesen wurde. ist auch aus unseren elektroncn- optischen Aui'naiimen ersichtlich. Ob die Insuitizienz der osmotischen Regulation z. B. bedingt ist durch einen ATP-Mangel oder durch lonenverschiebungen, kann aus den elektronenoptischen Ergebnisscn nicht beantwortct vverden. Die Fragen der Membranverminderung und Mem- branvermehrung sind moglicherweise verbunden niit Anderungen in der Atmungsintensitiit des Mitochon- driums. Aus bisherigen Untersuchungen zum Wir- kungsmcchanismus der Schilddriisenhormone ist be- kannt, daB bei der Hyperthyreose, trotz reichlichem O^-Angebot. eine verminderte Fahigkeit besteht, O^ aufzunehmen. Die eigentliche Ursache der Mito- chondrienschwellung besteht jedoch in einer Labili- sierung der enzymatischen Aktivitiit. wie eigene bio- chemische Untersuchungen bestiitigt haben. Die Unterschiede in der Morphologic der Mitochondrien entsprechen hierbei verschiedenen Labilitiitsstufen der Multienzymsysteme. Es bleibt die Frage zu erdrtern, ob mit den elektronenoptischen Bet'unden bewiesen werden kann, daB die Modifikationen am Chondriom und an den a-Cytomembranen dem ,,in vivo"-Zustand entsprechen. (Jber die abschlieBende Prufung der morphologischen Veriinderungen in Be- ziehung zu den biochemischen Ergebnissen wird ausfUhrlich an anderer Stelle berichtet. LiTERATUR 1. Alhi, H. und AnhUN, I., Elcklrolyt- und Fermcnthaus- halt dcr hypcrthyrcotischcn Leber. Biochem. Z. 324, 364-384 (1953). 2. Beykr, R. E., Low, H., und I.knsiik, L., Ihc ctVcci of thyroxine on mitochondrial stability. Proc. Norwegian Biochem. Soc, Oslo, 2-3 (1956). 3. — Acta Clu'iii. Scancl. (1957, in press). 4. Fawcett, W., Observations on the cytology and electron microscopy of hepatic cells. J. Nat. Cancer Inst. 15, 1475-1502 (1955). 5. Gansler, H. und Rouiller, Cm., Motlilicaiions ph>sio- logiques et pathologiques du chondriome. (Ltudc au microscope electronique.) Sc/iweiz. Z. allgem. Pathol. II. Bakteriol. 19, 217-243 (1956). 6. Klemperer, H. G., The uncoupling of oxidative phos- phorylation in rat-liver mitochondria by thyroxine, triiodothyronine and related substances. Biochem. J. 60, 122 128 (1955). 7. — The uptake of thyroxine and triiodothyronine by rat-liver mitochondria, ibid. 60, 128-135 (1955). 8. Martius, C. und Hess, B., The mode of action of thyroxin. Arch. Biochem. Biophys. 33, 486^87 (1951). 9. — LJber den Wirkungsmechanismus des Schilddriisen- hormons. Biochem. Z. 326, 191-203 (1955). 10. MoLBERT und GuERRiTORE, ziticrt nach BCrnNER, F., Hemmung der Oxydationen als pathogenctisches Prinzip. Klin. Wochschr. 34, 777-781 (1956). 11. Palade, G. E. und Siekevitz, P., Liver microsomes. An integrated morphological and biochemical study. /. Biophys. Biochem. Cytol. 2, 171-201 (1956). 12. Takahashi, K. und Iwase, S., On artifacts appearing in the histochemical fixation of glykogen. /. Biophys. Biochem. Cytol. 1, 391-399 (1955). 13. Tapley, D. F., Cooper, C, und Lehninger, A. L., The action of thyroxin on mitochondria and oxidative phosphorylation. Biochim. Biophys. Acta 18, 597- 598 (1955). Die Entstehung, (die Vermehrung un(d die AbscheicJung geformter Sekrete der Mitochondrien von Paramecium K. E. WOHLFARTH-BOTTERMANN Zentral-Laboratoriiim fiir augewandte Vbennikroskopic der Univcr.sitcit Bonn Uber den Organellcharakter und die wichtigen Funktionen der Mitochondrien im Zellstoffwechsel besteht heute kein Zweifel mehr: Biochemische Untersuchungen haben die stoftliche Zusammenset- zung und die enzymatische Aktivitat dieser Zell- organellen aufgezeigt (5). Das Elektronenmikroskop konnte in den letzten Jahren erweisen. daB der hochgeordneten enzymatischen Funktion eine ebenso hohe strukturelle Ordnung zu Grunde liegt (6-10, 14, 15). Dagegen fehlen uns bis heute eindeutige Auf- schliisse iiber die Entstehung und die Vermehrung der Mitochondrien, was fiir die Beurtcilung eincr Kontinuitat des Chondrioms von Bedeutung ist. Es geht um die Frage, ob wir die Mitochondrien als Autoduplikanten zu betrachten haben oder ob sie etwa aus anderen cytoplasmatischen Strukturen entstehen konnen (1, 4). Bekanntlich wird versucht. gewisse Phanomene der plasmatischen Vererbung durch eine Autonomic des Chondrioms zu dcuten. Als Untersuchungsmaterial diente ein Klon von Para- incciuni caiidatiiiii. dcr in Hcudckokt kulli\icrt wurde. Nach Fixation mit Osmiumtctroxyd, Formalin odor mit dcm Gemisch nach Rcgaud, Champy und Maximow wurden die Zellen in einer ansleigcnden Alkoholrcihc enlwiissert und in Plcxiglas eingcbcttct. Diinnschnitte von 200 300 A Dickc geiangtcn ohnc Herausloscn des Einbcttungsmediums im Elektronenmikroskop zur Un- tersuchung. (Ultra-Mikrotom nach Sjostrand, Siemens- Ubcrmikroskop Typ 100 d, 80 kV.) Als bcstcs Fixicrungs- mittel erwics sich eine l"oige Losung von Osmiumtctroxyd bzw. cine 10"oige Formalinlosung, beidc mit dcm Acetat- veronalputTcr nach Michaclis auf cincn pU-\Vert zwischen 6,5 und 7,2 cingestellt. Als Kontrastmittel haben sich Phosphorwolframsaure (PWS P.,0,-24WO;, • xH.,0)undThalliumnitrat(Tl = TINO:,) bewiihrt, beidc jewcils zu i "„ in VO^o'gem Alkohol gelost (15). 138 K. E. WOHLFARTH-BOTTERMANN Fig. 1. Schematische Darstellung der Feinstruktur der Mito- chondrien von Paramecium caudattim. A, schematischer Dunnschnitt; B, raumliche Rekonstruktion. Es zeigte sich, daB die Mitochondrien von Para- nu'ciiim einen anderen Bauplan besitzen als die Mitochondrien aus Saugergeweben. Im Inneren der Mitochondrien finden sich als wesentliche Struktur- elemente Rdhren bzw. Kanale. Uber den Inhalt der TubuH mitochondriales lassen sich vorerst nur Vermutungen anstellen. Es erscheint moglich, daB sie eine Fliissigkeit enthalten, die im elektronenoptischen Bild bislang nicht zur Darstel- lung gebracht wurde. Zweifelsohne bewirken aber diese Rohrenstrukturen eine betriichtHche VergroBe- rung der reaktionsfahigen Oberflache innerhalb der Mitochondrien. Wir kommen fur den Aufbau der Mitochondrien von Paramecium zu einem Schema wie nach Fig. 1, das gleichzeitig die Unterschiede im Aufbau von Paramecium-Mitochondrien einerseits und dem bis- lang bekanntgewordenen Strukturprinzip von Sau- ger-Mitochondrien (6, 8, II) andererseits deutlich macht. Da man bei Paramecium unter otpimalen Kultur- bedingungen bis zu 4Zellteilungen pro Tag erzeugen kann, wurden nun Tiere mit hoher Teilungsrate iiberpriift in der Hoffnung, Aufschliisse iiber die Entstehung oder Vermehrung der Mitochondrien zu gewinnen. Es lieB sich eine ziemlich geschlossene Entwicklungsreihe von Mitochondrien nachweisen. Aus kleinen cytoplasmatischen Blaschen entstehen groBere Gebilde, die wir als ,,Promitochondrien" bezeichnen konnen. Diese Promitochondrien entwickeln Mikrotubuli, die durch weiteres Wachstum in normal groBe Tubuli mitochondriales ubergehen, wobei gleich- zeitig das Mitochondrium selbst heranwiichst. Diese Entwicklungsstadien finden sich nur im Endoplasma. Uber die Natur der cytoplasmatischen Blaschen kann noch keine sichere Aussage gemacht werden. Mit Sicherheit sind die cytoplasmatischen Blaschen nicht mit ,,Mikrosomen" oder „Biosomen" zu iden- tifizieren (nach Eichenberger (3) sollen die Mito- chondrien durch Umwandlung von Mikrosomen ent- stehen). Unsere heutigen Forschungsmittel erlauben noch nicht, die etwa 700 A groBen ,, cytoplasmatischen Blaschen", aus denen die Mitochondrien entstehen, naher zu charakterisieren. Die Frage nach der Konti- nuitat und Autonomic des Chondrioms kann noch nicht endgultig beantwortet werden. Die Untersuchung von Zellen mit hoher Teilungs- rate erbrachte auch weitere Befunde iiber den Ver- mehrungsmodus der Mitochondrien. In den licht- mikroskopischen Untersuchungen hatten Danneel und Giittes (2) sowie Steiner (13) bereits eine Tei- lungsfahigkeit der Mitochondrien verschiedener Zel- len wahrscheinlich gemacht. Wir finden bei Parame- cium Stadien, die sich sowohl als Querteilungen als auch eine Art Knospung interpretieren lassen. Von einer gleichmaBigen Verteilung der Substanz des Mitochondriums kann bei solchen Durchschnii- rungen keine Rede sein. Vielmehr lassen sich alle Zwischenformen zwischen Teilung und Abschniirung verfolgen. Von besonderem Interesse ist eine Erscheinung, die sich wiederum an Zellen mit hoher Teilungsrate besonders deutlich zeigt. Man kann an zahlreichen Mitochondrien verfolgen, daB sich die begrenzende Doppelmembran ofTnet und Tubuli ins Cytoplasma entlaBt. Dieser Befund diirfte von Bedeutung fur den Funktionsmechanismus der Mitochondrien sein, wenn sich zeigen lieBe, daB es sich hier wirklich urn eine Art sekretorischer Tiitigkeit der Mitochondrien handelt. Die Deutung muBte mit besonderer Vorsicht ge- schehen, da uns ja das Elektronenmikroskop immer nur einzelne Zustandsbilder, nicht aber den konti- nuierlichen Ablauf eines Vorganges in der Zelle zu beobachten gestattet. An Hand eines groBen Untersuchungsmaterials lieB sich erharten, daB man wirklich von einer Abscheidung geformter Sekrete bei den Mitochon- drien sprechen kann. Die Richtigkeit dieser AufTas- sung wird durch folgende Befunde bestiitigt: 1) Der Aquivalenzgrad der elektronenoptischen Bil- der zum lebenden Zustand kann aus der Erfah- rung zahlreicher Reihenversuche mit verschiede- nen Fixierern, pH-Werten und Kontrastierungen als gut bezeichnet werden. 2) im Cytoplasma lassen sich auch in der niiheren Umgebung von Mitochondrien Strukturen nach- weisen, die den Tubuli mitochondriales auBer- ordentlich iihnlich sind. 3) Eine Sekretion von Tubuli konnte nur selten an Zellen mit geringer Teilungsrate beobachtet wer- den, sie tritt gehiiuft bei Zellen mit hoher Teilungs- rate auf, bei denen der StofTwechsel erhoht ist. 4) Und letztlich liiBt sich an entleerten Mitochon- Changes in ihc Ciliary Epithelium 139 -Endoplosma VKl^ Ektoplasma - Fig. 2. Schematische Darstellung der Entstehung (A), der Vermelirung (B) und der Abscheidung (C) geformter Sekrete der Mitochondrien von Paramecium caiidati/m. drien eine Regeneration der Tubuli verfolgen, und zwar bilden sich in diesen entleerten Mitochon- drien zunachst wie in den Promitochondrien Mi- krotubuli, die wieder zu normalen Tubuli mito- chondriales heranwachsen. Wir haben hier also eine echte Regeneration ein- zelner Mitochondrien vor uns. Die Ergebnisse iiber Entstehung, Vermehrung und einen Funktionsmechanismus der Mitochondrien lassen sich schematisch wie in Fig. 2 zusammen- fassen. Ein Abdruck der Belegaufnahmen war im Rahmen dieses Tagungsberichtes nicht moglich. Ihre VerolTentlichung erfolgt an anderer Stelle. Die Untersuchungen wurden durch cine Sachbcihilfe der Deutschen F-orschungsgemcinschafl ermoglicht. Frl. Chr. Brodt dankc ich wiederum fur wcrtvolle technische Assistenz. LiTF.RATUR 1. Ai.TMANN, H. W. u. a.. Das Cytoplasma. In Handbuch der Allgemcinen Palhologie, II, 1. Berlin (Goltingcn)- Heidelberg, 1955. 2. Dannfel, R. und GGttes, E., Naturwissenschaften 38, 5, 117 (1951). 3. EiCHENBERGER, M., E.xptl. Cell Research 4, 275 (1953). 4. Frey-Wyssling, A., Protopiasmatologia II, A 2. Wien, 1955. 5. LiNonrRG, O. und Ernsti r, L., Protopiasmatologia III, A4. Wien, 1954. 6. Palade, G. E., /. Histocheni. Cytodiem. 1, 188 (1953). 7. Powers, E. L., Emret, C. F., und Roth, L. E., Biol. Bull. 108, 2, 182 (1955). 8. Rhodin, J., Correlation of Ultrastructural Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Karolinska Institutct, Stockholm, 1954. 9. Sedar, a. W. und Portir, K. R., /. Biophys. Biochem. Cylol. 1, 583 (1955). 10. Sjostrand, F. S., Nature 171, 30 (1953). 11. — International Union of Biological Sciences, Series B 21. Groningen, 1954. 12. Sjostrand, F. S. and Rhodin, J., Expil. Cell Research 4, 426 (1953). 13. Steiner, M., Naturwissenschaften 41, H. 8, 191 (1954). 14. Wohlfarth-Bottermann, K. E., Z. Naturforsch. (1957, im Druck). 15. — • Diese Proceedings, S. 124. Changes in the UUrastructure of the Ciliary Epithelium during Inhibition of the Secretion of Aqueous Humour A. Holmberg The Lahonitory for Biological Ultraslruclurc Research of tite Department of Aiuiloniy, Kaioliiisi\a Inslitiilet, Sloeklwlni It has been shown by many investigators (I, 2, 4, 6, 7, 8, 9, 10, 11) that the epithelium of the ciliary body plays an important role in the formation of aqueous humour, partly as a barrier between blood and aqueous humour, and partly as an actively secret- ing epithelium. This investigation coinprises only the innermost, non-pigmented layer of the ciliary epithe- lium. Rabbits were used as experimental animals. The tissue has been fixed in vivo by injection of 1 "„ osmic acid in veronal-acetate buffer into the posterior chamber of the eye. The animals were kept under general anestesia (nembutal). After 15-20 minutes part of the ciliary body was removed and the fixation continued in \itri> for further Zl-A hours. The animals, treated with Diamox, were partly nephrectomized, partly not. The nephrccto- mized animals were given 10 mg Diamox per kg body weight intravenously, the others 100 mg Diamox per kg body weight. As control animals, besides the normal ones, animals were used which instead of Diamox were given Diazil in equal amounts. The fixation was carried out after different intervals after the injection. The normal stnietiire. — In this report brief men- tion will be made of some of the cell components which seem to undergo changes after inhibition of the secretion. Most characteristic for the non-pigmented epithe- lium of the ciliary body arc the ;'>-cytoniemhranes. (hg. 1). They stand out as iriplc-laycred membranes with two osmiophilic components, each about 40 A thick, and one central osmiophobic layer about 70 A thick. At the cell surface one can clearly see how the membranes are continuous with the cell membrane. Thus the /^-cytomcmbranes may be interpreted as folds of the cell membrane, but it seems probable 140 A. HOLMBERG ■*- //- Fig. 1. /?-cytomembranes in the non-pigmented epithelium of the cihary body. Magnification 70,000. that the central, osmiophobic layer, because of its regular thickness, is not a simple extracellular tissue space. The membranes are localized only to the apical parts of the cell, and do not extend deeper into the cell than to the apical part of the nucleus. Outside the cell membrane a basement membrane is observed 250-350 A thick, separated from the cell membrane by an interspace 400-700 A wide. The mitochondria are relatively small with a mean width of only 0. 1 8 //. Most of them are built up in the common way, with the inner double membranes oriented mainly perpendicularly to the outer double membrane. In some cases, however, the inner mem- branes extend parallel to the long axis of the mito- chondrion and along its whole length. In none of these cases have any connections between the inner and outer membranes been seen. The Golgi apparatus (fig. 2) is without exception encountered in the basal part of the cells. In each cell one can observe 1-3 Golgi units, each consisting of the well-known components: 2-3 rows of double membranes, 7-cytomembranes, a few larger vacuoles, and several small vesicles. Each single osmiophilic component of the double membranes is about 50 A thick. Both the vacuoles and the vesicles are bounded by a single membrane, about 50 A thick. Changes in the idtrastructure after injection of Diamox. — Diamox produces changes in the non- pigmented epithelium in at least three of its compo- nents: the cytoplasm, the mitochondria and the Golgi apparatus. The most marked changes in the cytoplasm is seen within 30 minutes after injection of Diamox. The cells are filled with small vesicles, in the deeper layers mostly very regularly spread, but close to the surface often congregated in large groups. It is striking how the vesicles in these groups are arranged in rows or concentric circles, resembling the arrangement of the /i-cytomembranes. After more than 30 minutes after injection of Diamox the amount of vesicles decreases and at intervals longer than 60 minutes they seem to be practically absent from the cytoplasm. Under the influence of Diamox there is a pro- nounced increase in the width of the /7//7()f//o//fl'r/a from the normal 0.18 /< to about 0.25 /<. This increase is due to a real increase of the volume of the ground substance of the mitochondria and not to such a swelling as is characteristic for the post-mortem changes. The alteration of the mitochondria appears very soon; it is seen already within 15 minutes after injection of Diamox and persists for a long time. Thus 2 hours after injection the width of the mito- chondria is still about 0.25 //. As far as can be seen Changes in l/.c Ciliary Ep it helium 141 Fig. 2. Golgi apparatus from a normal animal. One Golgi unit with several membranes, a tew large vacuoles and some small vesicles. Magnification 53,000. Table 1 . The vacuolization of the Golgi apparatus after Diamox. Number Experiment Number of Golgi Number of vacu- of regions oles per Golgi 'mals observed region per section per cell ani Normal animals; no 12 nephrectomy Control animals after 3 nephrectomy (Diazil) 30 min. after 10 mg 3 Diamox per kg body weight; nephrectomy 60 min. after 10 mg 3 Diamox per kg body weight; nephrectomy 120 min. after 10 mg 3 Diamox per kg body weight 30 min. after 100 mg 3 Diamox per kg body weight; no nephrectomy 31 78±10 7 79 + 1 1 4 86±18 7 93±12 13 193-12 11 173 ■ 11 4 228 z 40 6 184+13 4 142+ 7 5 155-15 6 79 • 10 6 116:* 15 9 118r 7 6 195r23 6 174+16 4 198 + 25 86 r 4 198±16 160±12 104 ■ i: 189: now, there are no other changes in the mitochondria, but the analysis is not yet finished. In the Golgi apparatus there is a great increase in the amount of the small vesicles (fig. 3). A quantita- tive analysis has been performed by counting all vesicles seen in the Golgi apparatus on one section through a cell. The result is presented in Table 1. This informs of the fact that the changes in the Golgi apparatus appear very soon and that the number of vesicles decreases in animals killed at longer intervals after the injection of Diamox. Thus, after two hours the amount of vesicles is about the same as in the control animals. Another change in the Golgi apparatus concerns the Golgi membranes, which normallv lie in 2-3 rows with 4-5 membrane pairs in each, sometimes split up into several units with onl\ 1 3 mem- brane pairs (fig. 3). Finally it must be pointed out that in the control animals no such variations have been observed in the cytoplasm, the mitochondria or in the Golgi appara- tus. Discussion. — Concerning the cell structure under normal conditions it should onl\ be emphasized that the organisation of the apical parts of the cells, with the ;J-cytomembranes and the basement mem- 142 A. HOLMBERG ^ ■V ■■'■ -*-■ t. 1^ ~m*K Fig. 3. Golgi apparatus from an animal fixed 15 minutes after injection of 10 mg Diamox per kg body weight. Several Golgi units (arrows) with numerous small vesicles. Magnification 50 000. brane outside the cell surface membrane, is also a characteristic of the epithelium of the proximal convoluted tubule of the kidney (12, 15). Comparing these epithelia from a functional point of view one finds that they are similar to a certain extent. The mode of action of Diamox has not yet been fully elucidated, but it is reasonable to assume that its inhibition of carbonic anhydrase will interfere with the formation of aqueous humour. Tonographic studies (3) have also shown that the inflow of aqueous humour into the eye is inhibited with about 60 "„. Are the changes seen in the ciliary epithelium in fact due to the effect of Diamox and are these changes responsible for the inhibition of the secre- tion? After nephrectomy the diuretic effect of Diamox is absent. The use of a control substance (Diazil), which is physically and chemically similar to Diamox but is lacking any effect on the intraocular tension, other secondary effects can be excluded. Therefore it seems highly problable that the changes seen actually are due to the inhibitory effect of Diamox on the secretion. As far as 1 know, no functional changes in the Golgi apparatus have been observed earlier by means of electron microscopy. It is beyond all doubt that there is a marked alteration in the Golgi apparatus of the ciliary epithelium concerning both the amount of vesicles and the arrangement of the membranes. The interpretation of these changes is somewhat difficult at this stage of the investigation, but they may indicate that the Golgi apparatus really takes part in the secretion according to the hypothesis presented in light microscopic investigations. The fact that in the earlier stages of Diamox inhibition a high amount of small vesicles accumu- lates in the cytoplasm of the cells can be explained by a sudden blockage of the outflow of aqueous humor from the cell surface to the posterior cham- ber and its accumulation in vesicles. From these arguments it follows that the changes in the cyto- plasm seen after Diamox injection may be secondary to the inhibition of the secretion. References 1. Barany, E., Acta Phys. Scaiul. 13, 55 (1947). 2. — ibicL 13, 81 (1947). 3. Becker, B., Amer. J. Ophlh. 11, 599 (1955). 4. Davson et al., J. Physiol. 108, 203 (1949). 5. Duke-Elder, S. and Davson, H., B. J. Opiuh. 33, 452 (1949). 6. Friedenwald, J., Arch, of Ophth. 210, 761 (1938). 7. — B. J. Ophlh. 28, 503 (1944). L'appareil de Golgi des Protozoaires 143 8. — Amer. J. Ophtli. 32, 9 (1949). 9. — ihkl. 39, 59 (1955). 10. KiNSEY, V. E. and Barany, E., Atm-r. J. Ophth. 32. 189 (1949). 11. Palm, E., Acta Ophih. 24, 189 (1946). 12. Rhodin, J., Correlation of Ultraslructiiral Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Stockholm, 1954. 13. Ross, E., J. Phys. 114, 229 (1951). 14. Sjo.srRANi:), F. S. and Hanzon, V., E.xptl. Cell Research 7, 415 (1954). 15. Sjosirani), F. S. and Rhodin, .1., E.\pll. Cell Research 4, 426 (1953). L'appareil de Golgi des Protozoaires et son Lillra-structure comparee a celle des Metazoaires P. -P. Grasse Lcihoratoire de Microscopic Elcctroiiiqiic appliqiicc ci la Biologic, 105 Boulevard Raspail, Paris L'appareil de Golgi des Protozoaires a ete Tobjet de recherches approfondies mais des incertitudes quant a sa structure et a ses inomologies, ont persiste jusqu'a I'epoque, toute recente, oti le microscope electronique a permis de reconnaitre la Constance structurale de cet organite. Des 1925, Duboscq et Grasse ont affirme que l'appareil parabasal des Flagelles (organite decouvert par Janicki en 1911) est I'homologue de I'idiozome des elements germinaux males des Metazoaires. Dans diverses publications, cette interpretation fut soumise au controle des faits (2, 3, 4). C'est au dictyosome, constituant normal de l'appareil de Golgi que Duboscq et Grasse ont finalement assimile l'appareil parabasal des Flagelles (Zooflagelles, Choanoflagelles, Euglenes, Cryptomonadines, etc.) et les corps osmioreducteurs de certains Rhizopodes (Enta/noeha) et Sporozoaires (Gregarines et Cocci- dies). Dans une note recente (5), j'ai eu I'occasion de montrer par la microscopic electronique la quasi- identite structurale de l'appareil de CJolgi des Zoo- flagelles et des Metazoaires. L'objet de la prcscnte communication est de preciser cette identite et d'en montrer la Constance. Quelle que soit I'especc de Zooflagelles a laquelle on s'adresse, les resultats sont les mcmes : l'appareil parabasal qu'il soit simple (Trimitiis, Foainci) ou complexe (Joenia) ou multiple (Trichonynip/ia, Spirotrichonyinpha) a la meme structure fondamen- tale. II se compose d'un nombre, variable scion les especes, de saccules tres aplatis, empiles les uns sur les autres (14 chez Trimitus, une trentaine chez Joenia annectens). Les saccules ont une parol forte- ment osmiophile et un contenu qui Test tres peu. Leurs bords sont generalement gonfles. Tout autour de la pile de saccules s'ohserve un essaim de vesicu- les de taille reguliere (100 m// env.) a parol osmio- Fig. 1. Foaina grassii, parabasal dont certains saccules ont les bords fortement gonfles. Grandissemenl direct 20.000; photographique 42.000. Fig. 2. Dictyosome dun o\oc\le d' Helix ponialia monlrant Tanalogie de structure avec Ic parabasal de Foaina. Grandis- semenl direct 17.300; photographique 37.000. 144 p. p. GRASSE Fig. 3. « Parabasalie » de Joenia annectens; Remarquer la coupe transversale (noire et compacte) du filament parabasal F. Grandissement direct 20.000; photographique 54.000. phile (moins que celle des saccules) et a contenu grisatre, ce sont les vesicides osniiop/iiles (fig. 1-2). Dans les especes oii un filament particulier (fila- ment parabasal lequel est engendre par le centrosome) relie I'organite au centrosome, les saccules s'orientent par rapport a celui-ci de telle sorte que le filament s'appuie sur le saccule superieur de la pile, lequel saccule est plus ou moins creuse en une gouttiere qui regoit le filament (fig. 4). Aucune autre liaison n'a ete observee entre les saccules et le filament. De toute evidence, les saccules forment la subs- tance chromophile de Tappareil parabasal; nous ne pensons pas que les vesicules osmiophiles etant donne leur petitesse soient visibles en microscopic optique, sauf lorsqu'elles sont extremement abon- dantes, ce qui arrive. Quant a la substance chromophobe, si nette chez les Protozoaires, il n'est pas possible de la reconnaitre dans tous les cas; nous croyons pouvoir affirmer qu'elle n'est pas un constituant constant et fonda- mental de I'appareil de Golgi. L'examen d'un nombre eleve de preparations conduit a Tinterpretation que voici : Les saccules osmiophiles les plus distaux, c'est a dire ceux qui, conventionnellement sont le plus eloignes du filament parabasal, se gonflent, tan- dis que leurs contours perdent toute nettete; ils deviennent une grande vesicule claire (fig. 3), laquelle nous parait exactement correspondre au lisere ou a la gouttelette chromophobe montre, sous le micro- scope optique, par le dictyosome, apres son impre- gnation osmique. Cette « secretion » ne nous a pas paru etre constante, ce qui explique Tabsence de substance chromophobe dans certains dictyosomes vus au microscope optique. Dans quelques preparations, concernant surtout des Foaina, on observe des vesicules non distales, mais a position intercalaire, dont la region marginale est extremement gonflee en bulle claire qui. liberee se confond, peut-etre, avec de la substance chromo- phobe (fig. 1). Fig. 4. Dictyosome ou parabasal de Tric/ionympha agilis (symbiote de Reticiilitemies liicifiigiis), montrant la produc- tion par perlage lateral, des vesicules osmiophiles, et de la substance chromophobe par « vesiculisation » totale des saccules distaux. Grandissement direct 20.000; photographi- que 43.000. Microscope electronique RCA EMU 3A. L'identite de structure de I'appareil parabasal et des dictyosomes de Metazoaires, parait totale : memes lamelles osmiophiles plates et empilees les unes sur les autres, meme mode de production de vesicules osmiophiles. Les difl"erences tiennent 1° au rapport de conti- guite de certains parabasaux avec un derive centro- somien, le filament parabasal, mais les dictyosomes eux-memes, sont tres souvent attires par le centro- some; 2° a la production de substance chromophobe, surtout par la vesiculisation totale des saccules dis- taux, laquelle n'a pu etre vue en toute certitude dans les dictyosomes'. La ressemblance va si loin que, dans un lot de photographies ou parabasaux et dictyosomes sont melanges, le biologiste non prevenu n'evite pas les confusions. L'appareil de Golgi, qu'il appartienne a un Proto- zoaire ou a un Metazoaire, presente les memes caracteristiques, a savoir : 1° une pile de saccules osmiophiles. 2° un nuage de vesicules osmiophiles, incontes- tablement issues des saccules. La substance chromophobe de tous derive pro- bablement de la vesiculisation totale de certains saccules, mais nous n'en avons la preuve que pour les Protozoaires. Au total, l'appareil de Golgi — aussi bien chez les Metazoaires que chez les Protozoaires — est un con- stituant cytoplasmique caracterise par sa structure et par son activite secretoire (vesicules osmiophiles, 1 Sur un appareil de Golgi aussi puissamment secreteur que celui des glandes multifides de VHelix pomatia, il ne nous a pas ete possible de suivre la transformation des saccules en produit de secretion (substance chromophobe), lequel se voit fort bien apres impregnation osmique sous le microscope optique. Morphology of the Golgi Apparatus in the Common Limpet 145 substance chromophobe), tout a fait distinct du chondriome et du « vacuome ». La question reste posee de savoir si rappareil de Golgi, organite constant de la cellule des Metazoai- res, existe chez tous les Protozoa ires. Nous inclinons a penser qu'il n'est pas present chez tous; nous n"avons pu le decouvrir chez des Flagellcs tcls que les Oxymonadidae. Mais Temploi systematique du microscope electronique le revelera peut-ctre dans des organismes ou il avait echappe au microscope optique, tel est le cas des Infusoires cilies, ou une de nos eleves, Madame Noirot, vicnt de le decouvrir chez un Entodinimn (Inl'usoire Ophryoscolecidae) (observation inedite), avec la meme structure qu'un parabasal, mais rclatisL-mcnt plus petit. BiBLIOGRAPHIE 1. DuDoscy, O. et Grassi'. P. -P., Conipt. rend. acad. sci., Paris, 180, 477 (1925). 2. — Arch. -ool. expll. et gen. 73, 381 (1933). 3. Grasse, P.-P., Conipt. rend. soc. tiiol., Paris 93, 1097 (1925). 4. — Arch. zool. e.xptl. et gen. 65, 345 (1926). 5. ~- Conipt. rend. acad. .sci., Paris, 242, 858 (1955). 6. Grasse, P.-P. et Carasso, N., Natitre 179, 31 (1956). 7. Janicki, C, Biol. Zentr. 31, 321 (19 11). The Morphology of the Golgi Apparatus in Neurones and Epithelial Cells of the Common Limpet Patella vulgata D. Lacy Depart iiu'iit of Zoology and Comparative .Anatomy, St. Bartholomew's Medical College, London Widely different views are held about the mor- phology of the Golgi apparatus in cells of molluscs (2. 11, 8). The following is a summary of work carried out to determine the morphology of the Golgi apparatus in neurones and epithelial cells of the common limpet. Patella vidgata. The neurones studied were from the pedal, pleural and visceral ganglia. The epithelial tissue lies adjacent to the ganglia. As revealed by Kolatchev's method (5) the Golgi apparatus in neurones of the limpet consists basi- cally of a system of black filaments. In most cells the filaments anastomose to form networks of \arious complexity (hg. 1 ). In relatively few cells the\ remain discrete. Associated with some of the networks and some of the discrete filaments is a weakly osmio- philic substance. In small cells (about 5 // in dia- meter) the networks arc confined to a small region Fig. 1. Photomicrograph of a neurone treated according to Kolatchev's tech- nique. Magnification 1700. Fig. 2. Electron micrograph of a very small and possibly young neurone about 5 fi in diameter. The Golgi apparatus is a small compact inclusion lying next to the nucleus. This may be a primiti\e or embryonic condition. Magni- fication 11,000. Fig. 3. Electron micrograph showing the fine structure of a Golgi filament in a neurone. Magnification 45,000. 10-568204 Electron Microscop'j 146 D. LACY Fig. 4. Electron miciograpli of a small part of the highly differentiated Golgi zone of a neurone. The line A-B in fig. 1 passes throughasimilar region. Magnification 21,000. of the cell body lying next to the nucleus. In larger neurones they extend from the nucleus (which is excentrically placed) into much of the cell body. Additional details of the structure of the Golgi filaments are revealed when cells are treated by Kolatchev"s method and examined by electron mi- croscopy. The Golgi apparatus is found to consist of chromophilic and chromophobic components as in vertebrate cells (3, 6). Examination of neurones fixed in buffered osmic acid (7) shows that the Golgi filaments are located in a highly diflFerentiated zone of the cell body (fig. 4). The filaments consist of paired anastomosing mem- branes which enclose two substances; a dense material lying within narrow folds of the membranes, and an osmiophobic substance lying within small dilations of the membranes (fig. 3). Lying next to the filaments and scattered about much of the differentiated zone are numerous small (Golgi) vesicles. Some discrete Golgi vacuoles are also present. Comparison of material treated by Kolatchev's method with that fixed in buff"ered osmic acid shows that the chromophilic component of the Golgi apparatus corresponds to the membranes together with the dense substance they enclose. The chromo- phobic component of the Golgi apparatus corre- sponds to the osmiophobic substance lying within dilations of the membrane. Finally the weakly osmiophilic substance (seen in Kolatchev's prepara- tions examined by light microscopy) arises from the impregnation of the Golgi vesicles. The weakly osmiophilic substance is probably the "archoplasm" of previous workers (2). Filaments are not seen in very small neurones fixed in buffered osmic acid. However, these cells contain a compact organelle, lying close to the nucleus, with an ultra-structure similar to that described for the filaments (fig. 2). Kolatchev's method also reveals certain spheroidal bodies. These are light to dark brown in colour and •#..^# 3 Fig. 5. Electron micrograph showing the Golgi apparatus in an epithelial cell. The apparatus is more compact than in neurones but has the same basic ultra-structure (compare with fig. 3). Magnification 41,000. Figs. 2-5 are micrographs of tissue fixed in buffered osmic acid, pH 7.4. G.A. = Golgi apparatus G.f. = Golgi filament G.m. = Golgi membrane G.v. = Golgi "vacuole" (osmiophobic substance lying within a dilation of the folded Golgi membrane) G.vs =^= Golgi vesicles G.Z = Golgi zone M = Mitochondrion N = Nucleus S.b = Spheroidal body ("lipochondrion") The Cytoplasm of the Sea Urchin Egg 147 are located around the Golgi zone. They are revealed by Baker's Sudan black method ( 1 ) and correspond to the "lipochondria" of some workers (8). Epithelial cells contain an organelle, lying to one side of the nucleus, with an ultra-structure similar to that described for neurones (tig. 5). This organelle is identified as the Golgi apparatus. When the above results are considered alongside previous work (4, 3, 6, 9, 10) it would appear that the Golgi apparatus is a distinct type of organelle, morphologically well defined, which exists in the cells of animals belonging to such diverse classes as mammals and gastropods. In all probability, there- fore, it is a universal constituent of animal cells. I am greatly indebted to Dr. V. E. Cosslett for the use of the Siemens Elmiskop I at the Cavendish Laboratory, Cambridge, and to Mr. R. Home of that Laboratory for his most skillful assistance in the examination of speci- mens. Reference.s 1. Bakir, J. R., Qiiort. J. Microscop. Sci. 85, 1 (1949). 2. Bkamhii I , F. W. R. and Gatenby, J. B., Proc. Roy. Soc. Diihliii 17, 275 (1923). 3. Dai ION, A. J. and Fhlix, M. D., Am. J. Anal. 94. 171 (1954). 4. Gatenbv, J. B., J. Roy. Microscop. Soc. 74, 134 (1954). 5. Gati-nhy, J. B. and Bfams, H. W., The Microtomists' Vadc MccLim. Churchill, London, 1950. 6. Lacy, D. and Challice, C. E., /// Mitochondria and other Cell Inclusions. S.E.B. Symposium, Oxford, 1955. 7. Palade, G. E., J. E.xpll. Med. 95, 285 (1952). 8. RoQUE, A. L., /. Roy. Microscop. Soc. 74, 188 (1954). 9. Sjostrand, F. S. and Hanzon, V., Exptl. Cell Research 7, 415 (1954). 10. , E.xpcrieiilia. 10, 367 (1954). I 1. Thomas, O. L., Quart. J. Microscop. Sci. 88, 445 (1947). Basophilic Structures in the Cytoplasm of the Sea Urchin Egg B. A. Afzelius The Laboratory for Biological Ultrastnicture Research of the Department of Anatomy, Karoliiiska Inslitiitet, and the Wenner-Gren Institute for Experimental Biology, Stockholm The sea urchin egg is an ideal subject for a study of the microscopic and submicroscopic organization of the nucleic acids. The nucleic acid content in the egg is high, giving the cytoplasm an intense baso- philia, especially in the young oocyte (4). The visco- sity of the egg cytoplasm is low, permitting a rather complete stratifying of the cell components within the intact egg by centrifugal methods (9, 10), and thus furnishing a means of comparing the submicro- scopic appearance of cytoplasmic fractions with their light microscopic staining properties. This paper is a preliminary report of an electron microscopical study of some of the nucleic-acid- containing structures in the sea urchin egg. Eggs of the species Echinus esciilentiis, Psainniechiniis tuilia- ris, and Strongylocentrotiis droehachiensis were used. The microscopically detectable basophilic elements were investigated in f^xed material by the Feulgcn tech- nique, methyl green-pyronin staining controlled by ribo- nuclease-treated sections (5), gallocyanin-chromalun staining (7) and other techniques considered to be less specific for nucleic acids. The fixation fluid used for most of the experiments was Carnoy's fluid, as no loss of the nucleic acids occurs following this fixation (17). However, since cytoplasmic fixation in Carnoy's fluid was rather poor, comparison tests were performed on material fixed in osmium tetroxide, formaldehyde, Regaud"s fluid, or Bouin's fluid. Also eggs that were vitally stained with toluidine blue were studied in the living condition and in sectioned material fixed according to the method of Izquierdo (14). The electron microscopical part of the work was performed with the sectioning technique according to Sjostrand (19) after fixation in 1 per cent osmium tetroxide in sea water (2). An RCA EMU 2c electron microscope was used. Light microscopical observations. — Feulgen stai- ning gave a negative result with regard to both the pronucleus of the egg and the different strata of the centrifuged egg. Similarly, there were no detectable methyl green staining of sea urchin eggs with the methyl green-pyronin technique. Staining methods for ribonucleic acid (RNA) loca- lization on the other hand indicated the presence of appreciable amounts of RNA in the egg cytoplasm. When centrifuged eggs were studied, the cortex (which is not affected by centrifugation) and the layers of clear cytoplasm and mitochondria were found to contain the bulk of the ribonucleic acids. Also, in the bottom of the mitochondrial zone, which in the eggs examined was the heaviest one, there were particles with a size ranging from 0.8 to 3 //, which stained readily with the RNA-specific stains. Other particles of similar size and staining characteristics did not move to the centrifugal pole of the eggs but remained apparently attached to the outer surface of the nuclear membrane. Similar results were obtained by vital staining with toluidine blue. The clear la\er, the layer of heavy bodies at the bottom of the egg and the bodies attached to the nuclear membrane all stained intensely, the two layers being sharply delimited. However, the mito- 148 B. A. AFZELIUS Fig. I. The two RNA-containing bodies of microscopic dimensions in the egg cytoplasm can be compared in this picture. A portion of a "yolk nucleus" with a-cytomembranes enclosing a few mitochondria is seen at the right and two triangular heavy bodies to the left of this inclusion. The picture is from the sea urchin species Styungy/ocentroriis droehacliiensis. Magnification ■ 30,000. chondria were not or very faintly stained by this technique. Another cell component that stained with the RNA-specific stains was found occasionally in some oocytes of Swedish sea urchins and regularly in the oocytes of the species Strongylocentrotus droebachien- sis. This particle can be described as an elliptical body consisting of concentric membranes which often enclose a particle resembling a yolk granule or a few mitochondria. The total diameter is 5 /< or less. When centrifuged, these bodies go to the clear layer. They are sometimes called "yolk nuclei", which is not a particularly happy name as this term has been used for different structures in different oocyte types and as they have probably nothing to do with yolk synthesis. Electron microscopical observations. — Two of the RNA-containing structures described above were recognized with certainty in the electron micro- scope on ultra-thin sections; the "yolk nuclei" of the oocytes and the heavy bodies of the mature egg. In the "yolk nucleus", whose membranous struc- ture is vaguely discernible with the light microscope, the individual membranes were seen to have a thickness of 60 A and to be associated with granules approximately 150 A in diameter. The membranes, like those of the exocrine pancreatic cells (20) were apparently arranged in pairs, with the granules located on their outer surfaces. The yolk granules and mitochondria that were typically enclosed by these membranes exhibited no peculiarities in their fine structure. Turning to the mature egg the heavy bodies were examined in sections from the centrifugal pole of a stratified egg. Each heavy body was seen to be a dense aggregation of tiny granules (approximately 150 A) surrounded by a membrane of exactly the same appearance as the nuclear membrane. The shape of the heavy bodies as seen in sections was variable; the type most often observed possessed a regularly triangular outline, but bodies of a lentiform outline were also common, and some, especially the larger ones, appeared polygonal or rounded. The membranes did not seem to enclose the granular mass completely, but left small gaps at the angles. Occasionally particles of the same appearance have been found on the nucleo-cytoplasmic border in mature eggs or in oocytes. They were enclosed by a The Cytoplasm of the Sea Urchin Egg 149 nuclear membrane on both sides or were situated within outpouchings of the nuclear membrane. The RNA-containing clear cytoplasm contained so many different structures that it was impossible to ascertain exactly what structure or structures were responsible for the basophilic staining. There seemed to be no particular structure that was confined to this layer upon centrifugation. Many cell compo- nents, however, were concentrated to this zone, among them the small dense granules with an approxi- mate diameter of 150 A (which normally were more or less uniformly dispersed throughout the cyto- plasm of the egg), and vesicles and membranous systems, some of which were dotted with small gra- nules on their outer surfaces. Discussion. — Quantitative biochemistry has pro- vided data concerning the nucleic acid content of the sea urchin eggs. Perhaps the most remarkable result from such studies is the demonstration that the DNA-content in the egg is approximately 25 times higher than that of the spermatozoon (8, 13). This is interpreted to mean that there may exist a cytoplasmic DNA-store in the egg cell (13, 21; see, however, 8). The RNA content of the egg is approxi- mately 30 times as great as that of the DNA (3, 18). Cytochemical localization of the two nucleic acids has been performed on living eggs by ultra-violet absorption at the wavelength 2537 A by Harvey and Lavin (11, 12). There is good agreement between these absorption pictures and pictures obtained by cytochemical staining for nucleic acids. The centri- fugal pole could not be investigated by the method of Harvey and Lavin because the Arliacia egg contains pigment that moves to the centrifugal pole and ab- sorbs ultra-violet light as well as visible and infra- red. The nucleus of the mature egg is Feulgen-nega- tive (see (15), and for a differing opinion (6)), and showed no detectable absorption in ultra-violet light. In the present investigation the RNA has been localized not only in the clear zone of the stratified egg but also in particles in the centrifugal pole. There are some reasons indicating that these heavy particles are derived from the nucleus. They are occasionally seen within bleb-like outpouchings of the nuclear membrane, and in the cytoplasm they are surrounded by a membrane that very closely resembles the nuclear membrane. Like the nucleolus they exhibit very great density on centrifugation. It is suggested that either fragments from the nucleolus or else other RNA-containing entities within the nucleus are expelled through the nuclear membrane together with portions of the membrane that act as envelopes, it has been shown in an earlier paper that fragments of the nuclear membrane may exist free in the egg cytoplasm (1); they are then often arranged in groups of three to ten membranes approximately parallel to each other. Three important basophilic entities have been recognized in the cytoplasm of the sea urchin egg; heavy bodies, "yolk nuclei", and the RNA-contai- ning component of the clear cytoplasmic layer of the stralilied egg. It is a notable feature that all three entities contain a large number of granules about 150 A in diameter. There is thus nothing that speaks against the assumption that the 150 A granu- les are responsible for the basophilic staining. Pa- lade (16) demonstrated a good correlation between cytoplasmic basophilia and "a small particulate component of the cytoplasm" consisting of dense granules of a diameter ranging between 80 and 300 A (normally 100 to 150 A). In the sea urchin egg most of the dense particles appear free in the cytoplasm, and only in the "yolk nuclei" are they arranged on membranes in a way characteristic of many verte- brate tissue cells, to form a-cytomembranes accor- ding to the terminology of Sjostrand (19). The author wishes to thank Dr. G. Gustafson, Kristi- neberg Zoological Station, and Bestyrer D. Rustad, Trondheinis Biological Station, for help and supply of material; Dr. F. Sjostrand, Associate Professor of the Department of Anatomy, for placing the resources of his laboratory at the author's disposal; and Professor J. Runnstrom, the Wenner-Gren Institute, for his unfail- ing support and interest. Financial support from the Swedish Natural Science Research Council is also grate- fully acknowledged. References 1. Afzelius, B. a.. Exptl. Cell Resecirch 8, 147 (1955). 2. — ibid. 10, 257 (1956). 3. Agrell, I., Conipr. rein/, soe. biol. 144, 1322 (1955). 4. Brachet, J., Embryologie chimique. Paris, 1947. 5. — Quart. J. Microscop. Sci. 94. 1 (1953). 6. Burgos, M. H., Exptl. Celt Research 9, 360 (1955). 7. EiNARSON, L., Exptl. Celt Research 1, 359 (1949). 8. Elson, D. and Chargaff, P., Lxperienria 8. 143 (1952). 9. Harvey, E. B., Biol. Butt. 81, 114 (1941). 10. — The American Arbacia and other Sea Urchins. Princeton, 1956. 11. Harvey, E. B. and Lavin, G. I., Biol. Bull. 86. 163 (1944). 12. — Exptl. Cell Research 2, 393 (1951). 13. HoFF-JoRciFNSFN. E., /// KiTCHiNG, RcccHt Dc\clopment in Cell Physiolog>. London, 1954. 14. IzQuiERDO, L., Compt. rend. soc. biol. 148, 1504 (1954). 15. Marsuak, A. and Marshak., C, Exptl. Celt Research 5, 288 (1953). 16. Palade, G. E., /. Biophys. Biochem. Cytol. 1, 59 (1955). 17. Sandritter, W. and Hartleib, J., Experientia II. 313 (1955). 18. Schmidt, G., Hecht, L., and Thannhauser, S. J.. J. Gen. Physiol, n, 203 (1948). 19. Sjostrand, F. S., ;'// Oster and Pollister, Physical Techniques in Biological Research. New York, 1956. 20. Sjostrand, F. S. and Hanzon, V., Exptl. Cell Research 7, 393 (1954). 21. Zeuthen, S.,Publ.Staz.Zool. ;V«/?o/;23 Suppl., 47 (1951). Functional Changes of the Free Cell Surface Membrane of the Intestinal Absorbing Cell F. S. Sjostrand and H. Zetterqvist The Laboratory for Biological Ultrastriictiire Research of the Department of Anatomy, Karolinsha Institiitet, Stockholm FvER since it was observed, one hundred years ago, that the free border of the intestinal absorbing cell was cross-striated, the question of how absorption takes place has been eagerly discussed. Many authors have considered that the cross-striation is caused by a system of small canals through which particles with a diameter of less than 0.5 micron can pass, while others advanced the opinion that the striation depends upon rodlets or fibres extending from the cytoplasm in order to increase the absorption area of the cell. In the latter case the absorption of fat, for instance, should occur in an aqueous solution after hydrolysis. With the development of electron microscopical techniques and methods of thin sectioning it has been shown in osmium-fixed material that the free border of the cell is built up of small finger-like extensions of the cytoplasm (1, 2, 3, 4). At high resolution it is observed that these extensions are covered by a continuous plasma membrane without pores or interruptions either between the bases of the exten- sions or at their tops (4). The membrane is visualized as two opaque lines separated by a less opaque space, the total thickness of the membrane being about 100 A (4). The structure of this plasma membrane (fig. 1) has now been investigated at absorption of different food stuffs. The experimental animals — adult white mice weighing between 20 and 30 grams — were fed after at least 12 hours starving and 4 hours without water. In this preliminary investigation the following kinds of food were used: as carbohydrate food, crumbs of white bread; as protein, native hen egg- white; and as fat, heavy cream. The animals ate the Figs. 1-2. Cross sections through the brush border of intestinal epithelium of mouse. Fig. 1 . Starving animal. Magnification >< 150,000. Fig. 2. Animal fed with egg white, one hour before killing. Magnification 160,000. The Hepatic Sinusoidal Endothelial Cell 151 carbohydrate food spontaneously while the protein and the fat was given by stomach tube. After differ- ent lengths of time, usually 1 or 2 hours, the ani- mals are decapitated and specimens from the proxi- mal jejunum are taken for fixation in the usual way. The results obtained are based on studies of speci- mens from 9 animals, 3 for each kind of foodstufT. That absorption actually takes place can be ob- served on survey pictures of fat absorption where the cells are crowded with fat droplets. In studying the plasma membrane the earlier known distinct, double-contoured structure can no longer be observed. In most places the membrane now appears as a single opaque layer with dilTuse outlines the thickness of which seems to vary with the substance being absorbed. Thus, at carbohydrate absorption the membrane appears thinner than at fat absorption, while the comparatively thickest structure is obtained at protein absorption (fig. 2), the total thickness in the last case being about 100 A, i.e. the same as in starving animals. No more exact measurements are performed because of the diffuse outlines of the membrane. The results obtained show that absorption from the intestinal lumen occurs through a cell membrane which appears to be able to change its structural organization under different functional states. References 1 . Dalton, a. J., Kahler, H., Strifbich, M. J., and Lloyd, B., J. Null. Cancer Inst. 11, 439 (1950). 2. Grangi K, B. and Baker, R. F., Anat. Record 103, 459 (1949). 3. — ibid. 107, 423 (1950). 4. Zetterqvist, H., The Ultrastructural Organization of the Columnar Absorbing Cells of the Mouse Jejunum. Thesis. Karolinska Institulct. Stockiiolm, 1956. The Hepatic Sinusoidal Endothelial Cell and Its Histological Relationships' H. F. Parks Department of Anatomy, University of Rochester School of Medicine and Dentistry. Rochester, !\'ew Yorlc, and the Lalwratory for Biological Ultrastructiire Research of the Department of Anatomy, Karolinsl\a Institntet, Stoclxholm t/VER since Hoffman and v. Recklinghausen (4) observed the phagocytic character of the endothe- lium of the hepatic sinusoid and v. Kupffer (6) identified his "Sternzellen" as potent phagocytes, this endothelium has been a subject of considerable attention. Failure of silver precipitation methods to demonstrate cell outlines has suggested that the endothelium is either syncytial or discontinuous (2). Finally, the existence of a perisinusoidal space is still being claimed ( I) and denied (3) in contemporary writings. These several points of interest emphasize the need for continued electron-microscopic study of sinusoidal endothelium and its histological rela- tions. This report deals with the external shape of sinu- soidal endothelial cells, their relation to one another and to hepatic cells, and to the physiological implica- tions of these relationships. The observations herein recorded were made as a companion study to an investigation of phagocytosis also reported in this volume (5). Details of tissue prepara- tion may be found in that paper. The endothelial lining of the sinusoid appeared in places as a single sheet of cytoplasm of variable thickness (fig. 1), the thinnest places being less than 200 A thick. In other places it was bilaminar, pre- ^ Partially supported by a grant from the National Science Foundation, Washington, D.C., U.S.A. sumably because of extensive overlapping of the cytoplasm of two neighbouring cells (hg. 2). A few sections were seen in which it was tri-laminar and appeared to consist of numerous overlapping sheet- like processes of cytoplasm (described below). Shape of cells. — Hepatic sinusoidal endothelial cells of the mouse are very complicated in shape. It will be easily appreciated that no exact description of cell shape can be given because of the limitations of interpretation imposed by the thinness of sections suitable for electron microscopic study; however, cer- tain morphological characteristics have been noted. Sinusoidal endothelial cells, at least many of them, are not typical squamous cells like those lining ordi- nary blood capillaries. Irregularity of shape and long cytoplasmic processes are discernible with the light microscope in ordinary histological sections. Finer cytoplasmic extensions have also been seen in this study; sheet-like and trabecular processes. The expression "sheet-like" is intended to describe narrower extensions (I to 3 /< wide and .^^ to I n thick) of the usual thin sheet of cytoplasm surround- ing the nucleus of an endothelial cell, (it is of course possible that they are really tangential sections of the periphery of more or less rounded cells.) Such structures were seen apparently hanging free in the lumen, or beneath the main stratum of endothelial cell cytoplasm lining the sinusoid, or actually con- stituting the endothelial lining in places. 152 H. F. PARKS V, • ?• » ".- r^' k ""94 ^ ^■' -tk ;.*d^ibt^^ -«i^- '"jGytaSh.^vs. Fig. 1. Section through portion of hepatic sinusoid. Erythro- cyte at bottom right. Hepatic cells at left and top. Lumen is lined by a single sheet of cytoplasm resting on hepatic-cell villi extending through perisinusoidal space. Phagocytosed gold particles are seen in membrane-enclosed vacuoles. Magnification 26,000. '"•^i. ■•»&'d A * ft Fig. 2. Section through wall of hepatic sinusoid. Erythrocyte at right. Hepatic cell on left. Endothelium is in two layers. Pseudopod-like processes seen in lumen and between endo- thelial laminae. Magnification -^ 48,000. Trabecular cytoplasmic processes were recogniz- able only in transverse section where they appeared irregularly round, oval, triangular, etc. Some tra- beculae were situated between sheets of cytoplasm I'orming a bilaminar endothelial lining; others pro- jected into the lumen as pseudopods, where some of them lay close to the sinusoidal lining, as though bent downstream by the moving blood (fig. 2). Among the trabeculae located in the lumen, some were lying in correspondingly-shaped grooves in larger masses of endothelial cell cytoplasm so that they did not bulge into the lumen, but presented one surface to it. In some instances the surface depression that lodged a trabecula was more than a groove; the lips of the "groove" were not fused together, but came into contact with one another above the trabecula, thus forming a tunnel through which the trabecula passed. Finally, many sections were seen in which a trabecula was fitted into a canal surrounded by a continuous wall of cytoplasm, the intercellular space appearing as a closed circular cleft. It will be appreciated from this "piston-in-a-cylinder"" arrangement that endo- thelial cells not only have elongate projections but also form internal channels into which such projec- tions can be fitted. Relation of endothelial cell to endothelial cell. — This relationship is easily described for those areas lined by thin sheets of cytoplasm. One process merely overlapped another, and the intercellular space apparently allowed blood plasma free access to the perisinusoidal space, as was evident from the fact that colloid particles passed between over- lapping cells with ease (5). In the vicinity of the main cell body the rela- tionship was sometimes more complicated, and is not yet fully understood. In many cases the body of one cell merely overlapped the attenuated cyto- plasmic extension of another or a thin cytoplasmic process of one cell was fitted into an appropriately shaped recess in another. In some places, however, two closely apposed (parallel) cell membranes ran a complicated, sharply-curving course strongly The Hepatic Sinusoidal Endothelial Cell 153 suggesting the appearance of intimate interdigita- tion of two neighbouring cells. It is not known whether, or to what extent, this relationship exists, because it was occasionally possible to follow a cell membrane far enough to get the impression that the whole formation represented a complicated infolding of the membrane of a single cell. One other type of intercellular relationship has been described above; that of a trabecula from one cell fitting into a tunnel in another cell. Relation of endothelial cell to subjacent structures. — The surface of the hepatic cell was characterized by a large number of simple and branched villous extensions (2) of variable length (usually about 0.15-0.5 //) and about 0.025 /( in diameter. In some sections a number of villi were cut longi- tudinally and seen to extend all the way from the main body of the hepatic cell below to the endothelial cell above, while the spaces between villi contained numerous transversely-sectioned cyto- plasmic processes (fig. 1). Most of these latter proc- esses were considered to be hepatic-cell villi or their branches. Endothelial cells rested directly on this bed of villi: no basement membrane was seen between endothelial cell and hepatic cell. The mechanical attachment of endothelial cell to hepatic cell appears to be effected, at least in part, by an intermingling of villus-like processes of one cell with similar processes from the other. This relationship appeared not to be important in sec- tions through cytoplasmic extensions some distance from the endothelial cell nucleus; in fact only a few cytoplasmic villi other than those of hepatic cells were seen. However, in the vicinity of the endothelial cell nucleus, branched villi projecting into the perisi- nusoidal space from the endothelial cell often ap- peared to be very numerous. Von Kupffer (6) described and illustrated endothelial cell processes passing between two hepatic cells toward the bile canal; a single similar observation has been made in this study, and it appears that this may be an important anchoring mechanism. A special relationship be- tween the reticular (collagenous) fibers in the perisi- nusoidal space and endothelial cells has not yet been observed. An incidental observation on the surface of the hepatic cell next to the perisinusoldal space seems worthy of mention here. Small membrane-enclosed vaculoes have been seen in hepatic cells containing one or more rounded bodies about 350 A in diameter and somewhat denser than the ground cytoplasm. In some instances these vacuoles were seen opening into the perisinusoldal space and apparently dis- charging their content of rounded bodies. Cell membranes. — The plasma membranes of both the endothelial cell and hepatic cell villus have been seen to be double in places, and apparently single in others. The double-membraned condition is not restricted to the luminal surface of the endothelial cell; it has been seen on the basal side also. This study has shown that hepatic sinusoidal endothelial cells of the mouse are most unusual in shape. Nothing has been seen that is inconsistent with the bi/arre pictures published by v. Kupffer (6). The facts that there is extensive overlapping of cyto- plasmic processes and in some cases trabecular proc- esses of one cell pass through tunnels in other cells tend to reconcile the idea of continuity of endothelial lining with the irregularities of cell shape. In the absence of a basement membrane and inter- cellular cement, endothelial cells appear to be anchored in place by complicated relationships of their cytoplasmic processes with other endothelial cells and with subjacent hepatic cells. The following considerations argue that the perisi- nusoldal space is an anatomical entity of physio- logical significance: There is no basement membrane between endothelial lining and hepatic cell. There is no intercellular cement, as manifested by the fact that colloidal particles (and therefore presumably blood plasma) pass from the blood stream between endothelial cells into the space. The hepatic cell surface area is enlarged by villi even in places where the under surfaces of endothelial cytoplasmic sheets appear completely flat (the hepatic cell surface is thus apparently designed to carry on exchange of materials at a much greater rate than that of the cell interposed between it and the blood stream). Small rounded bodies are secreted by the hepatic cell into the fiuid filling the perisinusoldal space. If, as it appears from the above, blood plasma actually comes into direct contact with the hepatic cell, then the question of continuity or discontinuity of endothelial lining decreases in importance. It is a pleasure to express thanks to Dr. F. S. Sjostrand for the use of his laboratory facilities. References 1. Eppinger, H., Die Permeabilitats-Pathoiogie als die Lehre vom Krankheitsbeginn, p. 569. Wicn, 1949. 2. Fawcktt, D., J. Natl. Cancer Inst. 15 (Supplement), 1475-1502(1955). 3. Fellinger, K., Braunsteiner, H., and Pakesch, F., Mien. Klin. Hochschr. 65, 738-740 (1953). 4. Hoffman, F. and von Recklinghausen, F., Zenir. meet. IViss. 5, 481-482 (1867). 5. Parks, H. and Ciiiquoine, A. These Proceedings, p. 155. 6. VON Kupffer, C, Arcli. niikroskop. Aiuii. ii. luUHick- liingsmech. 54, 254-284 (1899). Observations on Early Stages of Phagocytosis of Colloidal Particles by Hepatic Phagocytes of the Mouse^ 1, 2 H. F. Parks and A. D. Chiquoine Department of Anatomy, University of Rochester School of Medicine and Dentistry, Rochester, N.Y., and the Laboratory for Biological Ultrastriictiirc Research Department of Anatomy, Karolinska Institiitet, Stockholm, and Department of Biology, Princeton University, Princeton, N.J. When colloidal suspensions of gold or mercury chloride are injected intravenously in the mouse, microscopically visible inclusions of the colloidal particles very rapidly become visible in hepatic phagocytes. Optically homogeneous aggregates from I // in diameter down to the limits of microscopic visibility are seen in animals killed as early as 15 seconds after injection. Larger spherical inclusions about 1 micron in diameter appear after about 2 minutes and are very numerous by 10 minutes following injection (1). They have been observed previously with the electron microscope and seen to be membrane-enclosed structures (3). The question of how inclusions of phagocytized colloidal particles are formed is presently being investigated, and mor- phological observations on hepatic phagocytes of animals killed during the first 2.5 minutes following injection are reported here. Hepatic tissue was taken from 50 white mice weighing 15 to 20 grams. Twenty-five were killed at intervals of 15 seconds to 2.5 minutes following intravenous injection of colloidal gold or mercury chloride. The others received no experimental treatment. Small strips of liver were removed in a cold room and placed in buffered (pH 7.2) isotonic I "o osmium tetroxide solution, where they were cut into blocks one mm or less in thickness with a stainless steel razor blade. After 15 minutes to 4 hours (usually one hour) in fixative, the tissue was rinsed 10 minutes to I hour in Tyrode's solution, dehydrated in a graded series of alcohols, and embedded in methacrylate (10 or 20 parts methyl to 90 or 80 parts butyl). The polymeriza- tion was catalyzed by 0.4 "^o powdered benzoyl peroxide. A considerable part of this material was discarded be- cause of improper polymerization of methacrylate in the center of the blocks of tissue. The sections were cut on a Sjostrand ultramicrotome and analyzed in an RCA EMU 2c electron microscope. (1) At 15 seconds following injection colloidal particles were seen in several relationships to endo- thelial phagocytes: Particles attached to cell surface. Some cells appar- ently have a "sticky" membrane that immobilizes particles coming in contact with it. These cells were seen with their surface largely covered with particles while neighboring cells showed few or no attached particles (fig. I ). Their surfaces were sometimes very irregular, characterized by numerous small pseudo- pod-like processes that were also covered with ^ Partially supported by the National Science Foundation, Washington, D.C., and the Whitehall Fund. - An earlier stage of this work was reported in an abstract in the Anatomical Record, V, 124 (1956 Meetings of the American Association of Anatomists). particles. Adherence to a cell membrane is of course a necessary condition to phagocytosis, and the rela- tionship here described is considered an early stage of the phagocytosis process. Particles located in depressions of the cell metn- brane. Some sections were seen in which the whole luminal surface of a cell did not show an affinity for particles, but particles were found associated with semicircular (in profile) depressions of the cell membrane. The fact that particles were in relation to the depressed portion but not to other parts of the cell membrane suggests that the depression repre- sents an invagination formed in response to the presence of particles on the membrane. A depression appearing semicircular in profile could be either a groove or a spherical invagination. The latter possi- bility seems likely in this case because circular profiles of membrane-enclosed groups of particles were usually seen in the cytoplasm in the vicinity of the turned-in portion of membrane; these were sections of spherical inclusions or cross-sections of spherical (or cylindrical) invaginations from the surface. This general picture suggests that some phagocytosis takes place by a localized spherical or cylindrical invagination of cell membrane that pinches off the surface membrane to form a spherical inclusion. A miniature example of this type of relationship is seen in fig. 1 , B. Particles in intracellular clefts. In some sections shallow to deep infoldings of cell membrane were seen enclosing narrow cleft-like intracellular spaces (cf. fig. 2). Some contained a large number of particles; others only a few. It cannot be stated whether such clefts pre-existed or were formed by invagination of the cell membrane in response to adherent colloidal particles. Particles in intercellular ( ?) spaces. In many sec- tions a cleft-like space between separate masses of cytoplasm was seen communicating both with the lumen of the sinusoid and the perisinusoidal space. Whether such a cleft was intra- or intercellular in a given instance could not be ascertained from a single section; it might be a cleft between two cytoplasmic processes of the same cell or between two separate cells. At any rate, particles were seen both in these clefts and in the perisinusoidal space. (2) At later stages of the phagocytosis process up to 2.5 minutes, few changes of a qualitative nature were seen. Phagocytes were more conspicuous be- Observations on Early Stages of Phagocytosis 155 ^ 1? B **• *? Fig. 1. Endothelial piiagocyte projecting into lumen of sinus- oid. Numerous gold particles are adhering to its surface 15 seconds after intravenous injection. Two pseudopods are seen in section at lower right. The endothelium on the oppo- site side of the lumen is not entirely devoid of phagocytic capacity: at A a particle has evidently been phagocytizcd, and at B two particles are apparently in process of being engulfed. Magnification 24,000. •^•r-^;"^'-- • ' A r.*.A. Fig. 2. Unusually potent phagocvte 2 minutes after injection of colloidal gold. Lumen on right; perisinusoidal space on left. One large branched cleft filled with gold particles is seen communicating with lumen. At A a gold-containing cleft communicates with the perisinusoidal space. At B a rounded profile probably represents a spherical, membrane- enclosed inclusion. Maunification 32,000. cause of their greater content of particles, many of which were contained in spherical or spheroidal mem- brane-enclosed inclusions of various sizes. In some instances the membrane of an inclusion was seen to be double. The most remarkable observation made on this material was on certain particle-con- taining cleft-like spaces (these may or may not have been present in cells at 15 seconds following injec- tion, but were not seen). These spaces ran a highly irregular, sometimes branching, sometimes exagge- ratedly serpentine course, suggesting that the section was passing through an area of complicated inter- digitation of two neighboring cells. However, in a few instances a strong impression was gained that the whole formation was intracellular; i.e., that the cleft represented an extremely complicated infolding of the surface membrane of a single ceil. Some clefts that were obviously intracellular were seen appar- ently breaking up into small spheroidal inclusions. some of which contained particles. Again, it was impossible to decide whether such clefts, if intracellu- lar, pre-existed or formed in response to the presence of foreign particles. The small amount of particulate material (often in widely separated groups) in some clefts suggested that the clefts had not been formed in response to the presence of the particles. It was further noted that a certain amount of phagocytosis of material in the clefts passing from lumen to perisinusoidal space had taken place, as evidenced by spheroidal inclusions of particles in the vicinity of these clefts. Also, some particles were seen in clefts on the basal side of phagocytes, suggest- ing that phagocytosis of material from the perisinu- soidal space was taking place (tig. 2). In a morphological description of sinusoidal endo- thelial cells (2) an intercellular relationship charac- terized by a trabecular process of one cell passing through a tunnel in another cell is described. The 156 M. S. C. BIRBECK AND E. H. MERCER narrow intercellular cleft between these two masses of cytoplasm, which appears circular or ovoid in cross section, very often contained numerous col- loidal particles within 2 minutes following injection. Since the discovery that very dilute colloidal vital dyes are segregated in highly concentrated spherical inclusions by phagocytes, there has been much specu- lation concerning the manner in which these inclu- sions are formed. It is now possible with the electron microscope to follow the progress of colloidal parti- cles as they enter the phagocyte, and thus gain a clearer idea of the mechanism involved. Though the present study is far from complete, it indicates that particles enter the phagocyte in spheroidal and in cleft-like depressions of the cell membrane that secondarily pinch off from the cell surface to form intracellular inclusions. The clefts, whether they pre- exist or form by invagination in response to the presence of particles, appear to represent the more important form of phagocytosis mechanism (fig. 2). References 1. Parks, H., Amir. Rec. 125, 1-16 (1956). 2. — These Proceedings, p. 152. 3. Parks, H., Peachey, L., and Chiquoine, A., Auat. Rec. 124 (1956) (Abstracts of meetings of American As- sociation of Anatomists). The Role of Cell Membranes in Morphogenesis M. S. C. BiRBECK and E. H. Mercer Chester Beatty Research Institute of Cancer Research: Royal Cancer Hospital, London S. W. 3 The study of morphogenesis by electron microscopy is likely to prove a formidable undertaking. However, the labour can be lessened by the choice of an appro- priate system in which there are present not only various examples of differentiated cells but, at the same time, cells in different stages of development. It is a further advantage to have the different stages distributed in the tissue in an easily recognised man- ner. Systems of cells satisfying these demands are not common. One, which is admirable for the purpose, is the follicle of the growing hair of the mammalian skin. Here we find at the end of a small tube, the outer root sheath, dipping down into the skin, a germinal matrix producing a steady stream of cells which, passing along the tube, differentiates into the six concentric cylinders of cells which form the hair and its enveloping sheaths. In the distance of a few 100 fi of the tube we find cells in the following stages: dividing cells, undifferentiated cells, cells in early differentiation and differentiated cells engaged in synthesis of their characteristic products. It is thus possible in a single electron microscopic section to find examples of cells in all stages of their devel- opment arranged in a linear sequence. Taking advantage of this situation, we have ex- amined in hair foUicles^fixed, embedded and sec- tioned by the now standard procedures — the events associated with early differentiation, which occur in the mid and upper regions of the bulb (see figure 1). By comparing the cells of the undifferentiated matrix with those, a few cell diameters further along the follicle which show definite signs of differentiation, we have been led to suspect that the contacts between the surfaces of the cells play a leading role in mor- phogenesis. (See also reviews (5) and (6).) PRE-KERATINOUS ZONE CONSTRICTION ZONE Fig. 1. Diagrammatic representation of the behaviour of cell membranes and morphogenetic developments in the hair follicle with reference to the hair cuticle. In the outline of the follicle, shown in the centre, the changes in the cu- ticular formation and the several zones of the follicle are indicated. On the L.H.S. is shown a series of cells (A to D) in which cell contacts are shown developing, spreading and deforming the cells. The Role of Cell Membranes in Morphogenesis 157 pNi i 'A Fig. 2. Electron micrograph of tiie mid-bulb region of tiie follicle showing early cell contacts and (probably) contact spread. The cells of the matrix contain many mitochondria, many agranular vesicles and large numbers of the dense particles thought to contain ribonucleic acid (3). There are none of the specialised intracellular inclusions which appear later at higher levels in the follicle. The surfaces of the cells are very convoluted and intercellular gaps are common. Numerous small finger shaped pseudopods project from the surfaces and often penetrate deeply into neighbouring cells. Close contact between the cells is limited to a few small areas where the two plasma membranes arc parallel and separated by a distance of about 120- 150 A. We conclude therefore that the surface mem- branes of the undifTerentiated cells are flexible and in active motion. Their adhesion is small and contact is both temporary and limited in area. The space between the membranes at these "adhesive contacts" is not by any means empty. A material of rather poor electron scattering power is observed both between the closely opposed surfaces and also spread out over the immediately adjacent surfaces. We sup- pose that this material is the cement responsible for the adhesion of the surfaces. Very little can be deduced concerning its chemical nature. The very slight reaction with the osmium fixative suggests that it is not protein or lipid in nature, but that it might be polysaccharide. In view of the important role we shall assign to it in morphogenesis, it is most regrettable that more is not known. When the cells further along the follicle are ex- amined the importance of cell contacts becomes apparent (figure 2). The areas of contact spread and. in a /ipper-like fashion, draw the cells together with the result that intercellular gaps are closed and the surface convolutions are smoothed out (figure I ). This development does not take place uniformly throughout the cell population. Contact spread occurs lirsl in the cylinder of cells, which will become the cuticle of the hair, and follows rapidly in those cells, between the cuticle and the outer sheath, which form the three layers of the inner sheath. However, in the cortex, the central cylinder of cells, contact spread does not occur until much later; the cells remain united only locally and intercellular gaps are common. Contact spread appears to have important mor- phogenetic consequences, some of which can be deduced from the appearance of the cells of the various layers in the mid and upper bulb. The cells of the cuticle assume an important place in all sub- sequent developments because between them adhe- sive contacts appear early, spread more extensively and seem to be stronger. The cuticle stands out from the surrounding cells partly because of the density of its membranes and also because of the cuboidal shape of the cells in longitudinal section. It appears that the surface adhesion between these cells is great enough to actually modify their shape. By drawing the membranes of contiguous cells together in zipper fashion (4) the mass of previously rounded cells is converted progressively into a columnar layer (figure I ), which divides the advancing stream of cells into two domains whose subsequent developments are in striking contrast. Inside the cuticular "barrier" is the cortex where the characteristic product of cell synthesis is fibrous keratin; outside we tind the inner root sheath where amorphous trichohyaline is the typical product. In the cortex contacts are localised and surface activity persists at least as high as the follicular con- striction. The melanocytes, which form the pigment of the hair, are situated largely near the tip of the papilla, and appear able to take advantage of the weak adhesion between the cortical cells to extend their long pigmented processes throughout the cor- tical space. The absence of gaps between the cuticle cells prevents the processes penetrating the cuticle and the inner root sheath beyond. The actual in- corporatiiin of the pigment granules into the cortical cells appears to be a consequence of the continued surface activity of these cells for we can see small pseudopods enveloping the ends of the pigmented processes in a manner suggestive of phagocytosis. Occasionally bundles of granules still enclosed in membranes have been found in cortical cells (I). 158 M. S. C. BIRBECK AND E. H. MERCER Fig. 3. Complex intercellular membranes between the cu- ticle cells. Fig. 4. Complex intercellular membranes between the Henle cells (inner root sheath). The dense, closely adhering membranes of the cuticular cylinder not only appear to form a barrier separating the domain of keratin formation (cortex) from the domain of trichohyaline formation (inner sheath), but appear also to prevent the cuticle itself from acquiring sufficient raw materials. For although syntesis is rapid in the cortex and the inner root sheath, in the cuticle itself, synthesis is delayed until above the constriction of the follicle when a form of amorphous keratin with a high cystine content ap- pears. A discussion of the structure of keratin and tricho- hyaline will be found elsewhere in this volume (2). These two proteins are responsible for the hardened and fibrous texture of the cells of the hair and sheath which develops rapidly above the follicular constric- tion. Simultaneously with the development of these intracellular products the intercellular structures also undergo a remarkable development. Until this level the adhesive contacts have consisted simply of the two plasma membranes and the intervening cement layer of 120-150 A thick. As the cells of the cuticle and sheath fill with protein, the plasma membranes suddenly dilate to a distance of 300-400 A and a series of complex layers form between them (figs. 3 and 4). The nature and function of these intercellular layers are obscure; possibly they serve to hold to- gether the hardened cells. The dense layers (one in the sheath cells and two or three in the cuticle) may be deposits of an electron dense tanning agent in a pervading less dense cement of the type described below. In the cortex the membranes also dilate and finally become cemented together but we have not found a layered structure. This investigation has been supported by grants to the Chester Beatty Research Institute (Institute of Cancer Research: Royal Cancer Hospital) from the British Empire Cancer Campaign, Jane Coffin Childs Memorial Fund for Medical Research, the Anna Fuller Fund, and the National Cancer Institute of the National Institutes of Health, U.S. Public Health Service. The authors are particularly grateful to Mr. K. G. Moreman for supplying the illustrations. References BiRBECK, M. S. C. and Mercer, E. H., E.xptl. Cell Research 10, 505 (1956). — These Proceedings, p. 159. Palade, G. E., /. Biophys. Biocheni. Cytol. 1, 567 (1955). ScHMiTT, F. O., Growth 5, 1 (1940). Waddington, C. H., Principles of Embryology. Allen & Unwin, London, 1956. 6. Weiss, P., /. Embryol. E.xptl. Morphol. 1, 182 (1943). Electron Microscopic, X-Ray and Birefringence Studies on the Proteins of the Hair Follicle M. S. C. BiRBECK and E. H. Mercer Chester Beatty Research Institute, Institute of Cancer Research: Royal Cancer Hospital. London, S. W. 3 (^ells derived from the mammalian epidermis are able to synthesise a variety of materials some of which are fibrous. In the hair follicle (fig. I), for example, two very different fibrous products appear, trichohyaline and keratin. The formation of these substances can be followed in the polarisation and electron microscopes and to a limited extent by means of x-ray diffraction. The hair follicle therefore provides an excellent opportunity for correlating the results of the three techniques. The Proteins of the Hah- Follicle 159 Fig. 1. Diagram of the hair follicle showing the development of the fibrous structure of the cortex. On the right-hand side is shown the rise in birefringence (A/;): in the centre the ,\-ray diffraction patterns at several levels are indicated in circles; on the left-hand side the changes in cell shape. The two birefringent systems are well defined when the follicle is viewed between crossed polaroids and measurements (3) show in the coitex a continuous rise in birefringence at the level of the follicular constriction. In the Henle layer of the sheath the change from the isotropic to the birefringent state occurs in a single sudden step, suggesting a rapid transformation of the cell contents. A slower rise occurs in the Huxley layer and the sheath cuticle. The comparison of precisely the same fields in the polarising microscope and the electron microscope has enabled tlio birefringence to be correlated with structure. The inner root sheath. This correlation is most complete in the Henle layer of the inner root sheath. By examining in the electron microscope the area in which birefringence first appears, we can see in some detail the formation of the fibrous form. The rounded, structureless, dense bodies, which appear lower down in the cytoplasm of the sheath cells and rapidly aggregate to form the large droplets recogni- sable in the light microscope as trichohyaline, are seen transforming into fibrils. The details of the actual transformation are shown in fig. 2. Fine fila- ments (about 100 A diameter) can be seen extending in both axial directions from the lenticular shaped droplets of the amorphous form. The transforma- tion occurs entirely intraccllularly and ditTers from anything previously described, although the forma- tion of fibrils from solutions of corpuscular mole- cules is well known. It would seem that the filaments "crystallise" out on the surfiice of the solid droplets and are pushed away from the surface by their continued growth. The nearest physical analogue would be the growth of single crystals from the molten state. Little is known of the chemical com- position of trichohyaline to assist our speculations and the amounts present in the hair are too small to make it possible to examine it by x-ray difTraction. In the other layers of the sheath (Huxley and sheath cuticle) a similar transformation occurs. However, it takes place more slowly and the amor- phous and fibrous forms can be seen together, in the same cells, for a distance of several hundred microns above the constriction. The cortex. — A very different course of events occurs in the cortex. Here the fibrous form appears di- rectly in the form of fine filaments (60-80 A in width) Fig. 2. Electron micrograph of a portion of a Henle cell of the inner root sheath at the point of transformation of the elongated trichohyaline droplets into filaments. (Longitudinal section.) Magnification 73,000. 160 M. S. C. BIRBECK AND E. H. MERCER Fig. 3. Electron micrograph showing fibrils of keratin appearing in a cortical cell in the upper bulb region (longitudinal section). Magnification 20.000. Fig. 4. A cross-section at high magnification of a fibril showing the component keratin filaments (light) on a dark ground — the cystine-rich matrix. Magnification 150,000. without any evidence of a non-fibrous precursor. Filaments can be detected electron microscopically in the bulb cells at a level below which the bire- fringence is strong enough to be demonstrated. Such filaments are oriented parallel to the fibre from their first appearance. These observations clearly show that the keratin is not synthesised as an amorphous precursor which is converted into a fibrous form by its passage through the narrow neck of the follicle. In the upper bulb, where the rise in birefringence takes place, the cells rapidly fill with filaments (fig. 3) and condense to form the rather definite structures recognisable in the light microscope as fibrils (0.1- 0.2 /< in diameter). At this level there is sufficient material present to enable an x-ray diffraction photo- graph to be made and a typical a-type pattern results (fig. 1). There is therefore little doubt that the long fine filaments are the structures responsible for this x-ray pattern which is of such interest to crystal- lographers (2). Cross-sections of the condensed fibrils (fig. 4), show that the filaments are embedded in a material which, after osmium fixation, has a greater electron scattering power than the filaments them- selves, i.e. the filament sections appear light on a dark ground. The most probable interpretation is that the osmium is here acting as a specific stain for cystine (cysteine) and that the S sites are concentrated in the interfilamentous regions. Since chemical an- alysis of dissolved hair (1) shows the presence of a fibrous component (a) with a low S content and an amorphous component (;') with a higher S content, we suggest that the a-component be identified with the fine filaments and the y with the interfilamentous cement. Fibrous keratin is thus seen to be a complex of "filaments plus matrix" rather than a single entity. An unsolved question is the contribution to the observed birefringence of intrinsic and form factors. Attempts to determine these contributions by the standard method of immersing the hair in a series of liquids is not possible, since these invariably fail to penetrate. Liquids, which do penetrate the hair, either react with it chemically or swell the entire structure. In either case the double refraction falls, usually irreversibly, to a low value. The structure of fine filaments embedded in a highly cross-linked matrix suggested by electron microscopy may explain these results. The hair is certainly a Wiener body, i.e. a system of oriented rodlets embedded in a matrix, probably with ditTerent optical constants, and there is likely to be a form contribution. But, since the matrix is by far the more cross-linked component, suitable imbibition liquids penetrating the matrix alone probably do not exist. This investigation has been supported by grants to the Chester Beatty Research Institute (Institute of Cancer Research: Royal Cancer Hospital) from the British Empire Cancer Campaign, Jane Coffin Childs Memorial Fund for Medical Research, the Anna Fuller Fund, and the National Cancer Institute of the National Institutes of Health, U.S. Public Health Service. The authors are particularly grateful to Mr. K. G. Moreman for supplying the illustrations. References 1. Alexander, P. and Hudson, R. F., Wool, Its Chemistry and Physics. Chapman and Hall, London, 1954. 2. AsTBURY, W. T., Proc. Roy. Soc. B 134, 303 (1947). Review. 3. Mercer, E. H., Biochim. et Biophys. Ada 3, 161 (1949). The Mechanism of Hemolysis Caused by Uhrasonic Irradiation. I W. RoMANOWSKi, A. Feltynowski, and J. Liiwin Human Physiology Depaitnient of the Academy of Medicine, Warsaw, and State Institute of Hygiene, IVarsaw The authors investigated the mechanism of he- molysis caused by the ultrasonic irradiation in several experiments under special conditions, and they have obtained some interesting results. The blood of the dog treated with hcparine and next diluted three times by Tyrode's liquid was irradiated for U min. from a generator of the power of 3 W/cm- and at the frequency of 800 Kc. Then the specimen was formed from the erythrocytes irradiated and from the controls. The specimens were shadowcast with chromium and were observed under the Siegbahn-Schoenander electron miscroscope. All the blood cells from the irradiated specimen were much smaller than the controls and attained ^ of their previous size. The blood cells had no charac- teristic "delta" form in the centre, and their shadows were longer than those of the controls. This indicated that the erythrocytes transformed into a spherical form. We find that some results of Sibuya (2) and those of Lindemann (1) though contrary to their inter- pretation, give evidence that the spherical form is one of the forms in which the erythrocytes change when subjected to ultrasonic treatment, and this change may precede the hemolysis. It is now difficult to decide whether it is caused by the denaturation by the ultrasonic radiation of a characteristic protein in the erythrocyte membrane which gives shape of a "biscuit", or by the change in pH or temperature, or by other factors. When irradiating with a smaller dose, changes similar to those described by Lindemann were observed, i.e. convex and concave deformations on the surface of the blood cell. The appearance of these forms can be explained by a non-uniform denatura- tion of the protein over the whole surface. References 1. Lindemann, B., Arcli. exptl. Put/iol. P/numakol. 209, 44 (1950). 2. Sibuya, after Bergmann, Der Ultraschall, p. 41. VDI- Verlag, Berlin, 1939. The Mechanism of Hemolysis Caused by Ultrasonic Irradiation. II W. RoMANOWSKi and A. Feltynowski Human Physiology Department of the Acadettiy of Medicine, Warsaw, and State Institute of Hygiene, Warsaw By means of an ultrasonic 3 W/cm- generator at the frequency of 800 Kc the authors irradiated mice in a special vessel submerged into another one tilled with oil (cooled with ice and water). The blood specimens were taken from the tail and then investigated under the electron microscope. The irradiation times were 9-1 3 min. (mortal dosage) and 5 min. After irradiating with a mortal dosage nearly all the erythrocytes were greatly changed. Their sizes were much smaller than those of the control ones, and their shapes were also changed. Some smaller or greater convex and concave deformations were observed. After five minutes* irradiation such changes have also been observed, but not on all blood cells. These deformations are thought to be caused by a non-uniform denaturation of the protein on the surface, which gives a "biscuit" form to the erythro- cytes. Similar changes in size and shape of the erythro- cytes were obtained in /// vitro experiments (preceding communication). 1 1 — 568204 Electron Microscopy L'ultrastructure de la membrane nucleaire des ovocytes de I'Araignee {Tegeiuiria domestica Clerk) J. Andre et Ch. Rouiller Laboratoire de Zoo log ie, Facitlte des Sciences, Clermont-Ferrand, et Institiit de Recherches sur Ie Cancer Gnstave Roussy, Villejuif Des les premieres recherches sur Tultrastructure de la membrane nucleaire, par Callan et Tomlin (4), les ovocytes se sont revelees etre un materiel de choix pour cette etude. Depuis, on a etendu a un assez grand nombre de types cellulaires, en les precisant et modifiant, les notions alors acquises, et Ton se represente a Theure actuelle la limite entre noyau et cytoplasme comme une enveloppe a double paroi percee de nombreux pores qui la feraient res- sembler a une ecumoire spherique (voir la biblio- graphie parue jusqu'a ce jour dans Zetterqvist (16)). Toutefois. dans les dernieres publications, qui sont presque simultanees (1,6, 16), il subsiste encore des differences d'interpretation des images observees, et il nous a paru utile de confronter les points de vue de ces auteurs avec les resultats obtenus sur un materiel favorable et nouveau, les ovocytes de I'Araignee des maisons Tegenaria domestica Clerk. Des Araignees fraichement capturees sont rapidement dissequees et des fragments d'ovaire sont fixes dans Ie liquide de Palade au pH 7,5. Apres lavage et deshydrata- tion, ils sont inclus dans du methacrylate de butyle et coupes a riiltra-microtome Servall-Porter. L'observation est faite aux microscopes electroniques Triib-Tiiuber KM4 de la Faculte des Sciences de Clermont-Ferrand, et RCA EMU 2E de Tlnstitut Gustave Roussy de Ville- juif. Les fonuations lamellaires de la membrane. Sur les coupes normales a la surface du noyau, les deux feuillets de la membrane se presentent comme deux lignes paralleles, epaisses de 50 A et distantes de 140 A (fig. 1). Ils paraissent identiques en epaisseur et opacite, contrairement a ce qui a ete souvent note (6, 9). Les cliches de tres bonne resolution per- mettent de penser que chacun d'eux, considere jusqu'ici comme simple, est en realite double, com- pose de deux lames osmiophiles separees par un espace plus clair (fig. 1). Les pores. A interval les relativement rapproches, les deux feuillets se rejoignent et laissent libre un pore. A ces endroits, ils s'ecartent d"abord un peu Tun de Tautre en s'epaississant nettement, et se raccordent (fig. 1 ). Au voisinage des « soudures » ainsi formees se trouvent des petits grains fonces, de sorte que ces regions possedent une forte densite electronique. Le pore ayant un diametre de 350 A et etant partiellement obstrue, la probabilite pour qu'une coupe d'epaisseur minimum 250 A passe par Tespace libre est tres faible. C'est pourquoi la plupart semblent fermes par un bouchon (6), une fine membrane ( 1 ), un diaphragme diffus ( 1 5) (fig. 1 ). Ces barrieres ne sont probablement que des portions de la soudure annulaire plus ou moins entamee par la coupe. Un examen attentif permet de trouver quelques pores dans lesquels aucune membrane transversale n'est visible. La lumiere apparait alors tres etroite (fig. 1). Le miage. S"il ne possede pas de fermeture, le pore est toutefois occupe par un « niiage » d'une substance diffuse qui s'etend des deux cotes de la membrane nucleaire. 11 est plus fonce que la matrix cytoplas- mique, et c'est sa presence qu'ont decele les traces Fig. 1. Coupe normale a la surface du noyau, N = Nucleo- plasme, C Cytoplasme. Les amas granulaires sont: (a) au voisinage de la membrane nucleaire et en relation avec le nuage; (/>) libres dans le cytoplasme; (<) a proximite des lamelles ergastoplasmiques; en (e) une particule dans I'espace perinucleaire. La membrane micleaire des ovocytes de I'Aiaignee 163 Fig. 2. Coupe prcsque tangcnlicllc. Rcmarquer I'apparcncc de eertains anneaux en roue dentee («), deux ccrcles concen- Iriques (/?), cercles de parlicules (c). densitometriques de Kautz et de Marsch (8). Sa morphologie est extremement variable. II est le plus souvent vaguement hemisphcrique ou corique d'un cote comme de Tautre de la membrane. Dans le nucleoplasme, il s"effiloche parfois jusqifa 1500 A. Generalement, il est plus opaque dans une region situee a la vcrticale de la soudure. C'est elle qu'Afze- lius a schematisee sous la forme de margelles de puits bordant le pore, mais ses limites sont extreme- ment diffuses. Toute Taire nuageuse, surtout dans sa portion plus foncee, et aux abords de la soudure, est semee de tres fines particules de taille comprise entre 50 et 100 A. Du cote du cytoplasme, le nuage, apres s'etre retreci, se dilate a nouveau en une sphere richement pourvue en particules identiques dont certaines toute- fois semblent confluer. Plus loin dans le cytoplasme, de tels amas granulaires sont encore visibles, et leurs grains, plus gros, dessinent parfois les rosettes, cercles, spirales connus dans Tergastoplasme (9) (fig. 1 ). Les anneaux. Ces figures ne sont visibles que sur des coupes tangentielles, malheureusement tres ra- res : pour les ovocytes ages dont la vesicule germina- tive est devenue enorme, il n'y en a pas plus de 1 sur 500. Cette rarete explique qu'elles soient longtemps restees inapergues ou que leur interpretation ait ete erronnee. Dans Tovocyte de Tegenaire, les anneaux sont remarquablement nombreux puisqu'ils atteignent une densite de 80 au micron carre, compte tenu de la compression due au rasoir. lis sont disposes, uni- formement sur tout le noyau, en rangecs assez regulieres, alternant generalement d'une rangce a I'autre (fig. 2). Leur diametre externe etant de 850 A, lis couvrent presque la moitie dc la surface du noyau. La distance moyenne entre deux centres voisins est de 1100 A. lis comprennent une couronne foncee large de 250 A entourant un centre clair de 350 A. La couronne possedc une structure complexe, dont rimage varie selon le niveau, ainsi qu'on peut le voir sur une coupe tres legerement oblique, quand on observe les rangs qui se succedent depuis le nucleoplasme jusqu'au cytoplasme. Dans les cas fa\c)rables, on peut meme distinguer que deux cercles concenlriques fonccs la limitcnt (fig. 2). Quoi qu'il en soit, des particules de 50 a 100 A parsement ces dilTerentes images, donnant Ic plus souvent des figu- res en cercles pointillcs ou en roues dentces. Les plus grosses d'entre dies ont souvent Paspect de vesicules creuses, ainsi que Rebhun (11) Ta dcja note, et il est remarquable que ces vesicules s'associent frequem- ment par couples donnant des figures en paires de lunettes. Le centre de Tanneau est en general homogenc, plus ou moins grise, mais on peut y trouver egale- mcnt quelques grains. L'espace perinueleaire. Alors qu'on le pense gene- ralement homogene (10), nous y avons trouve des granulations d'environ 50 A. Certaines d"entre elles sont contournees par les deux feuillets de la mem- brane du noyau, et ce fait porte a penser qu'il ne s'agit pas d"un artefact (fig. 1). La eonespondance entre les pores et les anneaux. Cette question est interprctee de fagon diftcrcntc par les divers auteurs. Pour Watson et pour Hague- nau et Bernhard, I'anneau est la vue en plan de la soudure, tandis que c'est celle du nuas/e fonce pour Afzelius, et Zetterqvist. Or, il convient de remarquer que Topacite du nuage est plus faible que celle de I'anneau, tandis que la soudure, region epaissie de la membrane, est au contraire tres diffusante aux electrons. De plus, la morphologie du nuage est moins uniforme que celle des anneaux, et notem- ment, I'existence occaisonnelle d'un contour net a ces derniers (deux cercles concenlriques) montre Cy/o£>/ajme 0.1 r • /Yoyd u Fig. 3. Schema d"un pore de la membrane nucleaire. 164 F. S. SJOSTRAND AND B. A. AFZELIUS qu'ils possedent une paroi franche. Cette derniere ne peut etre que la soudure car jamais le nuage ne se montre nettement limite. II semble done plus raison- nable d'assimiler Tanneau aux parois du pore pro- prement dit, et si le nuage peut se surajouter eflfective- ment a elles dans Timage definitive, il ne saurait la former a lui seul. Cette interpretation est pleinement confirmee par Texamen des coupes obliques qui ne peuvent s'expliquer que si la structure annulaire est intermembraneuse. La diiplicitc des feiiillets de la membrane nucleaire. La constatation que nous avons faite a ce sujet rejoint celle de Rhodin (13) sur les membranes des mitochondries et celle de Zetterqvist (16) sur les membranes cellulaires. Toutefois, lorsque le feuillet apparait double dans nos micrographies, il est comme dilate et moins contraste (fig. 1). C'est peut-etre un artefact de fixation. C'est peut-etre encore Tamorce d'un processus de delamination (11). Cette observation demande a etre generalisee avant de pouvoir en connaitre I'exacte signification. La generalite des pores. Les pores ont deja ete rencontres dans les noyaux interphasiques de nom- breux types cellulaires chez les Vertebres (6). Chez les Invertebres, ils ont ete rencontres chez un Insecte (2), des Echinodermes (1) et des Protistes (3, 7, 15). Enfin, leur decouverte toute recente dans les gametes d'un Champignon (14) nous laisse a penser que ces structures sont peut-etre un trait morphologique de toute cellule vivante. Le noyau de Tovocyte age de Tegenaire en possede environ 1,5 million. La surface de la lumiere de ces pores represente presque le jg de la surface du noyau. II semble naturel de rechercher leur role apres de Robertis (19) et Watson (15), dans les echanges nucleo-cytoplasmiques. Si cette hypothese etait verifiee, nous aurions dans le nuage entourant le pore et dans llamas granidaire qui le termine, Timage du transfert de substance demontre par voie spectrophotometrique par Cas- persson (5) et son ecole. BiBLIOGRAPHIE 1. Afzelius, B. a., E.xptl. Cell Research 8, 147 (1955). 2. Bahr, G. F. et Beerman, W., Exptl. Cell Research 6, 519 (1954). 3. Bairati, a. et Lehmann, F., Experientia 8, 60 (1952). 4. Callan, H. G. et Tomlin, S. G., Proc. Roy. Soc. B 137, 367 (1950). 5. Caspersson, T. O., Cell Growth and Cell function. A Cytochemical Study. New York, 1950. 6. Haguenau, F. et Bernhard, W., Bull. Cancer 42, 537 (1955). 7. Harris, P. et James, T., Experientia 8, 384 (1952). 8. Kautz et DE Marsch, Exptl. Cell Research S, 394 (\955). 9. Palade, G. E., /. Biophys. Biochem, Cytol. 1, 59(1955). 10. PoLiCARD, A. et Bessis, M., Conipt. rend, acacl. sci. 242, 2496 (1956). 11. Rebhun, L. I., /. Biophys. Biochem. Cytol. 2, 93 (1956). 12. DE Robertis, E., /. Histocheni. Cytochem. 2, 341 (1954). 13. Rhodin, J., Correlation of Ultrastructural Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Stockholm, 1954. 14. TuRiAN, G. et Kellenberger, E., Exptl. Cell Research (sous presse). 1 5. Watson, M. L., /. Biophys. Biochem. Cytol. 1, 257 (1955). 16. Zetterqvist, H., The Ultrastructural Organisation of the Columnar Absorbing Cells of the Mouse Jejunum. Stockholm, 1956. Electron Microscopy on Grasshopper Spermati(ds F. S. Sjostrand and B. A. Afzelius The Laboratory for Biological (Jltrastriictiire Research of the Department of Anatomy, Karolinska Institutet, Stockholm Spermatogenesis in the grasshopper takes place within the ducts of the testis. The consecutive stages follow each other in a regular order with the imma- ture spermatocytes at the bottom and the mature spermatozoa at the mouth of a duct. Because of this neat order and because the chromosomes are few and large, spermatogenesis in the grasshopper has been thoroughly studied with the light microscope. A comprehensive bibliography on this subject is found in Schroder's Handhnch der Entoiuologie (3). In the present study the Swedish grasshopper Chorthippus (Stauroderus) sp. was used. This species belongs to the grasshopper family Trn.xalidae which usually has a haploid chromosome numberof approxi- mately 10 (3). The shape of the late spermatid is similar to that of the spermatozoon of Tettigonia (Locusta) viridissima (as described by e.g. Retzius, (7)), the harpoon-like acrosome being the most striking characteristic. The nucleus of the spermatid in the stages examined is spade-shaped, the anterior region flattened and bordered by the acrosomal "harpoon barbs", while the posterior part has a triangular cross-section. The slender middle piece forms a smooth continuation of the head consisting of two mitochondrial filaments parallel to the tail filaments. In the short tail-end the tail filaments, but not the mitochondrial filaments, continue. This tail-end is thus analogous to the sperm tail of other animal groups. In this study with the electron microscope of ultra-thin sections attention was concentrated on the nuclear contents and the mitochondria. The Grasshopper Spermatids 165 ^ t .- Fig. 1. Survey picture of two spermatids of the grasshopper Chorthipptis. Cross-section through the anterior part of their heads. Magnification 46,000. ^, 1 /" 1^ Fig. 2. Detail of a head with part of the "caudal sheath" and the nuclear material. Magnification 103,000. acrosomal region consists of at least four different parts, the mutual relations of which are not entirely clear in this material. The acrosomal barbs have a homogeneous texture. The nucleus, as seen in fig. 1., has a strikingly regular ultrastructure. There are many uniformly thick opaque filaments in a parallel arrangement oriented along the long axis of the nucleus. In a transverse section at the broadest part of the nucleus approximately 3000 filaments, about 200 A in dia- meter, have been counted. The filaments are within restricted regions equally spaced in a hexagonal array giving a crystal-like appearance to that region of the nucleus. In longitudinal sections these filaments could be followed for a distance of several microns. There is no clear evidence that two adjacent filaments join at their ends. Towards the basal portion of the nucleus there is a general coarsening of the structure. Instead of the filaments dense conglomerates of irregularly sized opaque regions are observed near the centriole. The nuclear membrane shows a varying appearance in different regions. Partially it seems to consist of a single layer and partially a triple layered structure is evident. In the neck region and alongside the convex surface of the head of the spermatid a caudal sheath is found, resembling very closely the caudal sheath described in the cat spermatid by Burgos and Faw- cett (2), or that in the spermatozoon of the edible snail (6). This sheath is thus seen to consist of a cylindrical aggregation of apparently tubular fila- ments. Possibly the function of the caudal sheath is a supporting one as it covers the neck region, where the mechanical strains are at their greatest. The centriole exhibits no peculiarities and is not found in a nuclear indentation as is the centriole of the sea urchin spermatozoon (I). The relationship of the two mitochondrial fila- ments and the tail proper was studied on transverse sections of the middle piece. The space enclosed by the cell membrane is divided into two approximately equal half-cylinders, each having its individual mem- brane. One part consists of the eleven tail filaments in the traditional arrangement (4), two additional free filaments and a non-dense surrounding matrix material: the other part consists of the two mito- chondrial filaments embedded in a moderately dense matrix material. The non-dense tail part invariably 166 F. S. SJOSTRAND AND B. A. AFZELIUS Fig. 3. Length-section tiirougii one of the two mitochondria in the middle piece with the mitochondrial membranes as folds from a homogeneous part. Magnification 94,000. Fig. 4. Cross-section through the middle piece of the sper- matid showing the characteristic arrangement of the two mitochondria and the tail filaments. Note also the "caudal sheath" like tubiili near the mitochondria. Magnification X 112,000. bulges into the mitochondrial half along the central axis. The two central filaments are always in a plane, which would pass between the two mitochondrial filaments. As these latter are known to pass without twisting around each other or around the tail proper (7), it can be concluded that the central filaments also follow a straight course. Three equally thick filaments are visible in the light microscope in the grasshopper spermatozoon. Since two are obviously mitochondria, only one of the three is therefore responsible for the flagellar movements. The mitochondria, however, are not complete mitochondria as are those described by Sjostrand (8) and others. As seen in longitudinally sectioned middle pieces each of the two mitochon- drial filaments consists of one part in which the opaque material is homogeneous, and a second part, which shows a regular cross striated appearance. The cross striations are due to a stack of double membranes. These membranes are arranged with very regular spacing and are all of equal size, approxi- mately 230 A thick. The membranes appear as folds of a thin continuous membrane. Alongside the mitochondrial threads and within the matrix are tubular formations similar to those of the caudal sheath. The high degree of order thus revealed in the grasshopper spermatid makes this material espe- cially suitable for studies in which the ultrastructure as seen in the electron microscope is compared with x-ray analysis. So far, the technical problems en- countered in getting a clean preparation of oriented grasshopper sperm nuclei for x-ray diffraction have not been overcome (10). Birefringence and ultra- violet dichroism measurements performed on living locust spermatozoa, however, have shown that the physical properties of the head closely resemble those of microcrystalline fibers of pure sodium desoxyribonucleate (9). This study also suggests a crystalline or paracrystalline organization of the nucleoproteins within the grasshopper sperm head. It is not known how this organization may be correlated with the idea of the continuity of the chromosomes in the interphase nucleus. Grasse, Carasso and Favard (5, 6) in a study of the spermato- zoon of the edible snail described a similar longitu- The Acrosomal Reaction of the Sea Urchin Spermatozoon 167 dinal orientation of filaments in the nucleus and made the suggestion that these filaments represented the unwound and elongated chromatids. As the number of filaments is a few thousand and the num- ber of chromatids in the grasshopper approximately 20, a large number of turnings of the filaments at their ends should be found if their interpretation is correct. As we have not seen these connections, the idea of the chromosomes existing as extremely long folded threads is not supported by the present material. Another plausible explanation is that each chromosome splits into a great number of subtila- ments. References 1. Ai/nius, B. A., X.Xcllforsc/i. 42. 134(1955). 2. BuRCJos, M. H. and Fawcktt, D. W., J. Biophys. Bio- ciiem. Cytol. 1, 287 (1955). 3. Depdolla, p., in Schroder, Handbuch der Entomo- logie. Jena, 1928. 4. Fawcett, D. W., Laryngoscope M, 551 {\'-)5A). 5. Grasse, P-P., Carasso, N., and Favard, P., Compt. rend. acad. sci. 141. 1395 (1956). 6. — Annates Sci. \at. Zool. 18, 339 (1956). 7. Retzius, G., Biol. Unlers., N. F., 14, 55 (1909). 8. Sjostrand, F. S., Nature 171, 30(1953). 9. Wii.KiNS, M. H. F., Piit}f)t. Staz. Zool. Napoli 23, 104 (1951). 10. — personal communication, 1956. The Acrosomal Reaction of the Sea Urchin Spermatozoon B. A. Afzelius The Wenner-Gren Instiliite for Experinwiital Biology and the Laboratory for Biological Ullrastnictiire Research of the Department of Anatomy, Karolinska Institiitet. Stockholm 1 HE acrosomal globule, which in some mammalian spermatozoa flattens to form a "galea capitis" covering the anterior part of the nucleus (6), remains in the primitive or spherical condition in sea urchin spermatozoa. The acrosomal region of this sper- matozoon contains, in addition to the acrosomal globule, a mass of structureless or sometimes fibrous material situated within an apical indentation of the nucleus ( 1 ). This paper deals with the changes the acrosomal region undergoes while the sperm ap- proaches the egg surface or. in most but not all spermatozoa, when treated with egg water. "Egg water" is the term for sea water containing dissolved substances from the jelly coat surrounding the eggs. Figure 1 represents a Psammechintis miliaris spermatozoon fixed two minutes after addition of egg water. This spermatozoon has undergone an acrosomal transformation. The most notable feature of this reaction is the position of the acrosomal globule on the outer surface of the cell membrane. The mechanism for this transfer is unknown, since the cell membrane appears not to be ruptured at any point. Other changes have been noted in egg-water- treated spermatozoa: (a) The cell membrane is more blackened following fixation in osmium tetroxide than that of untreated spermatozoa, and is often separated from the enclosed structures by an empty space. As a rule the cell membrane is not ruptured. (/)) The middle piece, which appears to be a single huge mitochondrion, may have moved to a position alongside the nucleus and some of its internal double membranes may have swollen to form vesicles, (c) A portion of a tail may also have moved to a posi- tion at the side of the nucleus but inside the cell membrane that ordinarily closely invests the nucleus. To a much smaller extent this phenomenon has also been observed in untreated spermatozoa (as in fig. 3, ref. I). The possibility is recognized that some of these changes may represent abnormalities. In other experiments spermatozoa have been added to sea urchin eggs that showed a somewhat slow fertilization rate but were otherwise normal. After 2-4 minutes the eggs were fixed, dehydrated and em- bedded following the routine scheme (10). In sections of cortex and jelly coat of the eggs, spermatozoa were found in contact with the egg surface (first stage of fertilization). They were often found in a position perpendicular to the egg surface, in which cases the chance of getting a longitudinally sectioned spermatozoa was rather favourable. These spermato- zoa had all undergone an acrosomal reaction; i.e.. the acrosomal globule was always outside the cell membrane of the spermatozoon. Figure 2 is an example of an Echinus csciilentns spermatozoon in the immediate vicinity of an egg. The apical indenta- tion in the nucleus has the same dimensions as in spermatozoa that have not reacted. A fine filament, presumably formed from material originally situated in the apical nuclear indentation, is seen within the cell membrane in the most anterior part of the acrosomal region. This filament, which consists of a rather short rod of longitudinalls oriented fibrils, has apparently pushed the cell membrane before it, resulting in the formation of a pointed apex. The acrosomal globule outside the cell membrane is recognized by its homogeneity and density following osmium tetroxide fixation. It has become ring- shaped, and surrounds the pointed apex mentioned above. At a later stage in the acrosomal reaction the cell membrane has ruptured, allowing the ring-shaped 168 B. A. AFZELIUS Fig. I. The acrosomal region of a Psanvnechinus miliaris spermatozoon that has undergone the acrosomal reaction through treatment with egg water. Magnification 100,000. Fig. 2. The acrosomal region of an Echinus esciileiUiis spermatozoon that has undergone the acrosomal reaction in the immediate vicinity of an egg. Magnification 116,000. Fig. 3. The acrosomal region of a Strongyloceiitrotus droe- bachiensis spermatozoon that has undergone the acrosomal reaction with the acrosomal filament projecting from the sperm nucleus to a process on the egg surface. The nuclear indentation is obliquely cut. Magnification 96,000. Fig. 4. The acrosomal region of a Strotigylocentrotiis dioe- bachiensis spermatozoon that has undergone the acrosomal reaction and projects an acrosomal filament from the sperm nucleus to the egg surface at the bottom of the picture. The filament is broken into three fragments. Magnification > 84,000. acrosomal globule and its attached segment of cell membrane to separate from the sperm head. How- ever, the filament, which in many cases observed had elongated to a length exceeding 0.5 //, has retained its connection with both the sperm nucleus and the separated acrosomal region. Figures 3 and 4 illu- strate this stage of the acrosomal reaction in Strongy- locentrotus drocbachiensis spermatozoa. In figure 3 the filament and acrosomal ring are shown to advan- tage while the nucleus is not sectioned in such a way as to show its apical indentation. Figure 4 illustrates the nuclear indentation. In both pictures the acro- somal filament is fractured. It is not known whether this is due to imperfect fixation or to mechanical damage in subsequent treatment. Conclusions. — The two parts of the acrosomal region found in sea urchin spermatozoa, viz., the acrosomal globule and the material within the inden- tation in the nucleus, behave differently when the spermatozoon comes in contact with the egg or is treated with egg water. The acrosomal globule is expelled from the cell but evidently does not dissolve within the first 3 minutes at the temperature used during the experiments (4-10 C). The material within the nuclear indentation, on the other hand, remains on the inner side of the cell membrane and elongates to form the acrosomal filament which coexists with the expelled acrosomal globule. The phenomenon of the acrosomal reaction was first described by J. C. Dan (2) in sea urchin sperma- tozoa and later her studies were extended to a variety of other animal classes and to different cir- cumstances evoking the reaction (3, 4, 5, 11). The possibility to perform ultra-thin sectioning (accord- The Galea Capitis in llttnian Sperm 169 ing to the technique of Sjostrand (9, 10)) made it possible to differentiate the two parts of the acro- somal region and to follow the fate of these two parts and of the cell membrane during the acrosomal reaction. It also made possible the examination of spermatozoa in contact with the egg surface. It has been found that all spcrnuitozoa in contact with cgt^s exhibit the acrosonuil reaction. The probable im- portance of this phenomenon in fertilization is there- fore emphasized. Spermatozoa treated with egg water, on the other hand, do not show a full per- centage of acrosomal reaction. Regarding the functions of the two parts of the acrosome in the process of fertilization nothing in the present observations is contrary to the belief of Wada, Collier and Dan (II) that the acrosome contains the lysin that dissolves the egg membrane, although stages in this process could not be studied. More likely the acrosomal globule is the one which consists of and contains the egg membrane lysin. The impression gained by Rothschild and Tyler (8) from the electron microscopic study of whole mounts (unsectioned material) that the acrosomal globule forms the acrosomal filament seems to be erroneous. The author wishes to thank Dr. G. Gustafson, the Kristincbcrg Zoological Station, and lk'st>rcr I). Riistai-I, Trondhcim Biological Station, for the help and supply of material; Dr. F. Sjostrand, Associate Professor of the Dcpailnicnt of Anatomy, for placing the resources of his laboratory at the author's disposal; F^rofessor J. Runnstrom, the Wenncr-Gren Institute, for his unfailing support and interest. Financial support from the Swedish Natural Science Research Council is also gratefully acknowledged. References 1. Afzelius, B. K.,Z.ZeUforsch. 42, 134 (1955). 2. Dan, J. C, Biol. Bull. 103, 54 (1952). 3. — ibid. 107, 203 (1954). 4. — ibid. 107, 335 (1954). 5. Dan, J. C. and Wada, S. K., Biul. Bull. 109, 40 (1955). 6. Fawcett, D. W. and Burgos, M. H., Ciba Foundation, Colloquia on Ageing 2, 86. London, 1956. 7. Rhoimn, J., Correlation of Ultrastruclurai Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Stockholm, 1954. 8. RoiHSCHiLD, Lord and TviiR, A., Exptl. Cell Rcsccinh, Suppl. 3, 304 (1955). 9. Sjostrand, F. S., Nature, 168, 646 (1951). 10. — /■// Osrru and Poi iistfr. Physical Techniques in Biological Research, 111, 241. New York, 1956. 11. Wada, S. K., Collier, J. R., and Dan, J. C, E.xptl. Cell Research. 10, 168 (1956). The Structure of Galea Capitis in Human Sperm J. Schultz-Larsen and R. Hammen The University Institute of Human Genetics, the Department of Gynecology cmd Obstetrics at the University Hospital, Copenhagen, and the University Institute of Biophysics, Copenhagen As galea capitis, at least as far as animals are con- cerned, has shown itself to be important for sperm fertility, we have undertaken an investigation of this structure and its relationship to the acrosome as part of our systematic electron microscopic research on the normal morphology of the human sperm, in the course of this work, we have found a formation in the foremost part of the head of the sperm which to our knowledge has not been described previously in human sperm. Lively sperm from young men whose ejaculates had been subjected to thorough analysis and found to be normal were used. The ejaculates were diluted 10-15 times with Tyrode's solution, and fixed with 1 "o osmium tetroxide with a pH of 7.4 in isotonic ion concentration. Dehydration in alcohol and embedding in butylmetacry- late. Our experiments have demonstrated that impurities, which are present in large amounts in normal human sperm, are largely the cause of irregular polymerisation. When using chemical accelerators (bcnzoylperoxidc. 2:4,dichlorobenzoylperoxide, lauroylpcroxide)it has thus been necessary to remove impurities from the ejaculates prior to fixation by means of repeated centrifugation and suspension in pure Tyrode solution. These procedures have, however, proved to reduce the number of motile sperm considerably, and the membrane mentioned above is damaged. Polymerisation with ultra-violet light without chemical accelerators makes it possible to work with less pure preparations and yet achieve satisfactory preserva- tion. The sections have been cut with a Sjostrand ultra- microtome with razor blades. In preparing sections with the latter microtome, the method suggested by Sjostrand (II) has been closely followed. In the preparations made by merely spreading a suspension of sperm heads which had been subjected to the normal cleansing and fixation procedures, one usually cannot see the galea, and in no way gains any impression of the structure to be described below. Fig. I shows that the galea is a hood-like structure which reaches down o\er the forepart of the head. The hood is in this preparation torn off the head. The galea capitis consists of two layers; the inner one, nearest to the head, and the outer layer, each of which is about 50 A thick. The two lasers are separated by a space appro.ximately 400 A wide. The galea stretches from the foremost point of the head down to the head's equator, where the inner and outer layer form a cul de sac. Galea capitis is bounded on its outer side by a thin membrane which continues past the equator of the head to cover the basal part. This outer membrane is only 50 A thick, and its continuation forms the membrane that covers the middle-piece, the tail and the tail tip. 170 J. SCHULTZ-LARSEN AND R. HAMMEN Fig. 1. Longitudinal section of the head. Galea capitis con- sists of an outer membrane {a) and an inner one (h), which continues down and lines a cavity in the tip of the head (c). Galea capitis has been forced off the head. On the surface of galea capitis there is an outer membrane approximating 50 A which continues down to the basal part of the head (d). Magnification 22,000. The inner membrane of the galea capitis forms a pocket lining an acrosomal cavity. At the bottom of this acrosomal cavity is seen a denser struc- ture. The cavity is otherwise filled by a structureless mass that continues into the mass filling up the space between the two galea capitis membranes, and which gives less contrast than the other parts of the head. We have sectioned several somewhat shorter heads in a sagittal or oblique plane and have found similar structures, but with acrosomal cavities larger and oval in shape. Discussion. — Galea capitis in human sperm is exceptionally delicate and fine in structure — so deli- cate, in fact, that up to now there has been doubt as to whether such a structure was present in mature sperm (cf. 5, 6, 12). Electron microscopists, too, have denied its existence in humans (e.g. ref. 8). On a few of Retzius (9) pictures of mature human sperm one nevertheless discerns a contour just in front of the equator of the head, which would cor- respond to the rearmost boundary of a hood. Some irregularly shaped heads likewise suggest the presence of a galea in human sperm (cf. 3. plate I, figure 5), which may be hypertrophied relative to the head. Furthermore, Williams (13) is of the opinion that he has observed a loose galea in a single case. Phylogenetically the presence of galea in human sperm is a reasonable assumption. It has been pos- sible to prove the existence of a hood in a large number of animals from varying phylogenetic levels, though these hoods have been of diverse shapes and extent, and often occur together with a so-called perforatorium. The presence of a galea in human sperm is also strongly supported by the investiga- tions of Gatenby et al. (2) of spermiogenesis. As yet it is impossible to decide upon the function of galea capitis and the acrosomal part of the sperm head. As galea capitis always forms part of the spermatozoa of various types of animals, it might be supposed that, together with the acro- somal part of the head, it has an important part to play with regard to the possibility for the sperm to reach the egg-cell. References 1. Afzelius, Bjorn a., Z. Ze/lforscli. 42. 134(1955). 2. Gatenby, J., Bronte, J., and Beams, H. W., Quart. J. Microscop. Sci. 78, 1 (1936). 3. Hammen, Rich., Studies on Impaired Fertility in Man. Copenhagen and London, 1944. 4. Hammen, Rich., Schliltz-Larsen,Jorgen, and Carlsen, F. E., Feitil. Sreril. 5, 411 (1954). 5. Held, H., Ber. Verhain/I. sdc/is. Akad. Wiss. Leipzig, Math.-phys. Kl. 68, 205 (1916). 6. Marcus, H., Arch. Zellforsch. 15, 445 (1921). 7. Papanicolaou, George N. and Stockard, Charles R., J. Anat. 24, 37 (1918). 8. Reed, C. J. and Reed, B. P., Amir. Rec. 100, 1, 1948. 9. Retzius, Gustave, Biologische Untersuchungen. XV, 57 ft"., 1910. 10. Schultz-Larsen, Jorgen, Hammen, Rich., and Carl- sen, F. E., Acta Pathol. Microbiol. Scand. 35, 45 ( 1 954). 11. Sjostrand, F. S., E.yperientia 9, 14 (1953). 12. Stieve, H., Harn- und Geschlechtsapparat. II. Teil. /// W. v. MoLLENDORFF, Handb. der mikr. Anat. d. Menschen 7, 103 (1930). 13. Williams, W. W., Fertil. Steril. 1, 199 (1950). The Ultrastructure of the Thyroid Gland of the Mouse R. Ekholm and F. S. Sjostrand The Laboratory for Biological I'ltrastriictiirc Research of the Departnicm o/ Anuioiiiy. hciro/iiiskci hi.sririiict, Stockholm, ami the Department of Anatomy, University of Gothenburg This paper presents the results obtained mainly by studying the thyroid gland of mice kept in normal laboratory conditions. These results will be the basis of a further analysis of the structural changes in the thyroid in connection with the stimulating and inactivating of the gland. The thyroid has earlier been studied with aid of the electron microscope by several investigators (1,2, 4). The observations made by these authors are in general accord. Thus, they all give an account of the existence of microvilli on the follicular surface of the thyroid cells and a "lamellar"" or "canalicular"" structure in the cytoplasm. However, as regards the structure of the capillaries, opinions are at variance. Monroe suggests that, in places, the endothelium lining of the capillaries is discontinuous but Demp- sey and Peterson are of the opinion that close exa- mination always reveals a continuous capillary wall. Material and methods. — The thyroid gland of 40 adult white mice were examined. The fixation of the tissue specimens was performed in a blood isolon 1 per cent osmium tetroxide soliilion, butVered at pH 7.2 with veronal acetate, a modification of the osmium solution of Palade (6). The specimens were embedded in a mixture of /;-biityl and /;-methyl niethacrvlate mainly according to Newman et al. The uluathin sectioning was perft)rmed partly with the microtome designed by Sjostrand (7), partly with an instrument described by Kkholm &Zelan- der (3). The sections were examined partiv in an RCA EMU 2b, partly in an RCA EMU 3b microscope. Results. — The thyroid cells are organized accord- ing to a very regular pattern as revealed by a survey picture (tig. I). The cells are bordered by a plasma membrane which at the top surface, facing the follicle, protrudes in a greater or lesser number of microxilli of varying size and shape. At the base the cell membrane, associated with a basement membrane, is usually seen to run in close vicinity to a capillary. The cell nucleus is situated in the centre of the cell. A large number of mitochondria are evenly dis- tributed throughout the cells. The apical zone of the Fig. I. Survey picture of mouse thyroid cells with the follicle (uppermost), cell boundary (CB), mitochondria (M). intra- cellular cytoplasmic membranes (CM), nuclei (N) and capillary (C). Magnification 12,000. 172 R. EKHOLM AND F. S. SJOSTRAND cells contains small granules. Predominantly in the middle and basal parts of the cells more or less regu- lar systems of membranes are seen. The plasma membrane bounding the cytoplasm of the thyroid cell is in its major course observed as a single membrane having a mean thickness of about 80 A (79 ±3 A). The membranes of two adjacent cells run regularly in close relationship, the mean distance between the lines being about 145 A (146 ±7 A). The part of the plasma mem- brane that bounds the villi is thinner, about 60 A (58 ±5 A). Here the membrane, at least in some places, is observed as a double-edged one composed of two dark lines with a less opaque space in be- tween. The maximum length of the microvilli is about 0.4 // and the maximum diameter about 0.08 //. 77;t' mitochondria, usually rod-shaped, are scat- tered all over the cells apparently at random. The length of the mitochondria varies considerably in the sections but the width was found to be fairly constant with a mean value of 0.25 ±0.016 /*. The mitochondria are bordered by an outer double-edged membrane composed of two dark lines separated by a less opaque space. The mean distance between the centers of two dense lines is 1 10 ± 3 A and the total thickness of the membrane measured between the outer limits of the dark lines is 1 65 ± 4 A. The calculated thickness of the individual dark lines is consequently about 55 A and the width of the space between the lines about 55 A. Inside the mitochondria a system of inner mem- branes is observed. These membranes, mainly oriented perpendicularly to the long axis of the mitochondrion, also appear double-edged. Usually the inner membranes are in contact with the outer membrane at both ends but sometimes only one end of the inner membrane touches the outer, the other end being free. At the point of contact the central space of the inner membrane continues, in isolated cases, in the corresponding space of the surface membrane, but usually the two spaces are separated by a dark line. The measured total thickness of the internal mito- chondria membranes is 180 ± 3 A and the distance between the centers of the two dark components 125 +4 A. Consequently, the calculated thickness of the individual dense lines is about 55 A and the width of the space between the lines about 70 A. The intraceUular cytoplasmic membranes, by Sjo- strand (8) also named a-cytomembranes, appear as a thin basic membrane to which at one side small dark particles are attached. These particles are ir- regularly shaped but are usually fairly regularly inter- spaced (150-350 A) along the basic membrane. How- ever, within shorter distances the membranes are completely free from particles. The mean diameter of the particles is 145 ± 9 A and the mean thickness of the basic membranes is 55 ± 3 A. The general organization of the a-cytomembranes varies. In restricted areas they show a rather regular. parallel arrangement. In such areas the membranes seem to be arranged in pairs in such a way that two adjacent membranes face each other with the particle- covered side, the distance between the two compo- nents of each pair being rather constant (about 0.05 //). The smooth surfaces of the membranes are in these areas separated from each other by a wider and more irregular space. If the membranes are followed for a certain distance it is seen that the two nearest membranes from two adjacent pairs join and so close the space which is bounded by the smooth sides of the membranes. In other areas this regular arrangement is not obvious because the spaces between the smooth sides of the membranes are irregular in shape and vary very much in size. Granules. In the cytoplasm of glands from animals which have lived in normal laboratory conditions only few and small granules are found. But if the gland is slightly stimulated (by low room tempera- ture) the granules increase in number and size and it is possible to distinguish at least two types of granules. One type has a content of irregular aggre- gates of a dense fine granular material. Another type has a denser packed, evenly distributed fine granular content. Both kinds of granules are bounded by a single, about 50 A thick membrane. The Golgi apparatus is usually found in close rela- tion to the nucleus. It is made up of short membranes arranged in pairs, vacuoles and small vesicles. The cell nucleus has a content of osmium-impreg- nated particles, irregularly arranged but densely accumulated at the nuclear membrane and in some areas in the interior of the nucleus. The nuclear envelope appears in sections cut perpendicularly to the nuclear surface as a double-edged membrane, consisting of two dense layers which are separated from each other by a less opaque space. The dense layers are about 50 A thick and the space in between varies from about 80 to 250 A. The double edging of the membrane is, however, lost at intervals and within these areas the nuclear membrane is rather indistinct. The dimension of these areas is approxi- mately 400 A. The capillaries. The basal part of the thyroid cell usually borders on a capillary. Here the plasma membrane runs irregularly and is separated from the capillary endothelium by a space containing two basement membranes, one belonging to the thyroid cell and the other to the endothelium. The thickness of each of these membranes is 300 500 A. The thickness of the endothelial wall differs very much from place to place but the major part of the wall is very thin, 0.03-0.05 //. Within these parts a constant feature is the appearance of discontinuities of the endothelial lining (fig. 2). These discontinuities are located at varying distances from each other but never coincide with the boundary between two adjacent endothelial cells. They are, however, no true interruptions for they are always bridged by a Chick Embryo Thyroid 173 Fig. 2. Section through the basal part of a thyroid cell and a capillary showing endothelial discontinuities. BM = base- ment membrane: END -■^- endothelium: RB -- red blood corpuscle. Magnification 62,C00. more or less distinct fairly dense membrane and the endothelial basement membrane continues uninter- rupted over these areas. The width of the discon- tinuities is about 300 A. References 1. Braunsteiner, H., Fellinger, K., and Pakesch, F., Endocrinology 53, 123 (1953). 2. Dempsey, E. W. and Pi n rson, R. R., Endociinoloi;y 36, 46 (1955). 3. Ekholm, R. and Zelander, T., Experientia 12. 195 ( 1956). 4. Monroe, B. G., Anat. Rec. 116, 345 (1953). 5. Newman, S. B., Borysko, E., and Swerdiow, M., J. Research Natl. Bur. Standards 42,, 183 (1949). 6. Palade, G. E., J. E.xptl. Med. 95. 285 (1952). 7. Sjostrand, F. S., Experientia 9, 114 (1953). 8. — //( PoLLiSTER, A. W., Physical techniques in biological research. Vol. III. New York, 1956. Electron Microscopy of Chick Embryo Thyroi(d R. Stole, P. Blanquet, A. P. Lachapele, R. Maraud, and A. Magimel Fondation Bergonie and Facidte de Medecine, Bordeaux According to a previous investigation the differ- entiation of the chick embryo thyroid shows two successive periods (18). In the first one, before 12 days, the thyroid anlage concentrates very low quan- tities of radioiodine and is practically devoid of follicles. During the second one, the uptake of radio- iodine raises suddenly and strongly together with the differentiation of the follicular structure. Differ- ence between the two periods is closely associated with colloid synthesis. The present investigation was undertaken to de- termine to what extent this phenomenon could be related to the development of the c\toplasmic cana- licular structures which were described in the thyroid of adult mammals (4, 14. 15. 23. 24). Normal thyroid glands were obtained from 36 embryos 174 R. STOLL, P. BLANQUET, A. P. LACHAPELE, R. MARAUD AND A. MAGIMEL J*^* .9^ Fig. I. Thyroid from a 9 days old embryo. The gland is of a cordonal type and cells are practically devoid of canalicu- lar structures and colloid. Mitochondria and small vacuoles are visible. Magnification 5000. and chicks from 8 days of incubation to 13 days after hatching (Light Sussex). Thyroids were also removed from 18 embryos of 9 and 16 days previously treated at 3 days with thyroxine (0.001 mg) or tetramethylthiourea (1.5 mg). Choice and dosage of the drugs were established from our previous investigations. The glands were fixed in osmic acid, buffered according to Palade (1952) and embedded in n-butyl methacrylate. Sections 0.05 /< in thickness were cut with a Porter-Blum microtome and examined in a Philips electron microscope (Mod. 1951). In the diflferentiation of the thyroid, electron microscope shows two periods identical to those previously described. In the first one, before 12 days of incubation, the canalicular structures and the col- loid are very poor (fig. 1). At 8 days, the gland consists of cellular cords which form spherules progressively later. Cellular nuclei contain a nucleolus and are ovoid, sometimes bean-shaped. The cytoplasm possesses a clearly dis- cernible chondriome and numerous vacuoles variable in size and containing a pale substance. The cana- licular structures, very scarce and small, lie generally at the periphery of the cell and show a clear lengthened center surrounded by a thii osmophilic wall. Be- tween 2 or 3 epithelial cells, some small colloid droplets appear, containing filamentous protrusions, or microvilli inserted on the cellular membrane. The second period is characterized by the differ- entiation of follicular colloid associated with sudden and general distribution of the canalicular structure. Its beginning is already discernible in the thyroid of 12 days old embryo. Sometimes the cell, very poor in canaliculi, presents the aspect of the preceding period; sometimes the cell, apparently devoid of vacuoles, appears clear and possesses numerous canaliculi which are short and fine (fig. 2); sometimes intermediary aspects are present. Generally speaking. the small follicles are poor in canaliculi while the greatest are rich in such formations. The epithelial cords still persisting at this time, keep their initial structures. In the 13 days old embryo, the follicles increase in number and size while short canaliculi exist now in all the cells. Some osmiophilic projec- tions, issued from granules of the apical part of the surrounding cells, are seen in the colloid. Afterwards, the follicular structure is generalized in the thyroids removed from the 16, 17, 19 days old embryos. The long and twistened canaliculi fill then almost all the cytoplasm where they lie, following its larger amount. According to the disposition of the canaliculi, three states can be distinguished in the cells of the same follicle. Few cells are narrow and dark, because their canaliculi have almost joint walls and poor, lightly osmiophilic contents. The other cells are cuboidal in shape with a relatively clear appearance. This is caused by the canaliculi which are well developed and moderately filled with a pale ground substance. Among these elements, some cells show apical vacuoles which seem to be related to sectioning of highly dilated canaliculi, in the apical part of the cells. Between these vacuoles, osmiophilic granules, variable in size, are scattered. In all the cells, mitochondria are always covered with a canalicular sheet. After hatching, the thyroid shows its well known follicular structure. Around the colloid, with its various osmophilic filaments, the three varieties of previously described cells form the follicular wall where a last cellular state is well recognizable (fig. 3). In this case, the canaliculi are so much filled with the pale ground substance that the cell shows an aspect of colloid degeneration. On the other hand, according to the observations of Monroe (14) on • 6 / V- ^ ^•:. If. Fig. 2. Thyroid from a 12 days old embryo. The cells sur- rounding the follicular colloid present numerous fine cana- licular structures. Magnification 5000. Chick Embryo Thyroid 175 Fig. 3. Tli>roi(.i from a just-hatched chick. The canalicular structures are well developed. In the apical part of some cells, osmiophilic granules and vacuoles, developed from canaliculi, are scattered. Note a sinusoidal capillary with endothelial cells. Magnification -< 2000. the thyroid of the rat, the capillaries which surround the follicles of chick thyroid are of a sinusoidal type. During the first period, thyroxine and tetramethyl thiourea have no evident action on the thyroid. On the other hand, they produce specific changes in the second phase of the differentiation. Following treatment with thyroxine, a few cells appear normal, but the other present a washed-out appearance and are practically devoid of canalicular structures. The canaliculi which differentiate are very small and, according to Vidal (22), the formation of follicular colloid is very reduced. After administration of tetra- methylthiourea, the follicles present columnar cells surrounding a reduced follicular cavity, as described by Adams and Bull (I). The cytoplasm is clear and filled by canaliculi which are strongly twistened by an abundant pale substance. During the differentiation of the chick thyroid, the apparition of canalicular structures, permit to separate two periods which correspond to those previously observed after the use of radioiodinc (16, 18). In the first period, before 12 days, the colloid, the canaliculi and the uptake of radioiodine are very poor. In this case, the extracellular colloid droplets seen by electron microscope, correspond to the chromophobic colloid (21) and probably to the P.A.S. positive droplet*; of the classic microscopy (5, 19, 26). The role of the vacuoles and osmiophilic filaments in the colloid remain unknown. In the second period, from 12 days, the physio- logical and histological differentiation of the thyroid advances rapidly. Radioiodine storage and hormonal synthesis increase strongly (2, 16, 20, 26). Simul- tancousl> follicles ditVcrcnliatc and colloid develops (17, 21). Interrelationships exist between these two groups of phenomena (18). Indeed, previous works on the thyroid of adult mammals ha\c shown that the radioiodine is included in the follicular colloid, after a short passage through the glandular cells (II, 12). The radioiodinc is also stored in ihc colloid of the chick embryo thyroid (10). The iodine metabolism o\ the th\roid is therefore related to the colloid whose synthesis follows the apparition o\ the canalicular structures. Their sudden extension, between 12 and 13 days, explains the rapid increase of colloid dcsclopment and radioiodine uptake. The factors governing the formation of the cana- liculi are an interesting subject for research. The most important point seems to be the embryonic hypophysis as shown by the use of thyroxine and tetramethyl thiourea. Indeed, following treatment with thyroxine, the formation of canaliculi and, consequently, the col- loid elaboration are strongly inhibited. This hor- mone is known to act by the way o\ the hypophysis whose thyrotrophic function is reduced by a high level of thyroxine. On the other hand, the antithyroid compound produces an excessive canalicular devel- opment. This is due to the increase of the T.S.H. whose high level is demonstrated by the hypertrophy of the chick embryo thyroid (I, 8). In conclusion, the development of the canalicular structures depends on the hypophysis which so controls the histophysiological differentiation of the thyroid, after 12 days. This explains the modifications of the embryonic thyroid when the hypophyso- thyroideal axis is affected by hypophysectomy (6, 7, 13). antithyroid drugs (1,8), or in vitro culture (19). Therefore, other factors, as iodide, can intluence the canaliculi and colloid (unpublished data). Such a fact could explain the absence of thyroideal modi- fications sometimes observed after embryonic hypo- physectomy (25). Rfferences 1. Adams, A. E. and Bull, A. L., Amii. Rcc. 104. 421^38 (1949). 2. Blanqui r. P., Stoll, R.. Maraid, R., and Caput, L., Conipt. rend. soc. hiol. 147. 676 678 (1953). 3. Bkrnhard, v.. Haguenau, F., Gauthier, A., and Obprlinc;. C. Z. Zellforsch. 37. 281 (1952). 4. Braunstlim R. H.. Fkllinger, K.. and Pakfsch, F., Endocrinology 53. 123-133 (1953). 5. Carpenter, E., Beatty, J., and Chamhi ks, R. D., J. E.xpll. Zool. 121, 249-269 (1954). 6. FUGO, N. W., J. E.xptl. Zool. 85. 271 297 (1940). 7. FuGO, N. W. and Witschi, E., .-Xcta hiologica larvincaH (1938). 8. Grossowicz, N., Proc. Soc. E.xpil. Biol. Med. 32, 51 (1946). 9. Haguenau, F. and Bernhard, W., .Arch. Anat. Micr. et Morplwl. Exp. 44. 27-55 (1955). 10. Hans borough, L. A. and Mustapha Khan, /. Exptl. Zool. 116, 447-452 (1951). 176 B. V. HALL AND L. E. ROTH 11. Leblond, C. p., J. Aiuit. n, 149-152 (1943). 12. Leblond, C. P. and Gross, J., Emhcriiwlogy 43, 306-324 (1948). 13. Martindale, F. M., Anat. Rec. 79, 373-393 (1941). 14. Monroe, B. G., Anat. Rec. 116, 345-362 (1953). 15. NosEDA, J., Z.mikroskop. anat. Forscli. 60, 192-204(1954). 16. Stoll, R. and Blanquet, P., Ann. Endocrinol. 13, 569- 577 (1952). 17. Stoll, R., Blanquet, P., Maraud, R., and Capot, L., C.R. Ass. Anat. XL" reunion. 785 (1953). 18. Stoll, R., Maraud, R., Blanquet, P., Mounier, J., and Meyniel, G., Ann. Endocrinol. 17, 15-32 (1956). 19. Tlxier-Vidal, a., Compt. rend. soc. biol. 149, 1377-1379 (1955). 20. Trunnel, J. B. and Wade, F. T., /. Clin. Endocr. Metab. 15, 107-117 (1955). 21. Venzke, W. G., Am. J. Vet. Research 10, 272-2SI (1949). 22. ViDAL, A., Ann. Endocrinol. 14, 444-449 (1953). 23. Walthard, B., Bull. Scliwelz. Akad. Med. Wiss. (\955). 24. Weber, G., Zampl G., and Ignesti, V., Arch, de Vecchi 11, 297-308. (1954). 25. Wolff, E. and Stoll, R., Compt. rend. soc. biol. 126, 1215-1217 (1937). 26. Wollmann, S. H. and Zwilling, E., Endocrinology 52, 526-535(1953). Preliminary Studies on the Development an(d Differentiation of Cells and Structures of the Renal Corpuscle B. V. Hall and L. E. Roth Department of Zoology, University of Illinois, Urbana, and Division of Biology and Medicine, Argonne National Laboratories, Lemon t. III. Soon after Bowman published his classical descrip- tion of the Malpighian corpuscle, Gerlach ( 1 ) de- scribed the glomerular capillaries as lying between an invaginated double layer of cells extending from the inner wall of the capsule, as the intestine lies within the double layered peritoneum extending from the abdominal wall. Zimmermann (11) relied heavily on Gerlach's concept in developing his much disputed hypothesis of a glomerular "mesangium". More recently others have employed Gerlach's idea as a basis for a modified mesangial theory. In the modified theory (7), it is supposed that a special limited space, termed in a specific sense "intercapil- lary space"', lies between the limbs of hypothetical capillary loops. Both mesangial cells and intercapil- lary space are imagined to be separated from capsu- lar space by invaginated, two layered extensions of the basement membrane and cells of Bowman's capsule. Hall (3) has obtained many convincing photo- graphs which afford no evidence that the glomerulus is formed of simple capillary loops, as Vimtrup (10) thought. If glomerular capillaries are not ar- ranged as simple loops, then the whole theory that the Malpighian body is formed by invagination of capil- lary loops into a blind, terminal vesicle of a renal tubule must be held questionable. In this respect, it is important to note that the able embryologist Huber (6) stated nearly fifty years ago, "the renal corpuscle consists of a double-walled capsule, the glomerular capsule, usually spoken of as the in- vaginated end of the renal tubule, though it is not developed by invagination". Huber (5) agreed fully with Herring's (4) earlier conclusion that glomerular differentiation takes place largely by "proliferation" and "readjustments" of its cells. To obtain a clearer understanding of the nature of the cells and structures of the mature glomerulus and of the embryonic processes involved in their formation, we began some time ago to investigate developmental stages of the Malpighian body by examination of thin sections under the electron microscope. Immediately after decapitating 1-5-day-old rats, small blocks of tissue, about I mm on edge, contain- ing immature glomeruli were removed from cortical regions of their kidneys. The blocks were fixed in 1 '^o buffered OsOj solution, as described by Palade (8). Washed and alcohol-dehydrated tissues were embedded in a 19:1 butyl-methyl methracrylate mixture. The sections were cut with glass knives on a modified Minot International microtome at ^!.,^, and 1/40 /'. Before viewing them in a RCA-EMU-2 electron microscope, the sections were mounted on grids with formvar membranes and immersed in toluene for several minutes to remove the plastic in order to increase image contrast. The earliest stages were recognizable chiefly by the total pattern of all constituent cells, since indi- vidual cells showed few or no specializations. The cells were arranged in solid masses with two outer layers arranged somewhat concentrically around a central group of cells showing no discernible organi- zation. In these earliest stages, cell membranes were difficult to detect, especially between cells of the inner cell mass. However, the external surfaces of the cells of the inner mass (prospective endothelial cells) and of the inner concentric layer of cells ( Bowman's visceral layer, or prospective podocytes (2, 3)), where they were opposed to each other, appeared to develop prominent surface membranes very early. No defini- tive capillary basement membrane, or lamina densa (2, 3) was observed in these early stages. Only the prominent, opposed surface membranes of the pro- Cells and Structures of the Renal Corpuscle 177 Fig. 1. Photomontage of electron micrographs of an early stage of a developing, rat renal glomerulus, with prospective endothelial cells but without capillaries, or capillary structures. {A) Bowman's capsular epithelium, and thin capsular basement membrane characteristic of these early stages. {B) Initial capsular space, seemingly developed by separation of the two epithelial layers. (C) Prospective podocytes, or Bowman's visceral epithelial layer. (D) Prominent cell surface mem- branes between the outermost prospective endothelial cells and the prospective podocytes. Especially noteworthy are the absence of a capillary basement membrane, and the short, broad extensions of podocytic cytoplasm between the prospec- tive endothelial cells. (£") These extensions may play an important role in the organizing of prospective endothelial cells into the capillary system characteristic of the glomerulus. (F) Two cells of an immature arteriole. (C) Capsular connection with tubule. Note within the mass of prospective endothelial cells, the mitotic figure, ni. and the precocious erythrocytes, as f, and the complete absence of formed capillaries. Osmium fixed. Magnification 2065. spective endothelial and podocytic cells were seen bounding the narrow space which in a later stage would be occupied by lamina densa. During these early stages of development, the basement membrane of Bowman's capsule appeared very tenuous. It had much the appearance of a fused, thickened membrane on the external surface of the prospective parietal epithelium. Flattening of the outermost concentric layer of cells and the concomi- tant initial definite appearance of a limited capsular space was the next recognizable advance in the dif- ferentiation of the Malpighian corpuscle. Pronounced mitotic activity of the prospective endothelial cells characterized this stage (fig. i ). The flattening of the prospective parietal cells, and the enlargement, perhaps even the initiation, of Bowman's capsular space may depend upon the elaboration and accu- mulation of fluid between the parietal and visceral 12 — 568204 Electron Microscopy layers of Bowman's capsule. Capsular space appears to arise from separation of the parietal and visceral epithelial layers. It then enlarges, if invagination played a major \o\c in glomerular development capsular space should decrease. The.sc facts support the concept that in the rat the renal corpuscle is not developed by invagination. DitTerentiating podocytes, after developing pro- minent surface membranes, usually appeared par- tially separated from each other. It appears that membrane formation was important to the separa- tion process, increase in surface area of the inner cell mass, afforded b\ mitotic di\ision and growth of the prospective endothelial cells, may be an added important, mechanical factor tending to induce separation of the developing podocytes. The next observed advance in glomerular develop- ment was initiation of formation of complex folds 178 B. V. HALL AND L. E. ROTH Fig. 2. Photomontage of electron micrographs of a later stage in the development of the rat, renal glomerulus, demonstrat- ing the early formation of the lobular organization, and initial stages of capillary formation. These incompletely differen- tiated capillaries are characterized by a very limited and apparently discontinuous capillary lumen, and by large, connecting masses of still undiflerentiated endothelial cells, and by the presence of few differentiated endothelial cells. The capillary spaces appear to develop by the expansion and subsequent fusion of intra-cytoplasmic spaces which develop within the differentiating endothelial cells. (A) Capsular epithelium, and basement membrane, still characteristically thin. {B) Much enlarged capsular space. (If invagination played an important role in glomerular formation, this space should be reduced, not enlarged.) (C) Maturing podocytes, developing discrete processes, pedicels and trabeculae, and greatly increasing in size. (D) The cell surface membrane of the podocyte, which folds to form the numerous pedicels and trabeculae, while the definitive capillary basement membrane forms in intimate relationship with the surface membrane of the endothelial cells. (£) Prospective and differentiated endothelial cells. (F) Maturing arteriole. (G) Connection with tubule. Magnification x 2025. and processes in prospective podocytes. The mature podocyte (2, 3) is characterized by an extensive and complex system of relatively large processes, the trabeculae, and numerous, minute, interdigitating, terminal processes, the pedicels (foot-processes) which arise bilaterally from more or less parallel trabeculae radiating from the central mass of each podocyte. Differentiating podocytes acquired first, short, relatively broad processes, like those in fig. 1, which appear to fit into irregular spaces between the peripheral prospective endothelial cells. In later stages, it appears that these apparent processes have developed into ridges, or processes (the trabeculae), which are 30-40 // long in some instances. They seem to penetrate into the compact inner mass of prospec- tive endothelium and to separate prospective endo- thelial cells into connecting groups of primordial capillaries that differentiate later into definitive capil- laries (fig. 2), The first, short, crude podocytic pro- cesses seemed to have few or to lack completely discrete pedicels. The elongated processes, however, showed numerous, typical, bilaterally arranged pedi- cels. Podocytic processes appeared to develop by folding of the cell surface, and by extension of dis- crete, cytoplasmic processes. The entire elaborate system of large and small processes forms only from the surface of a podocyte opposed to the surface membrane of the differentiating endothelium. Al- though the present study does not establish with certainty that growth of podocytic processes into the unorganized mass of prospective endothelium plays an active role in organizing the primordial capillaries and lobules, the evidence of the electron micrographs is in full accord with such a view. Capillary lumina or enlarged cytoplasmic vesicles were not found in prospective endothelial cells in Cells and Structures of the Renal Corpuscle 179 the earliest stages of development, but in slightly later stages an occasional erythrocyte was present in the central mass of prospective endothelial cells. Lumina appeared to develop later in the solid pri- mordial capillaries. Lumina appeared to begin devel- opment with the appearance of large cytoplasmic vesicles within individual endothelial cells. Later as the vesicles appeared enlarged, the endothelial cyto- plasm, except around nuclei, appeared attenuated. The impression was gained that the continuous capillary lumina of mature capillaries is formed by the subsequent fusion of enlarged, individual cyto- plasmic vesicles. In figs. I and 2, it is to be noted that some vesicles appear filled by a fully difTeren- tiated erythrocyte. This may be interpreted to mean either that erythrocytes differentiate /// situ, or that the vesicles are apparent only and exist merely as tangential sections of lateral outpouchings of con- tinuous,patent capillary lumen filled with plasma and erythrocytes. The validity of either possiblity can be determined with certainty only by means of a large series of serial sections not available now. Yet, the presently available sections, including some serial ones, through about ten immature glomeruli have produced no support for the second interpretation. From a statistical point of view, it is very unlikely that all sections through many immature capillaries and several arterioles were cut tangentially so that all failed fortuitously to show evidence of a continuous, patent capillary lumen in any of the earliest stages of glomerular formation. The impression is readily gained from close scru- tiny of the electron micrographs that the lamina densa (definitive glomerular capillary basement mem- brane) forms only in close, intimate relationship with the smooth, continuous, endothelial cell surface membrane. It has not been found extending into or following the numerous folds of the podocytic cell surface membrane. Yet, it appeared rather frequently to extend or form to some degree between two endo- thelial cell surfaces within the walls of the primordial capillaries. These observations, along with the fre- quently observed close relationship of the lamina densa to endothelial cell nuclei, afi"ord support to the view that lamina densa formation is dependent upon the presence or activity of external surface membranes of the endothelial cells. The proximal surface membranes of the podocytic cells may form, or play a role in the formation of the lamina densa, but they often appeared much less intimately associ- ated with the developing lamina densa, than did the endothelial surface membranes. It is of interest to note that soon after its earliest appearance, the lamina densa had attained a width characteristic of the adult glomerulus. The basement membrane of Bowman's capsule in these same glo- meruli dig. 2) often appeared thinner than the lamina densa. In adult kidneys the basement membrane of Bowman's capsule usually appears much thicker than the lamina densa, often several times as thick. The continued increase in thickness of the capsular basement membrane, and its lamination and fibrous nature in fully matured Malpighian bodies appear to completely distinguish capsular basement mem- brane from lamina densa. The electron micrographs do not support the old belief that capillaries invaginate the blind end of a nephron, but tliey do show clearly that glomerular capillaries ditVerentiate in situ from an unorganized mass of prospective endothelial ceils, much as Her- ring (4) and ReinhofT(9) had found. Previously, Mall (2, 3) has shown clearly that the kmiina densa (definitive glomerular basement membrane) is the only basement membrane in the capillary wall, and in this study it is demonstrated that lamina densa forms only in direct and intimate contact with the surface membranes of prospective endothelial cells. It clearly follows that there can be no embryological basis for the old belief that a two-layered fold of Bowman's capsule extends from capillary to capil- lary, enclosing a mesangium and/or a limited inter- capillary space between limbs of imaginary capillary loops. If the preliminary data and interpretations reported here are fully substantiated by future stu- dies, then Gerlach's old idea of a glomerular perito- neum, and its modern counterparts, Zimmermann's "mesangium" and "intercapillary space" in the spe- cial sense of McManus and others, must be replaced by concepts of glomerular structure which are in accordance with the known embryological and anatomical facts. References 1. GtRLACH, J., Anil. Anal. Physiol, ii. wiss. Med., p. 378 (1845). 2. Hall, B. V., Proc. Vth Ann. Conf. on the Nephrotic Syndrome, p. I. Natl. Nephrosis Foundation, New York, 1953. 3. — Proc. Vlth Ann. Conf. on the Nephrotic Syndrome, p. 1. Natl. Nephrosis Foundation, New York, 1955. 4. HiRRiNG, P. T., /. Pathol. Bactcriol. 6. 459 (1900). 5. HuBiR, G. C, Am. J. Anal. 4, Suppl., I (1905). 6. — The Harvey Lectures, p. 100 (1909-10). 7. McManus, J. F. A., Am. J. Pathol. 24. 1259 (1948). 8. Paladh, G. E., /. E.xptl. Med. 95, 285 (1952). 9. RiiNHOiF, W. F., Johns Hopkins Hosp. Bull. 33, 392 (1922). 10. ViMTRUP, Bj., Am. J. Anal. 41. 123 (1^>:S). 11. Zimmirmann, K. W.,Z. mikroskop. anal, loisth. 32, 176 (1933). Further Studies on the Nephron Ultrastructure in Mouse: Terminal Part of Proximal Convolution J. Rhodin The Lahoratory for Biological Ultrastructure Research of the Department of Anatomy, Karolinska Institiitet, Stockholm Light and fluorescent microscopic studies (4) proved structural differences between different parts of the proximal convoluted tubule of the mouse nephron. Analyses of the nephron by means of electron microscopy (3, 7) dealt with the first part of the proximal convolution. The present report gives some data concerning the terminal part of the proximal convolution, and demonstrates characteris- tic features of its tubular cells, proving differences also at an ultrastructural level. White female mice were decapitated and within 2 minutes small pieces of kidney tissue were immersed in 1 per cent isotonic osmium tetroxide bufiered at pH 7.2 (1, 3, 5). The duration of the fixation was 60 minutes. After dehydration in alcohols and embedding in nietha- crylate, a Sjostrand Ultra Microtome was used for sec- tioning and an RCA EMU 2c electron microscope for examination of the sections. The tubular cells of the terminal part of the proxi- mal convolution have a height of 5-8 //, the value decreasing towards the first straight part of Henle's loop. In most cases, a free lumen is present which tallies with light microscopic findings after fixation with freeze-drying method. The luminal part of the cell is fitted with brush border extensions, all free from each other. This general arrangement of brush border extensions was described and illustrated with micrographs for the first part of the proximal con- volution (3). With the tubule dilated this arrange- ment is easier to interpret in the terminal part. The extensions are covered with a 60 A thick cell mem- brane. Between the bases of the brush border exten- sions are tubular invaginations of the surface mem- brane. However, they are less abundant than in the first part of the proximal convolution. The invagina- tions reach 0.1-0.2 /n into the cell. The bottom of the invaginations is usually slightly extended. The basal part of the cell faces the basement membrane with a 50 A thick plasma membrane. This membrane shows but few of the infoldings, /i-cytomembranes (6), so characteristic of the first part of the nephron. It has been noticed that there is a tendency for these folds to decrease both in number and height from the neck of the nephron towards the first straight part of Henle's loop. The scarcity of infoldings is a typical feature of the terminal part of the proximal convolution and facilitates the recognition of its tubular cells. The basement membrane is a homogeneous, 800 A thick, structureless layer. It is separated from the adjacent plasma membrane by a 180 A wide space. The mitochondria are few and of oval shape with a diameter some times exceeding 0.5 //. Their ultra- structure is identical with what has earlier been described in tubular cells (3,7). Microbodies with a diameter of 0.1-0.3 // are present with a single outer membrane, the thickness of which is 50 A. They lack inner membranes and thus can easily be distin- guished from mitochondria. A small Golgi apparatus can be found with pairs of membranes as the main component. Large opaque granules and large vacu- oles are present to the same extent as in the first part of the nephron. Their size is usually larger than 0.5 micron. Both large granules and the vacuoles are surrounded by a single membrane with a thickness of 50 A. No connections have been traced between large vacuoles and tubular invaginations. Thus, it Wtmlm ^^ ^ Fig. I. Brush border zone with a free lumen (L). At the bases of the brush border extensions (B) are widened tubular in- vaginations (V). In close connection a mitochondrion (M). Magnification 20,000. Further Studies on the Nephron Ultrastrueture in Mouse Fig. 2. Basal part of a cell from the terminal part of proximal convolution. Towards the basement membrane (BM) is seen the plasma membrane without infoldings, in places with a high density of the adjacent cytoplasm. Mitochondria (M) of round or oval shape. Some a-cytomembranes (arrow) with attached 160 A thick granules. Parts of Golgi apparatus (G) and nucleus (N) are seen. Magnification 30,000. does not seem probable that the tubular invagina- tions develop into vacuoles, which makes the sug- gestion by Pease (2) seem doubtful. The cytoplasm of the cells from the terminal part contains 160 A large granules arranged either singly, or in aggregations of 4-6. The granules are more scarce than in the first part of the proximal convolution, which gives the cytoplasm of the termi- nal part a less dense appearance. The granules have not been found in the brush border extensions, as claimed by Pease (2). There may be a difference between mouse and rat. Several high resolution electron micrographs of brush border extensions from the first part of the proximal convolution have been published (3), where no granules are to be seen. These findings indicate that the cytoplasm of the brush border differs markedly from that of the rest of the cell. This structural difference may be related to functional differentiation of the cells. In very rare cases short a-cytomembranes (6) with the typical attached 160 A granules occur in the cytoplasm. Occasionally, these may also be encountered in the cells of the first part of the proximal convolu- tion of the mouse kidney. This study of the mouse nephron has made it possible to correlate light and electron micro- scopic findings as far as the lumen of the terminal part of the proximal convolution is concerned. In both cases a wide lumen has been demonstrated, although two difterent methods of fixation were used. The question of a free lumen in the first part of the proximal convolution is not yet settled b> the statement of Pease (2) that a variation in cellular water content is responsible for the presence or absence of a free lumen. Of course, the present author does not deny that a free lumen exists //; vivo, but after fixation with both osmium tctroxide b\ immer- sion and by freeze-drying the lumina of the proximal convolution are aKva\s found to be closed, in con- tradistinction to this, using the same fixation meth- ods, the lumina of the terminal part of the proximal convolution are found to be open. According to Pease (2), a "hydration" of the tu- bule cells is responsible for the closing of the lumen when the common fixation technique of immersion is used. He believes that the cells "pick up" water and that they then should be regarded as "swollen" cells. However, if this be true, it seems likely that the density of the cytoplasm of a swollen cell should be less than that of a non swtillen cell. Still, in our 182 J. D. LEVER cases, the cytoplasm of the cells of the first part of the proximal convolution looks denser than that of the terminal part. A free lumen throughout both the first and termi- nal parts of the proximal convolution has been proved to exist after intravenous injection of phlori- zin as well as diodrast (Rhodin, unpublished obser- vations). Either injection caused a heavy diuresis. In these cases, the kidney was fixed according to the standard procedure described previously (3). The cells of the first part of the proximal convolution fixed after induction of diuresis show isolated brush border extensions, as well as "dome'' cells of normal appearance without brush border extensions. At the same time there is a clear-cut difference between cells of the first and terminal part of the proximal convolution as far as the density of the cytoplasm is concerned. In these diuretic kidneys, no "hydra- tion" (2) was found to close the lumen. The explana- tion is no doubt that the injected diuretics cause such heavy accumulation of water in the lumen that it expands the whole tubule. The importance of the technique used by Pease (2) is not to be questioned here. On the other hand, many facts prove that the common technique of immersing the tissue is also reliable, at least when a thorough knowledge of the normal light microscopic picture is employed as the basis for extended analyses with the electron micro- scope. What has to be stressed is the fact that not only a free lumen of the present investigated part is in most cases demonstrable but also characteristic cellular features like few, rounded mitochondria, poor den- sity of the cytoplasm, and few or almost absent in- foldings of the plasma membrane. This study has been supported through a grant from Stiftelsen Gustaf och Tyra Svenssons minne. References 1. Palade, G. E., /. Exptl. Med. 95, 285 (1952). 2. Pease, D. C, Anat. Record 121, 723 (1955). 3. Rhodin, J., Correlation of Ultrastructural Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Karolinska Institutet, Stockholm, 1954. 4. Sjostrand, F. S., Acta Anat., Siippl. 1, (1944). 5. — /. Cellular Coinp. Physiol. 42, 15 (1953). 6. — in Physical Techniques in Biological Research (Oster and PoLLisTER, eds.). Vol. Ill, p. 241. Academic Press, New York, 1956. 7. Sjostrand, F. S. and Rhodin, J., E.xptl. Cell Research 4, 426 (1953). The Fine Structure of Brown Adipose Tissue in the Rat: with Observations on the Cytological Changes Following Starvation and Adrenalectomy J. D. Lever Department of Anatomy, Cambridge University, Cambridge 1 HE main body of current opinion is that brown fat is merely a form of adipose tissue and not an endocrine gland as Cramer ( I ) believed. The works of Wertheimer and Shapiro (14) and Fawcett (3, 4, 5) suggest the active synthesis of glycogen and lipid within adipose tissue. In brown fat cells fuchsino- phile "secretion-like" granules have been described by Rasmussen (12) while Sheldon (13) believed in a direct transformation of mitochondria into lipid droplets. This paper presents a short light and electron microscopic study of brown fat in the normal, starved and adrenalectomised rat. Tissue for electron microscopy was fixed in Dalton's (2) dichromate- osmic solution for 45 mins., methacrylate-emhedded and sectioned at approximately 150 A. Normal brown fat — Light microscopy: The multi- locular lipid distribution in brown fat is well seen after Sudan black staining, the lipid droplets ranging in diameter from 0.5-7 p within any cell. It is con- sidered likely that fuchsinophile bodies, demonstra- ted by methyl green acid fuchsin staining, are mitochondria in oval or rounded shapes (0.5-1 // d.) grouped between and around the fat droplets. In size and position these fuchsinophile bodies exactly correspond to phospholipid-positive granules dem- onstrated by the Baker acid haematein stain. Normal brown fat — Electron microscopy: Apart from short invaginations into the cell, the plasma membrane is uncomplicated and is covered through- out its extent by a basement membrane. The endo- plasmic reticulum (10, 11) in brown fat is incon- spicuous and consists of a small saccular component, and a granular component (150 A d.) which is aligned along the outer walls of the saccular compo- nent or occurs independently in the cytoplasm. The nuclear membrane is bi laminar: there are no special features of the nucleoplasm. Throughout this study the Golgi apparatus was not convincingly demon- strated. Brown Adipose Tissue in f/ic Rai 183 b Fig. 1. Part of a normal brown fat cell. Mitochondria mostly have regular cristac some ot" which are beaded (B). Inter- cristal sections account for the appearance of such bodies as (A). Some mitochondria have vesicular internal membranes at one pole (V) and tiny vesicles are sometimes seen between the cristae (C). Lipid droplets lie freely in the cytoplasm and often interconnect. Endoplasmic reticular elements include a saccular component (S): the granular component (G) occurs independently, and in association with the saccular component. Magnification 61,000. The mitochondria are not markedly pleomorphic: although found anywhere in the cytoplasm they are commonly clustered around lipid droplets. From direct measurements their average cross-sectional diameter was calculated as 0.86 //. in some parts of its extent the mitochondrial enclosing membrane may be deficient, and often at points of contact with lipid droplets (fig. 3): osmiophile material may extend between droplet and mitochondrion at such a breach. Mitochondrial internal membranes consist for the most part of double membranes arranged in parallel cristae of varying length (tig. 1 ) but vesicular components (figs. 1 and 2) may be admixed with the cristae or concentrated towards one pole as observed in the parathyroid mitochondria (9). The intercristal distance may range from 250 to 850 A throughout the tissue, with a narrower variation within individual mitochondria. The thickness of the cristae (double membrane plus contained space) shows a sitnilar individual and general variation between 150 350 A. The broadest cristae are usually most widely separated and are frequently beaded Fig. 2. The mitochondrion (A) contains both cristae and vesicles. The body (B) contains intensely osmiophile vesicles and is intermediate in appearance between the mitochondrion (A) and the lipid droplet (L). Note vesicular components in lipid droplet (see fig. 4). Magnification 47,000. Fig. 3. The mitochondrion shown is freely open to lipid drops along the line A-A and at B. Magnification 47,000. 184 J. D. LEVER Fig. 4. In this field a vesicular component can be seen at (A) and is partially masked by the osmiophile material in the Jipid droplet. Magnification 47,000. Figs. 5-6. Fig. 5 is a small representative field of a normal brown fat (control) cell while fig. 6 is a comparable field from the same animal after 7 days starvation. There is a marked increase in the number of mitochondria per unit area during starvation (see text). Both figs. 5 and 6, magni- fication 24,000. (figs. I and 2). Between the internal membranes is a semiopaque matrix material and should the plane of section be intercristal the entire mitochondrial outline is then occupied by this matrix. In addition to obvious lipid droplets and mito- chondria, other bodies of an intermediate appearance are frequently found in proximity to the lipid drop- lets. There is evidence of an enclosing membrane to some but not all of these bodies, which resemble mitochondria in size and shape and contain intensely osmiophile internal vesicles (fig. 2). Fat droplets are unconfined by a membrane and lie freely within the cytoplasm (fig. 1). They are often seen to contain vesicular elements (fig. 4) resembling those already described within the "intermediate" bodies. Changes in starvation and after adrenalectomy: The reaction of brown fat is similar in both these circum- stances but the observed changes are more profound and of more rapid onset in starvation. These changes are: {a) A reduction in total lipid within the tissue and a concurrent diminution in droplet size as judged by light and electron microscopy. After 7 days starvation and in 21-28 days after adrenalec- tomy, there is: (/>) a reduction in cell volume (esti- mated with the light microscope by direct measure- ments of cross-sectional cell diameters) of from 25- 50 and (c) a two-fold (approximately) increase over normal in the number of mitochondria per unit area of fat cell (figs. 5 and 6). This last estimate was based on direct counts in electron micrographs of similar magnification. Discussion. — It is of considerable significance that the size range, as measured by the light microscope, of \-\ /I for the phospholipid-positive bodies and fuchsinophile bodies (mitochondria) and the smallest sudanophile droplets, should be comparable to mito- chondrial average size (0.86 /n d.) as estimated from electron micrographs. It is conceivable that the beading (figs. 1 and 2) of mitochondrial cristae in brown fat cells may precede Iheir fragmentation into filaments or vesicles. This would account for the mixture of these two types of internal membranes in many mitochondria. It has been suggested that lipid appears within the mitochon- dria of the adrenal cortex (6), the corpus luteum (7) and possibly the parathyroid (9). From the evidence of the present investigation it is considered probable that mitochondrial internal membranes in the brown fat cell become modified from a predominantly crystal to a vesicular form. The appearance of lipid within the mitochondria is concurrent with this modification. As already stated the lipid droplets within brown adipose tissue lie freely within the cytoplasm. There- fore if, in fact, lipid does appear within the mito- chondria then these bodies either discharge it freely into the cytoplasm through breaches in their limiting membranes or these membranes disintegrate around a contained droplet. Deficiencies in mitochondrial enclosing membranes have previously been reported (8) in the rat and hamster adrenal cortices. The series of events within mitochondria that has been described results in considerable alteration of their fine structure and, in all probability, in their complete destruction. An expendibility of mitochondria such as is postulated has already been described in the adrenal cortex (8). The observed increase in the number of mitochondria during starvation and following adrenalectomy must be viewed in the light of a concurrent cell shrinkage. In ordinary white adipose cells the cytoplasm and nucleus are displaced peripherally by the mass of fat and the crescentic shape of the nucleus suggests some degree of com- pression. In brown adipose tissue, if it be assumed that owing to the presence of intercellular fat droplets the other cell constituents are displaced or "com- pressed"; then, if most of this fat material is released from the cell, as occurs for example during starva- tion, the other cell constituents will become less "compressed". The explanation does not take into account the elasticity of the plasma membrane nor does it allow for any other change of state within Brown Adipose Tissue in the Rat 185 the cell which might accompany lipid release. Ar- guing thus, and making these assumptions, it would be expected that if the cytoplasm is less "compressed" in the brown fat cell of starvation, then the popula- tion density of mitochondria would decline, in fact there are many more mitochondria per unit area than in the normal controls and from what has been said it is reasonable to postulate an actual increased production of these bodies during starvation and following adrenalectomy. 1 am indebted to Professor .1. D. Boyd for his interest and criticism and to Dr. V. E. Cosslett for electron microscope facilities at the Cavendish Laboratory, Cam- bridge. Rrri RtNCES 2, 3 4 5 6 7, 8 9 10 I I 12 13 14 Cramer, W.. Brit. J. E.xptl. Patlwl. 1, 184-195 (1920). Dai.ton, a. J., Anat. Rei. 121, 281 (1955) (abstract). Favvcett, D. W., Svieme 105, 123 (1947). — Anat. Rec. 103, 450 (1949) (proc). ^ J. Morplwl. 90, 363-406 (1952). Lever, J. D., Ainer. J. Anat. 97, 409-430 (1955). Anat. Rec. 124, 111-126 (1956). — /. Biopliys. Biodieni. Cytol. 2, Suppl., 31 3-31 8 (1956). — J. Anat. 91, 73-81 (1957). Paeade, G. E. and Pokier. K. P.. J. Expil. Med. 100, 641-655 (1954). Porter, K. R., J. Lxptl. Med. 97, 727-749 (1953). Rasmussen, a. T., J. Morplwl. 38, 147-193 (1923). .SiuEOON, E. F., Anat. Rec. 28, 331-343 (1924). Wertheimer, E. and Shapiro, B., Pliysiol. Rev. 28, 451- 464(1948). VII NERVE CELLS AND RECEPTORS Neurofilaments et neurofibrilles dans les fibres nervetises cle la Sangsue R. COUTEAUX Lahoratoire de Biologic aniiimle (P.C.B.) , Faciilte cles Sciences, Paris, et Laboiatoire de Micioscopie electroiiique appliqiiee a la Biologic (C.N.R.S.) BiEN avant F introduction du microscope electro- nique en Cytologic nerveuse, le probleme de la nature des neurofibrilles avait deja fait Pobjet de nombreux debats, mais la mise en evidence au cours de ces dernieres annees, a Taide de ce nouvel instrument, de neurofilaments submicroscopiques a nettcment fait prevaloir Topinion que les neurofibril- les se forment au cours de la fixation par I'agglutina- tion en faisceaux de ces neurofilaments. Cette inter- pretation, quoique tres probable, a encore besoin d'etre etayee par des observations positives. Les neurofibrilles des cellules nerveuses de la Sangsue que les recherches d'Apathy, Cajal et Nageotte ont rendues depuis longtemps classiques nous ont paru pour aborder ce probleme un materiel particuliere- ment favorable et nous exposerons ci-dessous, tres sommairement, les premieres observations que nous avons pu faire sur la chaine nerveuse ventrale de cet animal. Materiel et technique. — C'est principalement sur les fibres nerveuses, amyeliniques, de la Sangsue (Hiriido medicinalis) qu'ont porte nos observations, les neurofib- rilles presentant a ce niveau une orientation particuliere- ment simple qui facilite beaucoup I'interpretation des coupes ultraminces. Ces fibres ont ete etudiees a la fois au niveau des ganglions et des connectifs, sur coupes transversales et longitudinales. La fixation a ete effectu- tuee par une solution de tetroxyde d'osmium a 1 %, tamponnee au pH 7,3-7,5 selon Palade (I). Une fois mise a decouvert par une dissection prealable, sur des animaux maintenus en extension, la chaine ventrale a ete d'abord fixee /'/; sitn, puis une portion de cette chaine, detachee de Tanimal, a ete placee secondairement dans le fixateur pour une duree de 2 a 3 heures en general, mais aussi dans quelques cas pour des durees plus longues, atteig- nant jusqu'a 24 heures. Les pieces ont ete enrobees au methacrylate et coupees a Tultramicrotome Porter-Blum. La plupart des observations ont ete faites sur un micro- scope RCA du type EMU 3A et certaines d'entre elles sur un microscope RCA du type EMU 3B. En vue de per- mettre une comparaison directe des aspects neurofibril- laires observes au microscope ordinaire et de ceux que fournit le microscope electronique, des impregnations metalliques ont ete executees sur les animaux des memes lots en recourant aux methodes a I'argent reduit utilisees par Cajal sur le meme materiel. Observations. — Sur une chaine ventrale les aspects que presente Taxoplasme des fibres nerveuses peuvent etre d'une fixation a Tautre ou meme d"un point a I'autre de la meme piece tres sensiblement differents. Nous n"envisagerons ici que les deux plus tranches d'entre eux. II est frequent de n'observer dans ces fibres, qu'elles soient de grand ou de petit diametre, aucune formation filamenteuse disposee longitudi- nalement et vraiment significative. Dans ce cas, en dehors des chondriosomes, generalement assez abon- dants, parfois tres allonges et orientes longitudinale- ment, et de corpuscules arrondis de taille et de densite variables, d'une interpretation actuellement difficile, on ne trouve dans I'axoplasme qu'un reseau grele et tres irregulier. Notons que De Robertis et Bennett (2, 3) usant du meme fixateur sur la chaine ventrale d'un autre annelide (Helodrilns caliginosa) a souligne la particuliere pauvrete des cellules nerveuses de cet animal en « neurofilaments ». L'autre aspect de Laxoplasme se rencontre beau- coup plus rarement que le precedent dans les pieces fixees normalement. Aux constituants de I'axoplasme mentionnes dans le premier s'en ajoutent ici d'autres. 11 s'agit de nombreux filaments nettement orientes longitudinalement, d'une longueur indefinie (il a ete possible de les suivre individuellement sur des distances superieures a 2 /n dans le cas de fibres fixees en forte extension, ou les filaments deviennent absolument rectilignes), d'un calibre uniforme d'en- viron 200 A. Ces filaments, de forme cylindrique, se montrent en general plus denses a leur peripheric que dans leur region axiale, mais il parait difficile sur ce type de preparations de decider si cette diffe- rence de densite traduit reellement une particularite de leur constitution. Une coupe transversale prati- quee dans une fibre nerveuse contenant ces filaments montrc qu'ils peuvent etre distribues de maniere assez uniforme dans toute I'etendue de I'axoplasme. Cette distribution se trouve au contraire complete- ment bouleversee dans d'autres fibres par I'appari- tion d'une ou plusieurs formations allongees, rela- tivement volumineuses, tres fortement osmioreduc- trices, disposees parallelement a I'axe des fibres et auxquelles les filaments voisins paraissent s'agreger. Le calibre de ces formations varie de I'une a l'autre, mais reste en general de I'ordre du dixieme de micron. Elles rappellent en tous points les neurofibrilles, telles que les revelent chez la Sangsue les methodes argen- tiques de Cajal, tres souvent reduites a une seule dans le cas des petites fibres. Les filaments submicroscopiques dissemines dans I'axoplasme que nous venons de decrire se ren- contrent assez rarement dans les pieces de chaine ven- trale de Sangsue fixees normalement. Leur apparition s'accompagnant en general d'alterations plus ou moins importantes d'autres structures, notamment du chondriome, nous avons ete amene a y voir une anomalie de la fixation et a varier systematiquement les conditions de la fixation en vue de deceler la cause de cette anomalie. Le fait qui nous parait pour I'instant le plus propre a elucider ce probleme, c'est Les fibres nerveuses dc la Sangsue !89 qu'en faisant preceder, par exemple de 5 minutes, Taction du tetroxyde d'osmium tamponne sur la chaine ventrale par celle de la solution ioniquement non bahincee qui sert de tampon (tampon de Micha- elis a I'acetate-veronal), on peut mettre en evidence de fagon tres aisee ces filaments submicroscopiques, mais naturellement au prix d'alterations plus ou moins graves des autres structures (fig. 1-5). Les images obtenues dans ces conditions permettent neanmoins d'analyser avec plus de precision le pro- cessus d'agglutination par lequel se constituent, au niveau de nombreuses fibres et selon un determinisme qui reste a elucider, des formations longitudinales du Fig. 1-5. Coupes de fibres nerveuses au niveau de la chaine ventrale de la Sangsue. Fixation precedee d'un traitment de 5 min. par le tampon de Michaelis au veronal-acetate au meme pH (concentration 0.028 A/). Fig. 1. Coupe transversale d'une grosse fibre nerveuse a un niveau tres proche du pericaryon dont elle est issue, ne presentant que des alterations relativement discretes; chon- driome et « neurofilaments » apparaisseiit distribues de maniere a peu pres uniforme. (14.000 .) Fig. 2. Coupe longitudinale d'une petite fibre interessant en quatre points de son Irajet une neurolibrille sinueuse. (18.000 ■:.) Fig. 3-5. Coupes longitudinales de petites fibres montrant le processus d'agglutination des neurofilaments et la constitu- tion de neurofibrilles fortement osmioreductrices. (Fig. 3, 43.000 ; fig. 4, 46.000 x ; fig. 5, 57.000 .) 1 190 HEDI GANSLER type neurofibrille. Si Ton substitue au traitement prealable par le tampon de Michaelis un traitement par Teau distillee, on obtient en dehors d'un gonfle- ment plus marque du chondriome une accentuation du phenomene d'agglutination et sa generalisation a toutes les fibres. BiBLIOGRAPHIE 1. Palade, G. E., /. Exptl. Med. 95, 295 (1952). 2. DE RoBERTis, E. D. P. et Bennett, H. Stanley, //; The Structure of Cells (Symposium Leiden 1954), p. 261. Noordlioff Publ., Groningen, 1955. 3. — /. Biophys. Biochein. Cytol. 1, 47 (1955). Elektronenmikroskopische Untersiichiingen an Grenzstrangganglien von menschlichem Operationsmaterial Hedi Gansler Rheinisch- Westfalisches Iiistitiit fiir Uheniiikroskopie, Diisseldoif Von Patienten, bei denen wegen peripherer Durch- blutungsstorungen eine Sympathectomie vorgenom- men wurde, entnahmen wir wiihrend der Operation Ganglienmaterial, das 1 Stunde in I'^oiger Osmium- saure nach Palade fixiert und in der iiblichen Weise eingebettet wurde. Wir gingen aus von den Ergebnissen der klassi- schen Histologic, wonach die priiganglionaren Fasern in Form von sog. Endkolben oder Endknopfen in direkten Kontakt treten soUen mit der Nervenzell- membran der postganglionaren Fasern. In dem von uns untersuchten Material konnten wir nun weder auf dickeren Schnitten im Phasenkontrastmikroskop noch auf Dunnschnitten im Elektronenmikroskop solche Kontaktstellen zwischen Zellmembranen und Nervenfasern beobachten. Das Cytoplasma der Nervenzellen ist stark vakuolisiert, d. h. das endo- plasmatische Retikulum zeigt sackformige Auswei- tungen. Die Mitochondrien sind meist hell ge- schwollen und lassen eine Innenstruktur weitgehend vermissen. Da in Gliazellen und in einigen Nerven- fasern normale kleine Mitochondrien zu sehen sind, mochten wir annehmen, daB es sich bei den beschrie- benen Nervenzellformen um intravital entstandene Veranderungen handelt. Im Cytoplasma ungeordnet verteilt sind kleine Fibrillen von etwa 40-50 A zu sehen, wie sie auch in Nervenfasern, nur viel dichter und immer zur Liingsachse der Faser orientiert, vor- kommen. AuBerdem liegen im Cytoplasma noch Anhaufungen von kleinen kreisformigen Membranen mit einem Durchmesser von etwa 400 A. Jede Ner- venzelle ist von Gliazellen umgeben, die im allge- meinen ein helleres, strukturarmeres Cytoplasma be- sitzen, das aber auch Fibrillen und kleine runde Membranen aufweist. Im iibrigen sieht man ausgedehnte Areale von markarmen Nervenfasern, die in alien moglichen Richtungen angeschnitten und immer von einer Gliahiille umgeben sind. Wie erstmalig von Gasser (1) beobachtet, sind die Nervenfasern ahnlich in Gliataschen eingebettet, wie die inneren Organe in das Peritoneum: An einer Stelle entsteht eine sog. Duplikatur der Gliazellmembran, die als ,,Mesaxon"' bezeichnet wird. Dadurch entsteht ein groBflachiger ultrakapillarer Spalt zwischen Nerven- faser und Extrazellularraum. Die die Nervenfaser um- hiillende Doppelmembran ist also nicht einheitlichen Ursprungs: die innere Membran wird von der Ner- venfaser selbst gebildet, wiihrend die iiuBere Mem- bran eine Gliazellmembran ist. Die Nervenfasern zeigen entweder die gleiche Struktur wie die vorher beschriebenen Nervenzellen oder sie bestehen vor- wiegend aus dicht gelagerten langs orientierten Fibrillen von etwa 40-50 A Durchmesser. Dazwi- schen vereinzelt Vakuolen und kleine runde Mem- branen. Es ist anzunehmen, daB es sich hierbei um zellferne Nervenfasern handelt. — Eine Unterteilung in Neuriten und Dendriten ist auf Grund unserer morphologischen Befunde nicht moglich. AuBer diesem typischen Bild der markarmen Ner- venfasern lieBen sich an unserem Material noch folgende Beobachtungen anstellen: 1 . Wahrend im allgemeinen zwei in einer Gliazelle eingebettete Nervenfasern immer durch eine Glia- briicke voneinander getrennt sind, sahen wir ofters eine zellnahe und eine zellferne Nervenfaser in direktem Kontakt miteinander, d. h. nur durch ihre eigenen Membranen voneinander getrennt; die auBere Membran, die von der Gliazelle gebildet wird, umfaBt beide Nervenfasern gemeinsam. In der zellfernen Nervenfaser sind an solchen Stellen besonders reichlich kleine runde Membranen, ,, sy- naptic vesicles" (2), zu sehen, wahrend die zellnahe Faser an solchen Kontaktstellen oft feinste rohr- chenformige Membranausstiilpungen besitzt. 2. An einzelnen Stellen erreicht die Nervenfaser das Niveau der Gliazelloberflache dergestalt, daB das „Mesaxon" auseinander gedrangt zu werden scheint und so eine breite Kontaktflache der Nerven- faser mit dem Extrazellular-Raum zustande kommt. 3. Auffallend ist ein zwischen den Gliazellen ausge- bildetes StraBen- bzw. Kanalsystem, das sich bis tief in die Gliazellen hinein verfolgen liiBt und an einzelnen Stellen bis an die Nervenfaser heranreicht. Es ist meist von einer homogenen Grundsubstanz ausgefiillt. Auf Flachschnitten durch Gliazellen ist Grenzstrangganglien von uwnsc/iliclicm Operationsmaterial 191 Abb. 1. Mehrere Gliazellanschnitte (G). In der Milte oben Knotenpunkt (K) eines exirazelluliiren Kanalsystems, dessen feinere Auslaufcr zwischen und in den Gliazellen eine homogene Grundsubstanz enthalten. Links iinten ein Axon (Al), das durch Auseinanderschieben des Mesaxon eine breite Kontaktflache zum Extrazellular-Raiim herstellt. Rechts unten eine weitere Nervenfaser (A 2), die auBer dem Mesaxon noch ein bis an das Axon heranreichendes Kanai- chen erkennen liiBt. VergroBerung: 25 000 Abb. 2. Zwei direkt aneinanderliegende Nervenfasern (A 1 und A2), die eine gemeinsame auBere Doppelmembran haben. VergroBerung 16000 ■< . dieses vielverzweigte und sich oft Uberschneidende StraBensystem besonders schon zu sehen. Unsere Befunde sprechen dafiir, daB eine Verbin- dung der prii- mit den postganglionaren Fasern auf mehrfache Weise inergestellt werden kann, wobei den die Nervenfasern umhiillenden Gliazellen eine be- sondere Bedeutung zuzukommen scheint. Es handelt sich hierbei um ini Verhiiltnis zu ihrem Kern auffal- lend groBe helleZellen, deren Oberfiachen amoboid- artige Cytoplasmaauslaufer besitzen, die auf Schnit- ten z. T. als zahlreiche kleine Zellanschnittc impo- nieren. LiTERATUR 1. Gasser, H. S., Cold Spring Harbor Symp. Quant. Biol. (1952). 2. DE RoBERTis, E. und Franchi, C. M., J. Biophys. Biochem. Cytol. (1956). Ultra-Structure des cellules visuelles du Gecko. Mise en evidence de prolongements cytoplasmiques infra-microscopiques au niveau du segment interne Nina Carasso Lahoicitoiie de Microscopie Electronique appliqiiee a la Biologic, Centre Ncilional de la Recherche Scientifique, Paris La retine du Gecko nocturne Tarentola mawitanica comprend uniquement des cellules visuelles a baton- nets. Ceux-ci sont de deux sortes : batonnets simples, de type classique, et batonnets doubles, formes d'un batonnet principal accole a un batonnet accessoire. A. Rochon-Duvigneaud (5), S. R. Det- willer (2), G. L. Walls (9) ont etudie la retine du Gecko en microscopie optique. La microscopie elec- tronique nous a permis d'y reconnaitre des caracteres nouveaux. 1° Description des cellules visuelles. Le segment externe des batonnets est forme de disques empiles dans lesquels nous n'avons pu reconnaitre deux membranes separees par un espace clair comme cela est le cas chez le Cobaye (6) et le Lapin (4). Seule la peripheric des disques comporte une sorte de gouttiere qui, sur coupe, a la forme d'un chas d'aiguille (fig. 1). Dans le cytoplasme se trouvent differentes encla- ves : tout d'abord rellipsoide, forme de mitochon- dries de tres grande taille, tassees les unes contre les autres (7); le paraholoi'de, homogene, separe du cytoplasme sous-jacent par une membrane; puis epars dans le cytoplasme situe au dessous du pa- raboloide, de fins granules accoles ou non a de minces membranes, que Ton peut assimiler a /Va;;^^^- toplasine ou reticulum endoplasmique. Cette maniere de voir se trouve confirmee par la forte basophilic du cytoplasme, surtout au voisinage du paraboloTde et du noyau, basophilic qui disparait apres action de la ribonuclease. Signalons enfin, dans la partie toute proximale du cytoplasme, au voisinage du noyau, la presence de doubles membranes et de grosses vesicules osmio- philes, que Ton peut sans doute rattacher a un appareil de Golgi assez atypique, dans lequel la partie lamellaire est des plus reduite et les vesicules osmiophiles, au contraire, tres nombreuses et hyper- trophiees. 2° Description des prolongements cytoplasmiques. La particularitc la plus marquantedes cellules visuelles du Gecko est cependant la presence, tout au long du segment interne et de la zone perinucleaire, de prolongements cytoplastniques ( 1 ) pouvant affecter une longueur assez considerable. Ces prolongements sont tout a fait independants du prolongement ter- minal classique qui met en rapport les cellules visuelles avec les cellules bipolaires sous-jacentes selon un mode synaptique qui a ete etudie en micro- scopie electronique par Sjostrand (8) et de Robertis (3, 4). Les prolongements, sur coupes transversales ou longitudinales, apparaissent a peu pres equidistants separes les uns des autres par des distances de 300 a 400 m/< (fig. 2 et 3). Dans certaines cellules a cyto- plasme par ailleurs beaucoup plus osmiophile, ces distances peuvent etre moindres (moins de 200 m^). Sur coupes fortement obliques, la longueur des pro- longements cytoplasmiques est beaucoup plus grande que sur coupes transversales. Leur orientation semble done tres oblique par rapport a Taxe des cellules visuelles. Sur coupes obliques passant au niveau de la limitante externe de la retine, on voit que les prolongements ne traversent pas cette limi- tante et s'arretent tout pres d'elle, sans cependant entrer en rapport avec elle. ,..,«-j;-; Fig. 1. Coupe d"iin segment externe de batonnet. Les disques sectionnes presentent, en peripherie, un renflement en chas d'aiguille. 80.000. Les cellules visuelles du Gecko 193 f r « • ■jz r- db X db .1 .- ) > ^ V :i^^:l_ ^ Fig. 2. Coupe transversale de batonnets retiniens de Gecko. Entre les deux fleches, les deux doubles batonnets voisins db contractent entre eux des rapports directs d"apposition. m, fine membrane de I'element interstitiel; de part et d'autre de c, aspect de canalicule, entre deux membranes. 30.000. Fig. 3. Coupe tres oblique de batonnets retiniens. Les vesi- cules de Telement interstitiel aflectent ici une ordonnance tres reguliere le long des prolongements cytoplasmiques. N, noyau d'un batonnet: /, limitante externe. 25.000. Les cellules visuelles et leurs prolongements sent limites par une membrane dont le contour apparait sur les micrographies comme etant simple et forte- ment osmiophile. Ce contour apparait souvent double par un lisere externe peu osmiophile et dis- continu qui appartient non pas aux cellules visuelles ma is a un systeme d'elements interstitiels complexe. Entre les cellules visuelles et leurs expansions se trouvent en eflFet des vesicules de taille diverse, de minces membranes delimitant des canalicules (fig. 2, de part et d'autre de c), ou s'appliquant contre le contour externe osmiophile des photoreceptcurs. II s'agit la vraisemblablement de formation nevro- gliques,en rapport peut-etre avec les cellules de M tiller qui, dans la retine, sont excessivement ramifiees. Les vesicules, de forme plus ou moins ellipsoidale selon I'incidence de la coupe, afTectent souvent sur les micrographies une ordonnance remarquablement reguliere (fig. 3). La presence de prolongements cytoplasmiques dans les cellules visuelles du Gecko, et les rapports qu'ils contractent entre eux et avec les elements in- terstitiel entrainent un certain nombre de conse- 13 — 568204 Electron Microscopy quences sur lesquelles nous attirerons a present Tattention. A. De la sorte se trouve realise un accroissement considerable des surfaces d'echange au niveau du cytoplasme des photoreceptcurs. B. Les rapports des plwtorecepteurs avec I'element interstitiel sont de deux sortes : il y a d'une part accolement etroit, a certains niveaux, du contour externe osmiophile des cellules visuelles aux minces membranes de Telement interstitiel (les cellules vi- suelles semblent alors, a premiere vue. limitces par une membrane a double contour a ces niveaux); d'autre part, il existe un alignement tres regulier des vesicules interstitielles le long des expansions cyto- plasmiques (la fig. 3 est a cet egard suggestive). C. Des rapports semblent exister entre cellules visuelles dans la zone des prolongements. Les pro- longements appartenant a des cellules visuelles voi- sines sont imbriques les uns dans les autres, ce qui apparait bien dans la partie droite de la fig. 2. Dans la tig. 3, la presence de files regulieres de vesicules interstitielles reliant entre elles les deux portions de cellules interessees par la coupe montre egalement 194 F. S. SJOSTRAND AND L. G. ELFVIN qu'il y a imbrication des prolongements longes par les vesicules. Ici, les prolongements n'ont pas ete interesses sur toute leur longueur. Un autre fait apparaissant sur la fig. 2 merite d'etre souligne : en certains points (region comprise entre les deux fleches), des cellules visuelles voisines peuvent contracter, au niveau des prolongements, des rapports directs d'apposition, sans interposition de Telement interstitiel. De tels rapports sont tres comparables a ceux observes, en microscopic elec- tronique, au niveau des synapses centrales. BiBLIOGRAPHIE 1. Carasso, N., Compt. rend. acad. sci. 242, 2988 (1956). 2. Detwiller, S. R., /. Comp. Neurol. 36, 125 (1923). 3. DE RoBERTis, E., Acta Neurol. Latiuoamericana 1 (1955). 4. — /. Biopliys. Biochein. Cytol. 2, 307 (1956). 5. RocHON-DuviGNEAUD, A., Aiiii. OcuHst. p. 16, (1917). 6. Sjostrand, F. S., /. Cellular Comp. Physiol. 42, 15 (1953). 7. — ibid, 42, 45 (1953). 8. — Z. wiss. Mikroskop. 62, 65 (1954). 9. Walls, G. L., Amer. J. Neurol. 17, 892 (1934). Some Observations on the Structure of the Retinal Receptors of the Toaci Eye as Revealed by the Electron Microscope F. S. Sjostrand and L. G. Elfvin The Laboratory for Biological Ultrastriictiire Research of the Deportment of Anatomy, Karolinska Institutet, Stockholm In earlier electron microscopic studies the receptor cells of the retina have been studied in guinea pig and perch eyes (4, 5) and in albino rat and mouse eyes (1, 2). In this paper the preliminary results are reported from a study of the rods and cones of the toad {Bufo bufo) retina. The retinas from adult animals were fixed in a buffered, isotonic 1 % osmium tetroxide solution (pH 7.4) at - 4 C. Embedding in methacrylate and sectioning on a Sjo- strand ultramicrotome. The receptors of the toad retina consist of rods, and single as well as double cones. The present analysis deals with the outer and inner segments. The rods. — The outer segment consists of a stack of double membrane disks similar to those described in the guinea pig and perch retina (3, 4). They differ, however, from the latter as to dimensions and the number of incisions extending from their periphery towards the center. The stack of disks is enclosed in a membrane representing the plasma membrane. The dimensions of the double membrane disks are: base diameter 5-6 //, total thickness 1 10 A. The osmiophilic layers of the two constituent mem- branes are about 40 A thick and the less osmiophilic layer separating the osmiophilic layers is about 30 A thick. The average spacing is 190 A. The inner segment contains the ellipsoid which occupies most of its distal part. The ellipsoid consists of densely packed mitochondria as in the perch retinal receptors (7). A Golgi apparatus has been observed just proximally to the ellipsoid. The outer and inner segments are connected through a number of fibrillar structures which are in contact with the plasma membrane of the outer segment and penetrate into the inner segment late- rally to the ellipsoid (fig. 1). The cones. — The outer segment of the single cone cell consists of a stack of double membrane disks similar to those of the rod outer segment. However, these disks show no incisions. A cilia-like structure similar to that described in the guinea pig retina (5) seems to constitute the only direct connection be- tween outer and inner segment. The outer segment is bounded by a plasma membrane. A number of cytoplasmic extensions from the inner segment sur- round the basal part of the outer segment. Fig. 1. Part of an oblique to transverse section through a rod outer segment. The fibrils, the surrounding membrane, and the outer segment incisions are easily seen. Magnifica- tion 23,000. Inset: Higher magnification of a fibril. Magnification 49,000. The Structure of the Retinal Receptors of the Toad Eye 195 Fig. 2. Longitudinal section of a proximal cell outer segment in a double cone. The double membrane disks are well demonstrated. Magnification 58,000. In the double cones we may distinguish between one outer segment with a more proximal location than the other. The proximal outer segment is larger than the distal one and reminds of a rod outer seg- ment. It is, however, lacking the incisions character- istic for the rod outer segment. The distal outer seg- ment resembles that of the single cones. In the single cones the dimensions of the double membrane disks are: total thickness 230 A, base diameter 2-3 //. The osmiophilic layers are 70 A thick and the interposed, less osmiophilic layer 95 A thick. The average spacing is 490 A. In the proximal cell of the double cones the dimen- sions of the double membrane disks are: total thick- ness 130 A, base diameter about 2 //. The average spacing is 220 A (fig. 2). In the /////('/• segments of the single cones a strongly osmiophilic oil droplet may occur. It is almost completely surrounded by mitochondria. In some of the mitochondria located in contact with the oil droplet a system of 8-10 concentrically oriented and very densely packed double membranes may be observed. The proximal component of the double cones contains distally in its inner segment an ellipsoid consisting of mitochondria. Proximally to the ellip- soid the well delimited paraboloid is observed. It consists of a homogeneous ground substance in which vacuoles and vesicles are present (fig. 3). The cytoplasm of the distal component proximal to the ellipsoid contains irregularly outlined vacuoles of varying sizes. No paraboloid is present. The inner segments of the two components of the double cones are in close contact with the osmiophilic layer of their plasma membranes separated by a less osmio- philic interspace the thickness of which varies from 60 to 240 A (fig. 4). This arrangement is similar to f' > f^ ^^y^lfj J ^ %^ ■'.' >•, t 1^ A: Fig. 3. A double cone. The inner segment of the proximal cell and part of the inner segment of the distal cell. Notice the ditVerence of the cytoplasm in the two cells. Magnifica- tion X 5600. that of the plasma membranes at retinal synaptic contacts (6). Electrophysiologic studies by Svaetichin (8) per- formed on twin cones in the fish retina have given support to the interpretation of twin cones as representing a system which is stimulated by two complementary colors. The contact surface between the inner segments has been interpreted by Svae- tichin to act in a similar way as a synapse. The morphology of the double cones of the toad retina makes such an assumption quite justifiable from a morphologic point of view. ^i " f V " .^ 'l/M n Fig. 4. The plasma membranes between the cells in a double cone. To the left the proximal cell with a part of its mitochon- dria aggregation and the paraboloid structure. Magnifica- tion 58,000. 196 References G. LION, C. MAERTENS AND G. VANDERMEERSSCHE 1. De Robertis, E., /. Biophys. Biochem. Cytol. 2. 319 (1956). 2. De Robertis, E. and Franchi, C. M., /. Biophys. Biochem. Cytol. 2, 307 (1956). 3. Sjostrand, F. S., /. Cell. Comp. Physiol. 33, 383 (1949). 4. — ibid. 42, 15 (1953). 5. — ibid. 42, 45 (1953). 6. — /. Appl. Phys. 24. 1422 (1953). 7. — unpublished. 8. SvAETiCHiN, G., Acta Physiol. Scand. 39, Suppl. 134. 17^6 (1956). Submicroscopic Morphology of the Retinal Pigment Epithelium G. Lion, C. Maertens and G. Vandermeerssche Centre de Micioscopie Electronique, Medical, Industrie! et Agricole, Briissels-Uccle The aim of the present work is to bring out the possibilities and Hmitations of the electron micro- scope in a study of the fine structures of the eye. One of the present authors has done some work with the Philips electron microscope Type EM 100 (1); this work was done in collaboration with the Chnique Ophtalmologique of the University of Ghent and was solely concerned with suspension and replica techniques. A comprehensive study of the different ocular tissues, but this time by means of the thin-sections technique, has been undertaken by the present authors. Two different types of electron microscopes are being used: the Philips EM 75 and the new type EM 100. The work done with the first electron micro- scope will be presented here while the rest of this work will be published elsewhere. Material has kindly been provided by the Institut d' Hygiene et d'Epidemiologie of the Ministry of Public Health, Brussels. The eyes of the following animals have been studied: monkey (M. Cynoiuol- gus), guinea pig and pig. It has thus been possible to bring out differences in structural details of the eyes of these animals and those studied by several other workers (1-4, 9-12) in this field (man, guinea pig, carp, perch, ox, frog). Our investigations were carried out on individual pigments which were fixed and embedded after isolation and on tissue fragments which were treated by the usual fixation and staining techniques. The isolation of the pigments has been described earlier (2, 3). They have been washed by repeated centrifugation and fixed in buf- fered 1 "^'o OSO4 solution (7). As embedding medium (6) a 6-to-4 mixture of butyl and methyl methacrylate has been used. This mixture seems to give the best results with an extremely hard material such as the pigments. Extreme care had to be taken during the preparation of the retinal pigment epithelium itself: it is almost impossible to avoid retinal detachment. The only way which has given rea- sonably good results is to fix the posterior hemisphere as a whole after the eye has been opened at the front end, without removing the complete content of the vitrous body. The orientation of the tissue blocks is made possible by a technique described in a paper by Ruska, Stuart, Winsser (8), and which consists mainly in keeping the embedded material during polymerisation in an atmos- phere of CO2 inside a test tube in the desired position. The fixation technique described above does not give complete satisfaction in so far as the exact fine structure of the rods is concerned, because of the presence of the remaining vitrous body (showing up the fixation of the retina), but it allows at least the observation of the distri- bution of the pigments in relation to the rods. The ave- rage thickness of the sections obtained with a Porter- Blum microtome was about 250 A. The pigments of the eyes of the monkey, pig and guinea pig show in sections the same typical elonga- ted forms with rounded extremities as those which have been observed in suspension preparations (1-4) (figs. 1-2). So far, only the pigments of carp-eyes have shown a marked difference in shape and size. The pigments of the eyes investigated here are usually grouped around some cellular residues but in a much smaller quantity than what has been observed in ox-eyes (1-4). Also the mosaic distribu- tion described in the ox-eye has not been observed here. Although the pigments are extremely hard to cut with glass-knives, good thin sections have been obtained and they seem to confirm the hypothesis of Frangois and co-workers that no internal struc- ture (2, 3) is present in the pigments. The parallel striations which can be seen in some of the micro- graphs are an artefact due to the hardness of the material. The sections of the tissue blocks show the arrange- Fig. 1. Pigments from the retinal epithelium of a pig-eye. Pd. -shadowed. Magnification 8000. Fig. 2. Thin section of pigments from the retinal epithelium of a pig-eye. The pigments are still attached to cellular residues. Magnification 6000. The Ultrastnictiin- of a Frog Muscle Spindle 197 merit of the pigments in relation to the rod extre- mities. The pigments are all grouped in the interior part of the pigment epithelium. The axes of the pigments remain parallel to the axes of the rods even where the angle of incidence of the latter is very small. So far it has been generally accepted (5) that the inner surfaces of the epithelial cells send out proto- plasmic processes around the rods and cones, thus providing each rod and cone with a pigment sheath. Following our observations with the electron micro- scope, it seems at first quite improbable that enough pigments would be present to form such a sheath. Not a single preparation has shown a place where the pigments were farther than the outermost part of the rods. Considering the importance of this problem, work has been started in order to elucidate this physiological aspect. RiFERENCES 1. I RAN^ois, J., Rabaey, M., and Vandermeerssche, G., Natuurw. Ti/dschr. 34, 191 (1952). 2. — Ann. d'Oc. 186, 896 (1953). 3. — Ophthalmologicii 126. 347 (1953). 4. — Modem Trends in Ophllialmology — Electron Mi- croscopy. 3rd Series. Butlerworili and Co. London, 1955. 5. Maximov, a. a. and Bloom, W., Textbook of Histology. 6th ed., p. 557. W. B. Saunders Co. 1953. 6. Newman, S. B., Borysko, E., and Swerdlow, M., /. Re- search Natl. Bur. Standards 43. 183 (1949). 7. Paiade, G. E., /. E.xpil. Med. 95, 285 (1952). 8. RusKA, H., Stuart, D. C, Jr, and Win.sser, J., Arch. ges. Virusforsch. 6, 5 (1956). 9. SjosTRAND. F. S., /. Appl. Phys. 19. 1188 (1948). 10. — J. CeUular Comp. Physiol. 33, 383 (1949). 11. — ibid. 42, 15 (1953). 12. — ihid. 42, 45 (1953). Preliminary Observations on the Ultrastructure of a Frog Muscle Spin(dle J. D. Robertson Department of Anatomy, University College, London Trog muscle spindles are sensory proprioceptive end organs consisting of bundles of small muscle fibers with afferent nerve endings encased in a distinctive connective tissue sheath. The spindle system of the m. ext. dig. long. IV of the frog (R. temporaria) has recently been the subject of a detailed histological study by Gray (2). The results of an electron microscope study of this same spindle system is reported here in preliminary form. The muscle was fixed in 1 *?o OSO4 at about 5C for about 90 minutes. The spindles were dissected out after fixation, rapidly dehydrated in alcohol, embedded in methacrylate and sectioned with a Porter Blum micro- tome. Successive thin transverse sections were examined with a Siemens electron microscope at levels varying from two to fifty // apart. 1. The spindle sheath. — The spindle sheath is divided into an outer (figs. 1-2) and an inner (fig. 3) component. The outer consists of several compact cytoplasmic layers about 1-3 // thick with scattered concentrically elongated nuclei and associated bundles of collagen fibrils. The layers are formed by elongated sheet-like cells about 0.2-0.3 /i thick bounded on their free surfaces by a double membrane complex about 300-500 A thick (fig. 2). Several of these thin cells are closely apposed to produce the compact cytoplasmic layers. The ap- posed cells are separated by simple double membranes about 250-300 A thick (fig. 2). In some regions the cells separate to form tissue spaces occupied fre- quently by collagen fibrils. It appears that the indi- vidual cells do not split into several layers to form the composite lamellated structures but rather that each cytoplasmic layer belongs to a separate cell; the cells are packed together like the leaves in the center of a head of cabbage. The thin cytoplasm of these cells contains scattered mitochondria and nu- merous round or oval vesicular-appearing structures about 200 500 A in diameter which sometimes appear to originate from the cell surface and some- times appear to form tubular channels across the cytoplasm (fig. 2). These tubular channels drop out in serial sections. A space about 5-10 /< wide lies between the outer and inner sheaths (hg. 1). This contains a few scattered collagen fibrils and occasionally a sheath cell is seen traversing the space to connect with the inner sheath. The inner sheath sometimes consists of only one cytoplasmic layer varying from 0.1 // down to 300 A in thickness (tig. 3). Sometimes two or three slightly overlapping layers are present but they arc not closely associated as in the outer sheath. Between the inner sheath investing each intrafusal fiber and the surface of the muscle fiber there is present a space about 1-3 microns thick which is occupied by a material of low density and generally amorphous or delicately fibrillar appearance in which terminal nerve fibers are situated (fig. 1). This ma- terial is sometimes partly removed by treatment with phosphotungstic acid (fig. 3) except for a thin 198 J. D. ROBERTSON The Ultrastructurc of a Frog Muscle Spindle 199 layer 150-300 A thick at the muscle and nerve sur- faces. This layer is the outer dense layer of the muscle and nerve double surface membrane complex. Scattered longitudinally disposed collagen fibrils are present in this perimuscular substance usually near the sheath cell surfaces. 2. The intrafusal nniscle fibers. — The intrafusal muscle fibers in the spindle region vary in diameter from 5-10 //. Their centrally or slightly eccentri- cally placed nuclei occupy the bulk of the sarcoplasm in the smaller fibers. The sarcoplasm of the larger fibers generally is filled by myofibrils (fig. 3). Mode- rate numbers of mitochondria are present. As the region of maximal innervation ("nuclear bag" region) is approached centrally placed closely spaced nuclei appear and mitochondria increase in relative number. Endoplasmic reticulum and vesicular-appearing bo- dies are sometimes prominent in the sarcoplasm (fig. 4). The diameter of the fiber decreases and the nucleocytoplasmic ratio increases. The number of myofibrils decreases and the scanty sarcoplasm some- times exceeds the myofibrillar area in cross section. However, no section has yet been obtained in which no myofilaments at all can be identified. The surface of the muscle fiber is thrown into numerous folds and ridges which partially envelope the terminal nerve twigs. These folds sometimes extend as finger- like processes passing entirely through the middle of the muscle fiber. These penetrating channels contain the perimuscular substance and sometimes axon or Schwann cell branches. 3. The juxtaterminal nerve fibers. — The spindle region contains numerous myelinated nerve fibers 5-10 // in diameter. These penetrate the outer and inner sheath by passing between the sheath cells. The endoneurial sheath is here replaced by the spindle sheath. Indeed, it appears that the spindle sheath may be simply a specialized endoneurial sheath. These fibers branch repeatedly as the inner sheath is penetrated and the myelin sheath is reduced in thickness to 0.1 //. The inner or outer surface- connecting membrane (SCM) is sometimes visible in these fibers (figs. 1 and 4), (ref. 1, 5). Inside the inner sheath increased numbers of unmyelinated libers arc seen with rare small mye- linated fibers. The former consists of Schwann cells containing one or several axons each with one SCM. In rare instances intermediate fibers arc seen con- sisting of a Schwann cell with one axon and one elongated SCM which partially surrounds the axon in a spiral fashion ( i ). in a few instances one axon is seen with two SCMs. To the lower right of fig. 4 may be seen an axon (Ax) which is only partly covered by its Schwann cell iSc/i). One SCM is clearly shown. The axon presumably is leaving the Schwann cell via the second SCM. 4. The nerve endings. — The axons and Schwann cells develop an unusual number of mitochondria near their termination and this appears to be more pronounced in the axons. Vesicular- or tubular- appearing bodies (fig. 3 inset) about 250-300 A in diameter are fairly prominent in Schwann cytoplasm and less so in axoplasm. Axoplasmic filaments are not prominent in the terminal axons despite their frequent prominence in the larger myelinated fibers. Endoplasmic reticulum is sometimes prominent in Schwann cytoplasm. The small terminal branches of both axons and Schwann cells appear so similar that sometimes it is difficult to identify them with certainty. Often a structure at first thought to be a terminal axon is proven to be a Schwann cell on deeper sectioning (fig. 3). Fig. 1. Transverse section of spindle showing two intrafusal muscle fibers and a part of a third. A portion of the outer spindle sheath io.s.) may be seen to the lower right and the middle right (Ns). Numerous collagen fibrils (r.) are present. Each intrafusal fiber contains a centrally placed nucleus (N,n) surrounded by a thin layer of sarcoplasm containing a few myofibrils. The surfaces of the muscle libers are irregularly convoluted and in contact with terminal nerve branches. Some of the terminal nerve structures lie in the diffusely dense perimuscular substance which is clearly shown about the central muscle fiber. The inner sheath appears as a very thin cytoplasmic layer in which nuclei (/Vs ) appear. A small myelinated nerve fiber appears inside the inner sheath to the upper right. The intrafusal fiber associated with this axon is enlarged in fig. 4. Magnification 3,100. Fig. 2. Enlargement of two layers of an outer spindle sheath. Each cytoplasmic layer c^ and c, is bordered above and below by a double surface membrane complex about 300-400 A thick. A double membrane about 250-300 A thick separates the two sheath cells. Tubular-appearing structures partially traverse each cell. These appear to be related to the surface membranes. Magnification 80,000. Fig. 3. Single intrafusal fiber stained with phosphotungstic acid. The edge of a sheath nucleus ( As ) appears to the upper left. The delicate inner sheath (/..?.) partially surrounds the muscle fiber. The perimuscular substance is practically absent. Several terminal nerve processes are seen in the perimuscular space. One Schwann cell (Sc/i) is seen in contact with the muscle and with an axon (.4.v) which also contacts the muscle. Two fibers which may be axons {? Ax) are in contact with the muscle above. A fiber containing numerous vesicular- appearing bodies is in contact with the muscle to the lower left. This was proven in serial sections to be a Schwann cell. The dotted area is enlarged in the inset to show the double membrane between Schwann cytoplasm and sarcoplasm. Note to the right the tubular extension of the nuclear double membrane into Schwann cytoplasm and its suggestive rela- tionship to the round and oval vesicular-appearing bodies in Schwann cytoplasm. Magnification 14,000. Inset, magni- fication 45,000. Fig. 4. Enlargement of the intrafusal muscle fiber to the upper right in fig. I. Myofibrils (nif) are present in the muscle fiber. Numerous Lmmvelinated nerve fibers are present in the scanty perimuscular substance. One myelinated fiber with an outer SCM appears to the right. Note the axon (Ax) which is apparently leaving its Schwann cell (Scli) to the lower right. One SCM is present in this liber. The terminal axon or Schwann cell in contact with the muscle fiber to the lower center is enlarged to the left to show the 200-300 A thick double membrane. Magnification 19,000. Inset, magnification 73,000. 200 J. D. ROBERTSON These terminal axons and Schwann cells contact the intrafusal muscle fibers in the nuclear bag region and an undiflFerentiated double membrane about 250 300 A thick is formed in each instance (figs. 3^). Motor endings have sometimes been observed in the same preparation on nearby extrafusal fibers. Here a five layered membrane complex 500-700 A thick resembling closely the type of membrane com- plex observed previously in rat (3) and lizard (4, 6) motor end plates is seen. In this kind of frog motor ending the junctional folds are largely absent. It appears that the difference between the motor and sensory membranes may be helpful in differentiating motor and sensory neuromuscular junctions if followed in serial sections. This is necessary because a sensory terminal may display the m.otor type of membrane complex as the nerve twig parts from the muscle (fig. 4). In the perimuscular substance are numerous ter- minal cytoplasmic profiles about 0.1 0.3 /i in diameter containing small mitochondria (figs. 3^). These profiles often drop out in serial sections. But occasionally they expand in size and become packed with mitochondria or rarely enlarge greatly and develop a nucleus. To date no very reliable criteria other than serial sectioning for the identification of these terminal perimuscular profiles as axons or Schwann cell processes has been applied. In summary the following kinds of axon, Schwann cell and muscle relationships have been observed inside the inner sheath: (1) Axon and Schwann cell together in perimuscular substance. (2) Axon and Schwann cell together in contact with the muscle surface. (3) Schwann cell alone in contact with muscle. (4) Schwann cell alone in perimuscular substance. (5) Numerous terminating axon or Schwann cell twigs in contact with the muscle surface and in the perimuscular substance. If criteria for positively identifying terminal axon and Schwann cell processes had been applied a coherent picture of a systematic mode of termination could probably be presented. In the absence of these no definitive statements may be made at present. But a tentative working hypothesis involving some unproven as- sumptions might be considered. It appears that the axons and Schwann cells run along the muscle sur- face for some distance with a simple double mem- brane relationship shared between all three kinds of cytoplasm. The axons and Schwann cells separate from each other while maintaining their individual synaptic contact with the muscle fibers. Then each leaves the muscle fiber separately, branches and terminates in the perimuscular substance free of any terminal contact with other cells. While none of the facts contradict this interpretation the assump- tions on which it is based require further investiga- tion. References 1. Geren, B. B., E.xptl. Cell Research 7, 558 (1954). 2. Gray, E. G., Proc. Roy. Soc. B (in press). 3. Palade, G. E., Anat. Rec. 118, 335 (1954). 4. Robertson, J. D., Anat. Rec. 118, 346 (1954). 5. — /. Biophys. Biochem. Cytol. 1, 271 (1955). 6. — ibid 2, 381 (1956). vni MUSCLE AND OTHER CONTRACTILE ELEMENTS Preliminary Observations on the Structure of Insect Flight Muscle H. E. Huxley and Jean Hanson Medical Research Council. Department of Biophysics, University College, London, and Medical Research Council, Biophysics Research Unit, Wheatstone Laboratory, King's College, London The structure of striated muscle from vertebrates has been analysed in considerable detail in recent years, particularly by the techniques of x-ray diffrac- tion, electron microscopy, and phase-contrast and interference light microscopy. A theory has been developed which accounts for all the observed fea- tures of the structure, including the pattern of cross- striations and the changes in that pattern during contraction, in terms of a series of overlapping, interdigitating arrays of longitudinal filaments which slide into each other when the muscle shortens (2, 3, 5, 6). Two types of filament appear to be present; they differ in their location, diameter, and protein composition. The thicker filaments extend from end to end of the A-bands and account for the high density and birefringence of that band; they consist largely of myosin. The thinner filaments extend from the Z-line, through the I-band, into the A-band, up to the edge of the H-zone; these filaments contain actin. When the muscle shortens, these "secondary"' filaments are apparently drawn further into the array of "primary" filaments which form the A-band. In the electron microscope, cross-sections of muscle show a double hexagonal array of filaments in the A-bands where the two arrays of filaments overlap (Huxley, 1953). Each secondary filament is located symmetrically between three primary fila- ments, so that each primary filament has six sec- ondary filaments around it which it shares with its six neighbouring primary filaments. Insect night muscle exhibits a pattern of cross- striations which is very similar in many respects to that of vertebrate striated muscle, but differs from it in that the I-bands are usually very short or entirely absent. However, such muscles undergo only very small changes in length (of the order of a few per cent) during activity, and it could be argued that if they contract by the same sort of process as has been suggested for vertebrate striated muscle, they do not need to have very long I-bands. That the process of contraction should be similar in all types of striated muscle, would seem likely, but remains to be proven. Recently, studies of insect flight muscle have been made by Hodge (1955) and by Hanson (I ). Hodge's observations on the effect of ATP on glycerinated flight muscle seemed to show a migration of A-substance to the Z-lines. Hanson, on the other hand, observed ATP-induced contrac- tions of insect flight muscle which seemed very analogous to those given by vertebrate striated muscle, and she has been unable to repeat Hodge's observations. She has also made many other obser- vations on insect flight muscle which show a very high degree of similarity to corresponding observa- tions on vertebrate striated muscle. On the basis of his electron-microscope and light- microscope observations, Hodge came to the con- clusion that a double array of filaments is not present in insect flight muscle, but that, instead, a system of cross-bridges exists between the filaments. He has demonstrated the existence of these bridges by ex- tremely elegant electron micrographs of cross-sections of flight muscle. The bridges appeared rather less well-organised in longitudinal sections, but no con- tinuous axial structure was visible in his sections in between the main axial filaments. We have now carried out the first stages of an electron microscope investigation of insect flight muscle. On the basic issue of whether or not two sets of filaments are present, the results are very clear- cut, and are in conflict with Hodge's conclusions; they appear to show that insect flight muscle has a structure which is highly analogous to that of verte- brate striated muscle, and it is these results which we shall now describe. Materials and methods. — Insect flight muscle from Cal- liphora was fixed in osmic acid, dehydrated in alcohol, embedded in methacrylate, and sectioned on a modified version of the Hodge-Huxley-Spiro (4) microtome. Sections were mounted on carbon films and examined in the Siemens Elmiskop I. Additional staining was usu- ally provided by 1 % phosphotungstic acid. Results. — Longitudinal sections of flight muscle fixed fresh usually show fibrils with virtually no I-bands. Z-lines are visible, and between them stretch longitudinal filaments (fig. I ). in cross-sections, these may be seen to form a very regular hexagonal array. Longitudinal sections often display a pattern of pseudo-striations when the plane of sectioning is not quite parallel to the layers of the lattice, and successive layers of filaments are seen. Examination of very thin longitudinal sections at high magnification and high resolution shows that mid-way between each two primary filaments, a secondary filament is present (fig. 2). The secondary filaments are connected to the primary filaments on either side by cross-bridges which occur at mode- rately regular intervals of about 100 A. This structure is observed with great consistency. The cross-bridges often appear to occur alternately on either side of the secondary filaments. In cross-sections of these fibrils, the same structure is very clearly visible (fig. 3). The predominating feature of the secondary material in such cross-sec- tions is the set of six secondary filaments around The Structure of Insect Flii;ht Muscle 203 Fig. 1. Longitudinal section of insect flight muscle, showing array of logitudinal lilaments. No I-bands are visible. Magnification 22,000. Fig. 2. Longitudinal section of same muscle at higher magnification, showing secondary filaments between the primary filaments; primary and secondary filaments appear to be connected together by a system of bridges. Magnification < 135,000. Fig. 3. Cross-section of insect flight muscle showing double array of filaments: six secondary filaments may be seen around each primary filament, and each secondary filament is located midway between two primary filaments. Magnifica- tion X 135,000. 204 F. S. SJOSTRAND AND EBBA ANDERSSON each of the primary filaments. The arrangement is, however, not completely identical with that in verte- brate muscle; in insect muscle, the secondary fila- ments are located midway between two primary filaments, and not symmetrically between three pri- mary filaments as in vertebrate striated muscle. The cross-sections also show in many places the bridges between the primary and secondary filaments. It is apparent from the cross-sectional views that longitudinal sections which are very thin and which pass successively through different layers of lattice should in some areas contain only secondary fila- ments. This is in fact often observed. Detailed exa- mination of such sections also confirms that the secondary filaments are not merely part of the pri- mary filaments in adjacent layers. Examination of the details of the structure in the neighbourhood of the Z-line is now proceeding. Discussion. — The appearance of the double array of filaments in insect flight muscle confirms that its structure bears many resemblances to that of verte- brate striated muscle, and suggests that the contrac- tion mechanism may be similar to the one suggested for the latter type of muscle. Hodge interpreted his results rather diff"erently. However, reference to his published electron micrographs shows that the cross- sections could well be interpreted in many areas as showing the secondary filaments heavily cross- bridged to the primary filaments, and located midway between each pair of them. Whether the "bridge" or the "filament" appearance of the secondary material predominates will depend very critically on the degree of staining, the section thickness, the orientation, and the exact plane of sectioning. The micrographs of longitudinal sections published by Hodge show the intact bridges much less clearly than do the cross- sections and the general contrast and orderliness is considerably less than on the cross-sections; the non-appearance of secondary filaments in these sections is not, we feel, completely conclusive. The system of cross-bridges clearly provides a means by which the secondary filaments could be pulled along relative to the primary ones, as indeed has already been suggested for vertebrate striated muscle. We have observed the filaments and the bridges together with such consistency that we feel that the secondary filaments are indeed a completely genuine structure. References 1. Hanson, J., /. Biophys. Biochem. Cytol. (1957, in press). 2. Hanson, J. and Huxley, H. E., Nature 111, 530 (1953). 3. — Symposia Soc. Exptl. Biol. 9, 228 (1954). 4. Hodge, A. J., Huxley, H. E., and Spiro, D., /. Exptl. Med. 99, 201 (1954). 5. Huxley, A. F. and Niedergerke, R., Nature 173, 971 (1954). 6. Huxley, H. E. and Hanson, J., Nature 173, 973 (1954). The Ultrastructure of Skeletal Muscle Myofilaments F. S. Sjostrand and Ebba Andersson The Laboratory for Biological Ultrastructure Research of the Department of Anatomy, Karolinska Institutet, Stockholm This study aims at a detailed analysis of the struc- tural organization of the individual myofilaments of skeletal muscle at various states of shortening of the sarcomere. Leg muscles and abdominal muscles from frog, intercostal muscles and abdominal muscles from mouse, were fixed in situ in the living anesthetized animal or rapidly after decapitation of the ex- perimental animal. The frog muscle tissue was cooled by keeping the living frogs in a cold room at ~3 to 5 C before dissecting free the muscle tis- sue. In most cases, a minute bundle of muscle fibers were dissected free in situ by means of fine glass needles and the rest of the muscle tissue removed. The excitability of the muscle tissue to direct electric stimulus was checked before the fixation. At the lowest temperatures, the excitability was greatly reduced. Contracted muscle tissue was obtained by continuous electric stimulation to tetanic contrac- tion during the first 15 minutes of fixation. A series of experiments were performed, in which the muscle tissue in situ was subjected to extraction with a 1 : 1 glycerin-water solution and the standard solutions for myosin and myosin-actin extraction. 1 % isotonic solutions of osmium tetroxide buf- fered to various pH were used as well as buflfered formalin solutions (10 "o). The l^ hours' fixation was performed at 1 to -2'C. Dehydration with ethyl or iso-propyl alcohol and embedding in methacrylate. For high resolution pictures, the ultrathin sections were mounted on metallized formvar nets and analyzed in an RCA EMU 2c electron microscope with double objective without objective aperture. The electron optical magnifica- tions were 40,000-60,000 times. In the extensive material of skeletal muscle tissue that has been analyzed, only one type of myofila- ments has been observed (8). The myofilaments run through the whole length of the sarcmoere without interruption and continue through the Z- line (fig. 1). The diameter of the myofilament is diff'erent in the A- and the I-bands, which partly V .+■. Skclctctl Muscle Myofilaments $^^ ■^ ^flX- -• ■ i* fe .^ 'f-; i ^ .^ ■WkA4' O.i/j. , S Fig. 1. Survey picture of longitudinal section through shortened sarcomeres of frog leg musc\Q [in. biceps Jemoiis). Magnification 92,000. Fig. 2. Higher magnification of longitudinal section through shortened sarcomere of /;;. biceps fenwiis of frog. Notice the thick A-band part of the myofilaments with irregular outlines, interconnecting cross-bridges, and the transversal orientation of more opaque regions within the myofilaments. Magnification < 160,000. Fig. 3. Longitudinal section through stretched muscle fiber with elongated sarcomeres. Notice the difference in diameter of the myofilaments as compared to fig. 2. Magnitlcation 160,000. 206 F. S. SJOSTRAND AND EBBA ANDERSSON •M'. T^ '^^ "^■^ t^ "IW 6 1 li f 'ji.; ■^ 'i')- ** .♦-• ,# % .^" ,:^ # ^ 4 O.Iju ^ f 5 ,?'■ *^ IT ^ -^''^;^ ' ^-«. ■t 3 # ^*# Skeletal Muscle Myofilaments 207 accounts for the difference in opacity existing be- tween these two bands. The diameter is about 40 A in the 1-band. In the A-bands, the diameter of the myofilaments is a function of the degree of shorten- ing, and varies between about 60 to 140 A. The contours of the myofilaments are smooth within the l-bands but are irregularly zig-zag-formed in the A-band. Between the thick parts of adjacent myofilaments, interconnecting bridges arc seen (fig. 2). The thickness of these cross connections may be as minute as 20-30 A. The distance between the cross bridges extending between two adjacent myofilaments varies with the degree of shortening from 90 A in sarcomeres which are shortened (to about 50 "„ of their equilibrium length) to 250 A in extended sarcomeres. Judging from the variations in the opacity of the individual myofilaments, these seem to represent rather complicated structural units. In the shortened sarcomeres, there are observed transversally oriented regions of high opacity alternating with less opaque sections along the myofilaments. The structural pattern observed in the myofilaments is rather ir- regular, but shows that on shortening there is a definite tendency of transversal orientation of material within the myofilaments. This fact in ad- dition to the tightly arranged cross-bridges gives the impression of a marked transversal orientation of the structural components in the shortened sarco- mere (fig. 2). The modulation of the opacity within the myofila- ments may be interpreted as due to their consisting of smaller structural subunits. Direct indications of such subunits may be observed, but the electron staining of the myofilaments is rather diffuse, which is obvious when comparing the resolution of the myofilament structure with that of more dis- crete structures in the pictures. In fig. 4, it is pos- sible to observe in some places that the myofila- ments are subdivided into smaller filamentous units with an estimated diameter of 10-20 A. This picture is from an extended sarcomere in a muscle that was fixed in a stretched state. The thickness of the A-band part of the myofilaments is only about 60 A as compared with 140 A in a shortened sarco- mere. It seems justifiable to correlate the longi- tudinal orientation of the subunits to this stretched state. When examining cross sections (figs. 5, 6), the hexagonal arrangement of the myofilaments within the myofibrils is revealed (fig. 5). Cross bridges connecting adjacent myofilaments are also ob- served. They do not pass straight over from myo- filament to myofilament, but show a more irregular course with nicks. This may be explained by the shrinkage perpendicularly to the length of the myo- lihiments that takes place during fixation and cm- bedding. At the nicks, the cross bridges may appear especially distinctly because they frequently are thicker at these sites. It might be these nicks in combination with superposition efTects in too thick sections of crossing bridges that have been inter- preted as cross sections of a second type of thinner myofilaments interposed between the thick myo- filaments within the A-band region (4). In high resolution pictures, it is striking to ob- serve the angular shape of the myofilaments in cross sections and the rather high frequency of triangular cross sections. The interconnecting bridges between adjacent myofilaments, in most cases, extend between the corners of the triangular cross sections. In addition to the triangular form of the cross sections, there are other less well defined forms. The triangular form seems to correspond to the thickest regions of the myofilaments (fig. 6). Measurements of the length of A- and I-bands as well as that of the sarcomeres at various degrees of shortening (I) have revealed that both I- and A- bands shorten with decreasing sarcomere length. The change is, however, more pronounced in the I- bands. As no definite new structural pattern than those characteristic for the I- and A-bands appear in connection with the shortening it seems justifiable to conclude that the I-band parts of the myofilaments structurally have changed to the organization characteristic for the A-bands. The variations in the diameter of the A-band part of the myofilaments with varying degree of shorten- ing, the variation in distance between the intercon- necting bridges, and the transformation of the I- band part of the myofilament into an A-band type of structure are interpreted as pointing to a change in the organization of the individual myofilaments as responsible for the shortening of the sarcomere. The ultrastructure of the myofilaments indicate that these filaments consist of smaller subunits with a diameter of 10-20 A. These units might represent thin ropes of Corey-Pauling's a-helices. The shorten- ing would then be due to a folding of the compound a-helices or to a change in the spiralization with a shortening of the pitch of the secondary helices. Assuming that there are three main subunits in each myofilament would give a simple explanation to their triangular cross sections (9). Fig. 4. Longitudinaf section through stretched muscle fiber. Indications of subunits measuring 10-20 A in some myofila- ments. Notice the small diameter of myofilaments as compared to fig. 2. Magnification : 360,000. Fig. 5. Cross section through //;. hkeiys fcmoris of the frog. Most of the section through the A-band region. In lower left corner the section cuts through the I-band region. Magnification 200,000. Fig. 6. Higher magnification of cross section through m. biceps femoris of the frog. Notice the considerable number of triangular cross sections (indicated with arrows). Magnification 300,000. 208 EBBA ANDERSSON In this investigation, it has not been possible to observe the two sets of myofilaments described by Huxley (4). An explanation for this mysterious discrepancy might be that we have studied muscle tissue that has been fixed in the fresh state, and have reduced the various stages in the preparatory tech- nique to a minimum. Huxley investigated glycerin- ated muscle tissue. In such material, we have observed a splitting of the A-band part of the myofi- lament into thinner branches, each of which shows a folded or spiralized structure. The interconnecting bridges are similar to those described by Hodge (3) in dipteran flight muscle but are thinner and more numerous in vertebrate muscle. With such contradictory results as those presented here and the observations and interpretations of Huxley and Hanson (2, 4, 5), it seems justifiable not to consider the morphologic background for the contractile process as definitely unveiled. References 1. Andersson, Ebba, Intenmt. J. Ultrastructiire Research, under preparation (1957;. 2. Hanson, J. and Huxley, H. E., Symposia Soc. Exptl. Biol. 9, 228 (1955). 3. Hodge, A. J., /. Biophys. Biocliem. Cytol. 1, 361 (1955). 4. Huxley, H. E., Biochim. Biophys. Acta 12, 387 (1953). 5. Huxley, H. E. and Hanson, J., Nature 173, 973 (1954). 6. Sjostrand, F. S., Science Tools 2, 25 (1955). 7. — Exptl. Cell Research 10, 657 (1956). 8. Sjostrand, F. S. and Andersson, Ebba, Exptl. Cell Research 11, 493 (1956). 9. — , Internat. J. (JItrastnutiire Research, under prepara- tion (1957). The Tubular System in the Striated Muscle Cell Ebba Andersson The Laboratory for Biological Ultrastructiire Research of the Department of Anatomy, Karoliiiska Institutet, Stockholm Already at the end of the 19th century Retzius (2) found in his material of gold-impregnated muscle tissue a network in the sarcoplasm between the myofibrils. Until recently rather little attention has been paid to the structure and function of this sarcoplasmic detail in the muscle cell, but with the development of the cell studies by means of electron microscopy this network has gained new interest. Bennett and Porter (I) have among others included this component in the rather all-round endoplasmic reticulum. The muscle material investigated is mainly taken from skeletal muscle from frog and mouse. The material is fixed /// situ in buttered isotonic osmium tetroxide solu- tions. Some material was stained with phosphotungstic acid dissolved in 70 % ethanol after the fixation. The preparations were embedded in a //-butyl-methyl metha- crylate mixture and sectioned with Sjostrand ultra- microtomes. The micrographs were taken with a RCA EMU 2c electron microscope. In a longitudinal section (fig. 1) the sarcoplasmic component is recognized as tubes passing in between the myofibrils. In the A-band region the tubes are oriented parallel to the axis of the myofibrils. In the 1-band region they run in various directions. partly perpendicular to the long axis of the myofi- brils. In most of the preparations the tubes are dilated in the I-band region. In a cross section (fig. 2) the tubes are cross- sectioned in the A-band area and more or less obliquely or longitudinally sectioned in the I-band area. The diameter of the undilated tubes is 250-300 A. The total thickness of the bounding membrane is about 60 A and is triple layered with the layers measuring 20 A in thickness. This is observed most easily in material stained with phosphotungstic acid. The tubes have a very intimate topographic relation to the myofilaments. In phosphotungstic acid stained material the bounding membrane of the tubes seems to be in direct contact with the myofila- ments. The relationship of the tubes to the mitochondria is very intimate. Any definite direct continuity be- tween tubes and mitochondria has, however, never been observed. For the analysis of this relationship the mouse muscle has been used with its large number of mitochondria. Fig. 1. Longitudinal section from frog skeletal muscle, fixed in osmic acid. The lubes are longitudinally sectioned in the darker A-band region and more or less cross-sectioned in the lighter I-band region. In the lower right part a longitudi- nally sectioned mitochondrion with its triple-layered outer and inner membranes. Magnification 32,000. Fig. 2. Cross section through skeletal muscle from frog. Osmium fixation. The section has passed through both A and I-band regions. In the I-band region the tubes are somewhat dilated. In the A-band region to the right a cross- sectioned mitochondrion. Down in the lower part is the sarcolemma. Magnification 36,000. The Tubular Svstem in the Striated Muscle Cell 209 ^. X b. \ ■ >».4p-- ■^ ■> N ^^ ST. ■^. ^ ^ >'. 1 > _ ■'>>t "> ' 'V*. ?^ ^ M I i- "sit. f/ '^. S'spw^*-, .^ 2 / ''C f- 3 0 %■-%' .'5 fcl V 4*^P .:»'■ 'S^ f iL /^ 14 — 568204 Electron Microscope 210 HEDI GANSLER The thickness of the tube wall and the cell surface membrane is of about the same dimensions. In some preparations where the tubes are dilated in the I-band region it seemed possible that the tubes were invagi- nations of the plasma membrane. This supposition has not as yet been confirmed, although an abundant material has been analyzed where special attention has been paid to those areas where the plasma membrane extends over the I-band regions. No defi- nite indications of such a relationship could be found in serial sections. As to the relation of the tubes to the Golgi appa- ratus it may be pointed out that the dimensions of the Golgi membranes and those of the tube walls are similar. A direct continuity has not been estab- lished between these two components. As to the function of these tubes there are two possibilities which are obvious. They may represent a structure for conduction of the excitation through the muscle fiber or they may represent a structure involved in the sarcoplasmic extra-mitochondrial metabolism. References 1. Bennett, H. and Porjer,K.R., Am. J. Anal. 93, 6\ (1953). 2. Retzius, G., /// Biologische Untersuchungen 1 (1881). Elektronenmikroskopische Untersuchungen am Uterusmuskel der Ratte Hedi Gansler Rheinisch-Westfalisches Iiistitnt fiir Vbermikroskopie, Diisseldorf WiR untersuchten die innere Ringmuskulatur von Ratten-Uteri, die unter den verschiedensten Ovarial- hormonbedingungen standen: Uteri von infantilen Ratten, von normalen Ratten in verschiedenen Zyklusphasen, von kastrierten und von Daueroestrus- Ratten (Kastration mit anschlieBender FH-Applika- tion). Nach Decapitieren entnahmen wir aus dem mittleren Teil eines Uterushorns eine kleine Scheibe, die in kleine Stiickchen zerschnitten wurde. An dicken Schnitten orientierten wir uns im Phasen- kontrast-Mikroskop iiber die Topographic des einge- betteten Materials und nahmen zur elektronenmikro- skopischen Untersuchung nur die Blockchen, bei denen das Endometrium mit langsgeschnittener in- nerer Ringmuskulatur zu sehen war. Das Material wurde in T'oiger Osmiumsaure nach Palade I St. bei Zimmertemperatur fixiert und in der iiblichen Weise eingebettet. Die Einteilung der glatten Muskelzelle in Endo-, Meso- und Exoplasma von Haggqvist ( I ) erwies sich als Schema auch fiir die Ultrastruktur der glatten Muskelzelle als sehr geeignet. Das Endo- plasma umfaBt Kern, Mitochondrien und endoplas- matisches Retikulum. Das Mesoplasma wird von den Myofibrillen gebildet, das Exoplasma besteht aus Zellmembran und angelagerten Kollagen- bzw. ela- stischen Fasern. Im infantilen Uterus beobachteten wir folgende Strukturen: Endoplasma: Kern lang gestreckt, parallel ver- laufende Doppelmembran, ohne Fiiltelungen und Einbuchtungen, Karyoplasma homogen. Wenig kleine Mitochondrien mit normaler Innenstruktur. Mesoplasma: Myofibrillen von etwa 50 A D und unbestimmter Lange, die eine bevorzugte Langs- orientierung erkennen lassen. Selten sieht man strich- formige Verdichtungen, die dadurch zustande kom- men, daB um die Myofibrillen eine elektronenoptisch dichtere Substanz angelagert ist. Die Myofibrillen in diesen Bereichen zeigen dieselbe Dimension und Orientierung. Exoplasma: Die einfachen Zellmembranen ver- laufen leicht gewellt, annahernd parallel; mitunter sieht man Membranliicken von etwa 1000-2000 A. An diesen Stellen besteht also ein direkter Kontakt zwischen dem Cytoplasma zweier benachbarter Zel- len. Myofibrillen konnten wir in solchen Kontaktstel- len nicht beobachten. Zwischen denZellen sind mehr Oder weniger zahlreich Kollagenfasern eingelagert. Sie scheinen an Membraneinbuchtungen biischel- weise aus der Zellmembran herauszuwachsen, also ein Produkt der Muskelzellen selbst zu sein. Binde- gewebszellen sind nur im Stroma des Endometrium bzw. der GefiiBschicht zwischen Ring- und Langs- muskulatur zu beobachten. Normaler Uterus: Die Auswertung der Vaginalab- striche erfolgte nach dem Schema von Long und Evans (2). Um die Veranderungen am Myometrium zu beschreiben, unterscheiden wir eine Pra- und eine Postostrusphase. Ein signifikanter Unterschied be- steht insofern, als im Post-Ostrus der gesamte Muskelzellverband aufgelockert ist, dieZellen heller und vakuolisiert sind. Prd-Ostrus: Endoplasma: Kerne meist mit zahl- reichen tiefen Einbuchtungen, die auf Grund der Befunde an dicken Schnitten eine spiralige Verdre- Der Utcrusmuskcl clcr Ratte 211 %. A ir>-- Abb. 1. Muskelzelle aus einem schwangeren Uterus: Kern- und Zellmembran langgestreckt. Das Mesoplasma besteht aus dicht gepackten, langs orientierten Myofibrillen ohne signifikanten Verdichtungen. Vergrosserung: 24600 x. Abb. 2. Muskelzelleauseinem schwangeren Uterus:Zellmem- bran stark gefiiltelt, die Myofibrillenzone zeigt zahlreiche Verdichtungen. Vergrosserung: 22000 hung der Kerne wahrscheinlich machen. Das Karyo- plasma ist inhomogen, die meist doppelte Kern- membran zeigt einen parallelen Verlauf. Kleine nor- niale Mitochondrien, Golgiapparat und endoplas- matisches Retikulum nicht auffallig. Mesoplasma: Myofibrillen wie zuvor beschrieben, aber zahlreiche spindelformige Verdichtungen von 150-300 A Breite. Zwischen den Myofibrillen verein- zelt kleine kreisformige Membranen, die u. E. von Einstulpungen der Zellmembran herriihren. Exoplasma: Membran leicht gefiiltelt, einfach kon- turiert, meist mit der Membran der Nachbarzelle parallel verlaufend, auch hier Membranliicken. Zwischen den Membranen entweder eine homogene Grundsubstanz oder Kollagcnfibrillcn. Die Mem- bran zeigt zahlreiche rohrchenformige intracytoplas- matische Einstiilpungen. Post-Ostrus: Kerne homogen, langgestreckt, Mito- chondrien hell geschwollen mit ungcordneten innen- membranen. Endoplasmat. Retikulum vakuolig er- weitert. Um die Membranen und frci im Cytoplasma viel Paladsche Kornchen. Mesoplasma: Myofibrillen reduziert, vorwiegend im peripheren Teil derZelle, Langsordnung nicht so eindeutig wie im Prii-Ostrus. Verdichtungen sind nur selten zu sehen. Exoplasma: Membranen gerade verlaufend, Ab- stande zwischen den einzelnen Zellen groBer, dazwi- schen zahlreiche Kollagenfibrillen. Schwangerc Ratten: 15. Tag (Abb. 1-2). Ab- weichend von der eingangs beschriebenen Tech- nik wurden die schwangeren Tiere narkotisiert und die Uteri /// .v/7// fixiert, um eine Kontraktion beim Aufschnciden der Uteri zu vermciden. In vcr- schiedenen BIdckchen fanden wir zwci verschiedene Zelltypen: Typ /: Endoplasma: Kerne langgestreckt, homo- gen, doppelte Kernmembran gerade verlaufend. Die AuBenmcmbran ist deutlich von der Inncn- mcmbran abgchobcn, milunter zeigt sic kleine Aus- buchtungcn in das Cytoplasma. Die Mitochondrien sind hell geschwollen, auBerdem sieht man zahlreiche Vakuolen auch zwischen den Myofibrillen und eine stark entwickelte Golgi-Zone. Mesoplasma: Myofibrillen wie zuvor beschrieben, nur fiillen sie einen groBeren Teil derZelle aus. Nur wenig schmalc und kontrastarme Verdichtungen. £".vo/?/a.s7;;a.Zellmcmbranen gcstreckt. parallel vcr- 212 F. GUBA UND G. HAJOSSI-KEREK laufend, wie zuvor beschrieben, mit zahlreichen rohrchenartigen Einstulpungen derZellmembran. In- terzelluliir ist eine homogene Grundsubstanz und Kollagen, das oft verwaschen erscheinf. dazwischen eingelagert sog. elastische Palcks. Diese Befunde sprechen u. E. im Sinne Wassermanns (3), der an einem prinzipiellen Unterschied von amorpher und strukturierter Interzellularsubstanz nicht langer fest- halten will. Typ II: Endoplasma: Kerne gefaltelt und einge- buchtet, Karyoplasma inhomogen. doppelte Kern- membran parallel verlaufend. Mitochondrien normal, Golgizone und endoplasmatisches Retikulum deut- lich vermehrt. Mesoplasma: Die Myofibrillen zeigen im Unter- schied zu Typ 1 zahlreiche breite, stark kontrastierte Verdichtungen, in denen man teils langs verlaufende, teils ungeordnete und auch quer verlaufende Fibril- len erkennen kann. Exopkisiua: Zellmembranen gewunden verlaufend mit zahlreichen Einstulpungen, sonst wie zuvor beschrieben. Kastrierte Rotten: Im Prinzip der gleiche Befund wie bei den infantilen Tieren. Dauer-Ostrus-Ratten: Endoplasma: Kerne homo- gen, langgestreckt, ohne Membranfalten und Ein- buchtungen. Karyoplasma hell, homogen. Mito- chondrien spiirlich tiber die ganze Zelle verteilt, endoplasmatisches Retikulum sparlich. Mesoplasma: Myofibrillen stark reduziert, oft noch in schmalen Biindeln im peripherenZellbereich zu sehen; auBerdem vereinzelt und ungeordnet lie- gende Fibrillen in einem hellen strukturarmen Cyto- plasma. Verdichtungen fast ganz verschwunden. Exoplasma: Zellmembranen langgestreckt, ohne intraplasmatische Einstulpungen. Zwischen den Zel- len reichlich koUagene Fasern. LiTERATUR 1. Haggqvist, Gosta, Hdb. d. mikrosk. Anat. d. Menschen, II, 4. Springer-Verlag, 1956. 2. Long, J. A. und Evans, H. M., The Oestrus-Cycle in the Albino-Rat. Memories of the Univ. of California, 1922. 3. Wassermann, F., Erg. Anat. ii. Entwickhingsgesch. 35, (1956). Elektronenmikroskopische Untersuchungen i'lber das Gewebe des glatten Muskels F. GuBA und G. Hajossi-Kerek Eleknonciiniikroskopische Abteiliing des Jnstitiits fiir Mefitechiiik und Instnimenteiiwesen der Ungarischeu Akademie der Wissenschaften, Budapest L)iE Zellen des glatten Muskels wurden mit einem Mikromanipulator isoliert, der sich in einer feuchten Kammer befand. EIn Plasmateilchen einer so isolier- ten Zelle ist in Abb. 1 ersichtlich. Das GerCist des Zytoplasmas besteht aus longitudinal verlaufenden Subtibrillcn, die einen Durchmesser von 700 A und einen periodischen inneren Aufbau haben. Die Sub- fibrillen anstomisieren im distalen Zellenteil. Auf Abb. I. Plasmateilchen einer mit Mikromanipulator isoiicr- ten Zelle. Abb. 2. Muskel-Subfibrillen der Schneckc {Helix ponuitia). Abb. 3. Muskelsubfibrillen des menschlichcn Uterus. Behand- lung nach Weber-Edsali. Abb. 4. Wie Abb. 3. Behandlung mit KSCN. Grund der Erfahrungen mit den isolierten glatten Muskelzellen wurdc ihre Identifizierung moglich, so daB die mit dem Waring-Blendor hergestellte Mus- kelsuspension, die in groBerer Menge fiir die che- mische Untersuchung bereitet wurde, mit Sicherheit benutzt werden konnte. In den so hergestellten Prii- paraten konnte festgestellt werden, daB die Gewebe der glatten Muskeln von Siiugern, Vogeln und Mol- lusken in gleicher Weise aus elementaren Subfibril- len bestehen und daB diese Bundel bilden. Es ist be- merkenswert, daB der Durchmesser der Subhbrillen bei den MoUusken urn 30 °o groBer ist als bei den Saugern und Vogeln. Die Subfibrillen sind in einer strukturlosen, homogenen Substanz eingebettet, die am deutlichsten im Magen der Vogel erkennbar ist. In unseren Priiparaten waren auBerdem auch stets Kollagenfasern zu finden. Die Fibrillensuspension wurde sowohl chemisch untersucht als auch auf die Objektfolie des Elek- tronenmikroskopes aufpriirariert, wobei sie der gleichen Salzbehandlung wie bei der chemischen Untersuchung ausgesetzt wurde. In Abb. 2 treten die Subfibrillen infolge der Wirkung der Hasselbach- schen Auslosung (2) deutlich hervor, da sich die Grundsubstanz ausgelost hatte. Auf die Auslosung Mechanism of Pignwut Migraiion within Teleost Melanophorcs Tabelle 1. Tabelli. 2. 213 Rind-Uterus (Trockensubstanz 180 — 15 mg/g 'Gesamtstickstoff 25 7 mg/g (miometnum) | ATP-Abspaltung 0,03//Af ATP/mgN/min (Trockensubstanz 160 10 mg/g //(V/.v /v)/;/^///V/ Ciesamtstickstofl' 20 3 mg/g ATP-Abspaltung 0,75//A/ ATP/mgN/min nach Edsall (1) und Weber (3) (Abb. 3) folgt dcr Abbau der subtibrillaren Struktur. Der in der Abbil- dung sichtbare intakte Faden vveist eine fur Kolla- genfasern charakteristische Struktur auf. Abb. 4 zeigt den Zustand nach Behandlung mit Kaliumsul- fozyanid. Die Behandlung hatte eine vollige Destruk- tion der Museklstruktur zur Folge. In der Tabelle 1 sind die Angaben fur die unbe- handelten Muskelsuspensionen zusammengefaBt. AufTallend ist die hohe ATP-Abspaltungsfiihigkeit der Schneckenmuskeln, was wahrscheinlich mit dem geringeren Bindegewebsgehalt des Schneckenmuskcls zusammenhiingt. Die Angaben der Tabelle 2, die den Zustand nach den in dieser Tabelle angefiihrten Behandlungen widerspiegeln, sind in Prozenten der entsprechenden Angaben der Tabelle 1 ausgedriickt. Aus der Tabelle 2 geht hervor, daB die fur Myosin charakteristische Komponente der Struktur nach der Edsall-Weberschen Auslosung eine tiefgreifende Veranderung erieidet. Es wurde auch die Struktur der ausgelosten Proteine mit dem Elektronenmikro- skop untersucht. Unter den ausgelosten Proteinen zeigt das nach Edsall-Weber ausgeloste einen fibril- laren Charakter. Dagegen ist in dem nach Hassel- bach gewonnenen Extrakt ein globulares Protein zu sehen. Vergleicht man die Angaben der chemischen Un- tersuchung mit den elektronenmikroskopischen Auf- Ausge- ATP Emp- ATP Ab- spal- tung Behandlung Material loster N find- lich- keit Uterus Wasserige miiimetrium 25 0 20 Extrakliim Helix ponicirici 40 0 65 Auslosung Uterus miomelrium 10 0 5 nach Hasselbach Helix poiiiatia 20 0 10 Auslosung Uterus miometrium 45 30 60 nach Edsall Helix poiuiiiia 40 14 30 Die Wertc sind in Pro/cnlcn des nativcn Miiskels angegeben. nahmen, so gelangl man zur Schlulifolgerung. dal3 sich in den glatten Muskel/.ellen der verschicdcnen Lebenwesen gleicherweisc die subfibrillare struktur findet. deren Baustoff wahrscheinlich Aktomyosin ist. Die Loslichkeit der das librillare System umgc- benden Grundsubstanz ist bei den verschiedenen Organismen unterschiedlich. Was ihre chcmische Natur anbelangt, so ist sie auf Grund ihrer Loslich- keit und ihrer Enzymaktivitiit myosin- b/w. myo- genartig. LlTERATUR 1. Edsall, J. T., /. Biol. Cheni. 89, 289 (1930). 2. Hasselbach, W., Z. Natiirforsch. 86, 449 (1953). 3. Weber, H. H., Biochem. Z. 158, 433 (1925). Mechanism of Pigment Migration within Teleost Mehmophores S. Falk and J. Rhodin The Lahoratory for Biological Ultrastnictitre Research of the Department of Anatomy. Karolinska Instiiiilet. Stockholm, and the Zoophysiologieal Insliliilion. University of Liiiul The colour changes of animals are brought about by the aid of special cells in the dermis, so-called chromatophores. These many-branched cells ccMitain the pigment in the form of small granules, which can perform distal and proximal migrations within the cell. Thus the pigment granules can be more or less dispersed or concentrated, in the former case the chromatophore is said to be dispersed and in the latter concentrated. The time required for these changes from fully dispersed pigment to fully con- centrated or the reverse varies from some minutes to a few hours depending on the dilTcrent animal spe- cies. According to the type of pigment the chromato- phores are called melanophores. lipophores, allo- phores and guanophores. The melanophorcs contain a dark brown or black pigment, melanin. Chromatophores are mainly to be found in Cepha- lopoda, Crustacea, C\clostomata, Eiasmobranchii, Teleostei, Amphibia, and Reptilia. The vertebrate chromatophores are stimulated directly by heat and light and indirectly by hormones and nerve action. 214 S. FALK AND J. RHODIN \tA ^IfiW^ ^ « »*?■ Fig. 1. A concentrated melanophore with nucleus (N) in the center. Part of the cell membrane at arrow. The inner cytoplasmic membrane (I) surrounds an inner sack. Magni- fication 3800. Fig. 2. Part of a concentrated melanophore. Nucleus (N), mitochondria (M), pigment granules (P), inner cytoplasmic membrane (I), fibrillar zone (F). Magnification 29,000. They can be simple or double innervated. Tn the former case they only possess concentrating nerve fibres. In the latter case both concentrating and dispersing nerve fibres are present. The principal hormones acting upon the colour changes are inter- medin, acetylcholine and nor-adrenaline, the former two generally dispersing the pigment granules, the latter concentrating them. Thus the chromatophores are a type of effectors. In this investigation 2-3 cm long females of the com- mon aquarium fish Lebistes reticulatus were used. By placing the fishes on black or white backgrounds during 15 minutes the melanophores in the dermis on the scales were fully dispersed respectively concentrated. Scales from the dorsal and dorso-lateral sides were then quickly removed and immediately fixed at < 4' C during 5 minutes in a solution consisting of 0.289 g veronal-sodium, 0.190 g sodium acetate, 0.315 g sodium chloride, 12 ml HCl 0.1 A', I g OsO, and distilled water to make 100 ml. This solution is isotonic in relation to the blood of the fishes. After fixation the scales were washed with NaCl 0.1 A" and dehydrated at +4°C over 70 per cent, 95 per cent and absolute alcohol, 10 minutes in each. The last change of absolute alcohol was carried out at room temperature. The scales were embedded in a mixture of «-butyl-metha- crylate and methyl-methacrylate containing 0.3 per cent benzoyl-peroxide as a catalyst. Then the mixture was polymerized at 40"C. The sections were made with the Sjostrand Ultramicro- tome. Without dissolving the embedding medium the sections were examined in an RCA EMU 2c electron microscope. The melanophore is surrounded by a thick cell membrane with a structure similar to that of base- ment membranes in mammalian cells. The thickness of this membrane is about 1200 A. No internal structure can be observed. Outside the thick cell membrane a less dense zone is often present. An inner cytoplasmic membrane constitutes an inner "sack". This sack contains the nucleus, mito- chondria and the dark pigment granules. The thick- ness of this inner cytoplasmic membrane is estimated to about 80 A (figs. 1-2). The nucleus is about 6 /< in diameter. The nuclear membrane consists of two opaque layers sepa- rated by a less dense zone with a total thickness of about 300 A. The nuclear content consists of about 170 A thick granules and a nucleolus, about 1 /< in diameter. The mitochondria are oval or rod-shaped and surrounded by an external double membrane. Inside this membrane is a system of internal double mem- branes arranged in an irregular way. No interme- diate forms between mitochondria and pigment granules have been observed in contrast to du Buy et al. (2). Within the sack there also appear special cyto- plasmic structures. They consist of more or less dense annular and irregularly formed single mem- branes. They may represent the Golgi apparatus, but they can also be special structures in the melano- phore cytoplasm. The spherical pigment granules, containing mela- nin, range from 0.25 /< to 0.55 /n in diameter, the mean value being 0.40 /<. They are surrounded by a membrane about 300 A thick. The granules generally contain a number of dense bodies, but may some- times appear homogeneous. Between the two cell membranes is a zone 0.5-3 ft wide. This zone is transversed by fibrils, about 80 A thick. In general the fibrils do not seem to be attached to either of the two cell membranes. They are pre- ferably arranged in a direction parallel to the cell Mechanism of Pigment Migration within Teleost Melanophores 215 Fig. 3. Fibrillar zone of a contracted melanophore, the width being about 3.5 microns. Magnification 20,000. Fig. 4. Fibrillar zone of a dispersed melanophore, the width being about 0.4 micron. Magnification 17,000. Fig. 5. Longitudinally cut fibrils. Magnification x 33,000. Fig. 6. Transversely cut fibrils. Magnification < 55,000. membranes. The width of this fibrillar zone varies with the pigment dispersion. By fully dispersed melanophores the width is about 0.5 n and by fully concentrated melanophores the width is about 3 n (figs. 3-6). The fibrils are regarded as representing contrac- tile structures, which as a basket mesh-work surround the inner sack. Thus when the fibrils contract, the Fig. 7. Schematic representation of the observations made on teleost melanophores. sack diminishes and the pigment granules are con- centrated. On the other hand when the fibrils relax, the sack can increase and the granules will be dis- persed. The suggestion is thus made that the migration of the pigment granules within the teleost melano- phores should be due to the contracting and relaxing fibrils in the zone between the two cell membranes. From a physiological point of view it also seems to be the concentration of the pigment granules that requires most energy. This investigation gives a completely new explana- tion of the mechanism of pigment migration within teleost melanophores. It abolishes many previous theories of pigment migration (1. 3-9). This work has been supported by a grant from the Kungliga Fysiografiska Sallskapet, Lund. References 1. Ballowitz, E., Pfl tigers Arch. ges. Physiol. 157, 165 (1914). 2. Du Buy, H. G., Woods, M. V.. Bdrk, D., and Lackey, M. D., /. Nat. Cancer Inst. 9. 325 (1949). 3. Franz, \.,Z.Zellforsch. 30, 194(1939). 4. Hooker, D., Amer. J. Anai. 16, 237 (1914). 5. KoLTZOFF, N. K., Compi. rend. acad. sci. URSS, N.S. 28, 554 (1940). 6. Marsland, D. a., Biol. Bull. Hoods Hole 87, 252 (1944). 7. Schmidt, W. J., Biol. Zentr. 39, 140 (1919). 8. Shanes, A. M. and Nigreli i, R. F.. Zoologia 26, 237 (1941). 9. Wright, P. A. and Sabal, J., Biol. Bull. 103, 312 (1952). The Pharyngeal Protein Fibres of the Ciliates Ch. Rouiller, E. Faure-Fremiet and M. Gauchery Laboratoires d' Enibryologie experiinentale et cle Medecine experimentale dii College de France, Paris; Iiistitiit de Recherches siir le Cancer Gustave Ronssy, Villejuif 1 HE Ciliates are distinguished amongst unicellular organisms by the diversity and complexity of their cytoplasmic differentiations. Electron microscopy allows us to make new studies both of the ultrastruc- ture and comparative morphology of various orga- nites such as myoid fibres and certain skeletal appa- ratus (3-9). Among the latter, we will examine here protein fibres which in the Gymnostomata and some Hypo- W stomata form below the mouth, in the endoplasm, a pharyngeal framework guiding the introduction of ingesta. The properties of these rigid and elastic fibres, birefringent and easily stained with mercuric bromophenol blue, are readily comparable to certain scleroprotein. In the prostomal and pleurostomal Gymnosto- mata (called Rhabdophorina by Corliss (1), from Faure-Fremiet (2)) this framework is built up from ^ .^- ,r- w • > .« * •! A V / : ^ Fig. I. Coleps hirtus. Cross section of the anterior part of the body. B, buccal depression. T, cross sections of the trichites constituting the pharyngeal framework around the buccal funnel. M, mitochondria. Around the buccal depres- sion there are some radial cytoplasmic lamellae (L) and many granules (G). Magnification 31,000. The Pharyngeal Protein Fibres of the Ciliates 217 9 Fig. 2. Nassiihi iiiirea. Cross section of a part of the pha- ryngeal basket. A, proteic rods; B, peripheral sheath with (C) aliform crests, and (D) attempts at cristallisation. Magnification 31,000. a nearly cylindrical bundle of protein rods — or trichites — quite distinct from the trichocysts wfiich may coexist with them. The transverse section on the level of the buccal depression of the Coleps hirtiis (fig. 1) shows the order of these trichites. They are themselves built up from a bundle of long parallel homogeneous fibrils. To return to transverse sections one may see that these fibrils are regularly arranged on a series of longitudinal parallel planes and so evenly spaced that the structure has a cristalline appearance. Comparable micrographs have been obtained from two species of Prorodon, but in one of these, P. nucleolatus, the arrangement of the fibrils is less regular, due to the presence, near the buccal extre- mity of the trichites, of an amorphous interstitial substance. In the hypostomal Gymnostomata (called Cyrto- phorina) the pharyngeal framework appears as a much more clearly defined organite, the basket, whose protein rods, in general thicker than the trichites, are united, near the mouth at least, by a peripheral sheath. This structure is shown by the different species of the genus Nassiila. In Nassiila aiirea Fig. 3. Nassiila ornata. Paracristallinc strLictiire of a pliaryn- geal proteic rod. Magnification 68,000. Fig. 4. Frontonia niari/ia. Cross section of tlie pharyngeal proteic fibres (F). M, mitocliondria. Magnification 30,000. (fig. 2) the twenty-five rods of the basket have an elliptical transverse section, and one of their side surfaces is extended into the pharyngeal cytoplasm by a fibrillar lamella. Their ultrastructure shows a dense and remarkably regular packing of longitu- dinal and parallel fibrillar planes. The peripheral sheath which unites the rods extends its outer surface with aliform crests. It is eqiiallN built up from an en- semble of longitudinal fibrils but these are loosely arranged, although there may be observed, from place to place, some attempts at cristallisation. Nassiila ornata (fig. 3) shows the same structure with thicker rods of triangular cross sections. The basket becomes more complicated in Chla- niydoclon where at the same time there may be ob- served a pharyngeal tube of lamellar structure which runs from the buccal epiplasm to the endoplasm. The protein rods include an axial mass which is appar- ently amorphous and around which the fibrils are assembled in a random manner. Their buccal extre- mity is dilTerentiated and capped by a dense protein sheath conical in form, resembling a tooth. All these teeth form a maxillary crown. This type of difTerentiation is accentuated in 218 CH. ROUILLER, E. FAURE-FREMIET AND M. GAUCHERY Fig. 5. Fruntonia marina. Longitudinal section of some plia- ryngeal proteic fibres (F). M, mitochondria. Magnification X 34,000. By their structure and their position, these fibres are readily comparable to those which form the pharyngeal basket of Nassula. This fact should be remembered if one wishes to study the comparative morphology and evolution of the Ciliates. We have elsewhere described similar fibres found around the buccal infundibulum of the peritrichous Ciliates (5). From the cytological point of view, we see that all the fibres examined are of proteic nature. They are rigid and elastic, birefringent and built up of elementary fibrils arranged to form a paracristal- line structure. The elementary fibrils seem to be of a fairly constant diameter, between 150 and 200 A. Each of them must be considered as an already complex macromolecular entirety. The elementary fibrils are always in a parallel arrangement but their mutual ordering may be either loose and irregular, or tight and regular in the transverse plane with respect to a definite pattern. In fact we have a system which we may name smecto- nematic since the fibrils are arranged in a parallel manner (as in the nematic mesomorphic states) but in addition on a series of parallel planes (the latter bringing to mind a smectic mesomorphic state). Unfortunately, there is some doubt as to the origin of the fibres, that is to say their formation process, their growth, their assemblage, their pattern-making. The case of Frontonia indicates a possible relation- ship with the ciliary rootlets similar to those previ- ously described in Stentor (3). It is not sure whether this is so with Nassula and Chlamydodon. Dysteria by the development of the pharyngeal tube and the reduction of the protein rods to two large stylets surmounted around the mouth by two com- plex maxillae. It is known that in certain Hymenostomata a set of pharyngeal fibres dip from the mouth into the endo- plasm. In Frontonia marina (figs. 4 and 5) these long and flexuous fibres appear on the same section, cut in various planes, beside of trichocysts and mito- chondria. Of a protein nature, they are here also built up from a set of longitudinal homogeneous fibrils, regularly spaced in series of parallel planes. References 121-140 (1956) 1. Corliss, J. O., System. Zool. 5, 68-91, and Arch. Prot. (1957, in press). 2. Faure-Fremiet, E., Bull. soc. zool. France 75, 109-122 (1950). 3. Faure-Fremiet, E. and Rouiller, Ch., Compt. rend. acad. sci. 241, 678-680(1955). 4. Faure-Fremiet, E., Rouiller, Ch., and Gauchery, M., /. Protozoal. (1957, in press). 5. — Bull. soc. zool. France 81, 77-85 (1956). 6. — Arch, d'anat. microscop. 45, 139-161 (1956). 7. Rouiller, Ch., Faure-Fremiet, E., and Gauchery, M., Compt. rend. acad. sci. 242, 180-182 (1956). 8. — /. Protozoal. 3, 188-193 (1956). 9. — Exptl. Cell Research 11, 527-541 (1956). IX COLLAGEN, CARTILAGE, BONE Kollagen U. HoFMANN und K. Kuhn Ediiard Zintl-Institiit fiir anorganische und physikalische Cheniie der Technischen Hochschide Darmstadt Im Jahre 1942 haben zum ersten Mai Hall. Jakus und Schmitt (5) sowie etwas spiiter Wolpers ( 1 5) eine Quer- streifung des Kollagens mit einer Periode von unge- fiihr 650 A im Elektronenmikroskop nachgewiesen. Diese ersten Aufnahmen zeigten nur einen Hell- und einen Dunkelstreifen pro Periode. Bald darauf gelang es, die Querstreifung zuniichst in drei (6, 16), spiiter in sechs bis sieben Banden aufzulosen ( 12, 17). Hofmann, Nemetschek und GraBmann (8)erreich- ten an Sehnenkollagen eine Auflosung in zuniichst acht, dann zehn und schlieBlich dreizehn Querstrei- fen (9). Um ein groBeres Versuchsmaterial zu erhalten, untersuchten wir im folgenden KoUagenfibrillen ver- schiedenster Herkunft. So zeigten Fibrillen aus Kalbshaut zehn Querstreifen pro Periode. Es lallt auf, daB die Periodenlange bei Kalbshaut durch- schnittlich 100 A kurzer ist als bei Sehnenkollagen. — Die Fibrillen nicht ausgereifter Kollagenfasern, so z. B. von embryonaler Achillessehne, sind diinn und kontrastarm und von einer storenden Zwischensub- stanz behaftet, die durch vorherige Behandlung mit Trypsin teilweise entfernt werden konnte. Wir konn- ten acht bis neun Querstreifen beobachten. — Auch Reticulinfibrillen zeigen eine periodische Querstrei- fung. Abb. 1 zeigt neun bis zehn Querstreifen pro Periode, deren Schwarzung aber vom Normalen abweicht, so daB eine Indizierung der einzelnen Streifen Schwierigkeiten macht. Nach Nemetschek, GraBmann und Hofmann (9) konnen aber auch bei normalem Sehnenkollagen die Intensitiiten der ein- zelnen Querstreifen betriichtlich schwanken. Behandelt man kollagenes Material mit verdunn- ten organischen Siiuren (19) oder mit sauren PuflFern (10), so geht ein Teil des Kollagens in Losung. Aus diesen Losungen kann man durch Dialysieren gegen verdijnnte Kochsalzlosung oder schwachal- kalische PufTerlosungen Fibrillen abscheiden, deren Querstreifungsmuster bis auf eine etwas abweichende Schwiirzung mit dem des normalen Kollagens ijber- einstimmen. Abb. 2 zeigt dreizehn Querstreifen. in Gegenwart von geringen Mengen kohlenhy- drathaltiger Substanzen, wie -/, -Glycoprotein oder PP*^ Chondroitinsulfat, kann (7) aus denselben KoUagen- losungen eine Kollagenart abgeschieden werden, die in der Natur nicht angetroffen wird, die sogenannten long-spacing-Fibrillen. Sie unterscheiden sich von normalem Kollagen durch eine weit groBere Periode von 1500 bis 3000 A und einem voUig anderen, und zwar symmetrischen, Querstreifungsmuster. In Gegenwart von Nucleinsaure oder Adenosin- triphosphorsiiure bildet sich noch eine andere Art, die sogenannten long-spacing-Segmente (13). Bei den bestaufgelosten KoUagenfibrillen mit drei- zehn Querstreifen diirften die kleinsten Einzelheiten, welche noch in Faserrichtung erkennbar sind, kaum groBer als 15 A sein. Die photometrische Ausmes- sung der einzelnen Fibrillen iiber alle sichtbaren Perioden hinweg zeigt, daB innerhalb einer einzelnen Fibrille die Lage der Schwiirzungsmaxima der ein- zelnen Querstreifen mit einem Fehler von weniger als 10 A festgelegt werden kann (8). Man stoBt so mit optischen Mitteln in Bereiche vor. welche nur noch von der GroBenordnung einiger Aminosiiuren sind. Fiir die Entstehung der Querstreifung hat man noch keine befriedigende Erkliirung gefunden. Zwei Deutungen sollen erwiihnt werden. Nach der einen werden die dunklen Querstreifen durch Anhiiufung schwerer und Phosphorwolframsaure (PWS) bevor- zugt bindende, also basische Aminosiiuren, gebildet. Nach der anderen werden die dunklen und hellen Querstreifen durch verschiedene Anordnung der EiweiBkettcn gebildet. Die Annahme (1), daB die ungeordneten und lockeren Teile der Periode die PWS bevorzugt ein- lagere und deswegen nach Anfiirbung dunkler er- scheine als die geordneten und an sich dichteren Teile, ist zweifelhaft, da man auch ohne PWS-Anfar- Abb. 1. Reticulumfibrille aus Katzenmilz mit Hyaluronidasc und anschlieBend mit PWS behandelt. lOfach unterteilte Querstreifung. Elektr.-opt. Vergr.: 42 000 , Endvergr. 120 000 , Periode 500 A. Abb. 2. Aus Citratlosung abgeschiedene Kollagen-Fibrille, mit PWS behandelt. 13fach unterteilte Querstreifung. (Unge- wohnliche Schwarzung einzelner Banden: a^ besonders stark und scharf geschwarzt, h^ intensiver als b.,). Elektr.-opt. Vergr.: 48 000 , Endvergr. 160 000 , Periode 800 A. KoUagen 221 bung durch stiirkere Bestrahlung im Hlektronon- mikroskop (8) und durch Erhitzen im Hochvakuum (9) eine hochunterteilte, wcnn auch nicht so scharfe Querstreifung erzielen kann. Nach der Annahme von Bear und Mitarheiter miiRte die Aufnahmc cin Ncga- tiv einer mit PWS angefarbten Fascr scin, was nicht /.utrifft. Die RoUe der PWS ist nocii ungekliirt. Wic wir finden, bindet Kollagen bis zu 80 "o seines Gewichtcs PWS. Diese liiBt sich aber leicht bis zu 40 "„ auswaschen, ohne da(3 dabei der Kontrast der Quer- streifung schwacher wird. Die PWS kann also nicht nur eine beschwerende Wirkung habcn, sondern scheint auch noch irgendwie chemisch auf die Kol- lagenfaser einzuwirken. In diesem Zusammenhang ist es interessant, daB Banga und Balo (18) gel'unden haben, daB Kollagen, welches mit PWS behandelt wurde, in Gegensatz zu nativem Kollagen von dem Enzym Elastase angegriffen wird. Viellcicht spielt hier die nicht unbetriichtliche Oxydationswirkung der PWS eine Rolle. Versuche in dieser Richtung sind bei uns im Gange, aber noch nicht abgeschlos- sen. Hofmann, Nemetschek und GraBmann (8) haben gezeigt. daB intensive Bestrahlung im Elektronen- mikroskop charakteristische Veranderungen her- vorruft. Zuniichst zerfallen die dunklen Querstreifen in dunkle. in der Langsrichtungder Faser orientierte Einzelstreifen. Bei starkerer Belastung des Objekts entstehen anstelle dieser dunklen Liingsstreiien Locher, das sogenannte ,,Gluhstrumpfstadium".Zu- ]etzt bilden sich dunkle, kugelahnliche Partikelchen, die unregelmiiBig iiber die Fibrille verteilt sind. Ahnliche Veranderungen erzeugt auch das Erhitzen der Faser im Hochvakuum (9). Die Kenntnis der chemischen Reaktionen, welche diese Veranderungen begleiten, ist von groBer Wich- tigkeit. Zu diesem Zweck haben wir Fasern mit charakteristischen Zersetzungsstadien, die wir durch Erhitzen im Hochvakuum herstellten, auf ihre Ami- nosaurezusammensetzung mit Hilfe zweidimensio- naler Papierchromatographie untersucht. Die beiden ersten Stadien, also das Auftreten einer unterteilten Querstreifung und der Zerfall dieser Streifen in langsgerichtete dunkle Partikelchen lassen nur die Aminosauren in normaler Zusammensetzung erken- nen. Erst bei fortgeschrittener Zersetzung bemerkt man auf den Chromatogrammcn dunkle Flecke mit sehr geringen Rf-Werten, wahrschcinlich Verkoh- lungsprodukte. Mit Sicherhcit liiBt sich sagen, daB bei den Belastungen, denen die Fibrillcn im Elek- tronenmikroskop normalerweise ausgesetzt sind, keine durchgreifenden chemischen Umsetzungen vor sich gehcn. Wie wir aber feststellten, werden mit Elektronen bestrahlte Fibrillen im Gegensatz zu nativem Kollagen von Trypsin angegrifTen. In den letzten Jahren ist ofters die Vermutung ausgesprochen worden, daB Kohlenhydratverbin- dungen eine wesentliche Rolie bei der Bildung und dem Aufbau von Kollagenfibrillen zukommt. So ent- halten Kollagenfascrn geringe Mengen von Kohien- hydraten (vsenigcr ais I "„) (4). Die Grundsuhstanz des Bindegewebes, in der sich die Kollagenfibrillen bilden, enthiilt Kohlenhydratvcrbindungen wie saure Polysacharide, z. B. Chondroitinsulfat. Solche Ver- bindungen beeinflussen die I- ihrillcnbildung wcsent- lich (7). Weiter wichtig ist der Bcfund, daB man Kollagen- fascrn durch Bchandlung mit Natriumperjodat und Phenyljodosoacetat in Kettenabschnitte aufteilen kann, die nur noch 20-25 Aminosauren lang sind (3). Diese spezifischen Qxydationsmittel greifen aber von den hicr in Betracht kommenden Substanzen nur Kohlenhydrate oder vorsichtiger ausgcdriickt kohlenhydratarlige Substanzen an. Die von Dettmer und Schwarz (2) entwickelte Perjodat-Silberurotropin-Reaktion, mit der man Kohlenhydrate im Elektronenmikroskop sichtbar machen kann, bcruht cbcnfalls auf der Oxydations- wirkung des Perjodats. Die bei der Oxydation entste- henden Aldehydgruppen lagern bei der folgenden Versilberung Silberkorner an. Mit der Perjodat-Silberurotropin-Reaktion konnte Pahlke (11) zeigen, daB die Finlagerung \on Silber- kornern bei embryonalen Sehnen vor allem in der amorphenZwischensubstanz erfolgt. Mit zunehmen- dem Alter scheiden sich die Silberkorner immer mehr an der Oberflache der Faser ab, um sich beim Erwachsenen schlieBlich periodisch in die Dunkel- teile der Querstreifen einzulagern. Nach Schwarz (14) gilt nur die periodische Innenversilberung als charakteristisch fiir ausgereifte Kollagenfibrillen, denn Quersti-eifung nach PWS-Anfiirbung zeigen auch embryonale und reticulare Fibrillen. Zur weiteren Klarung dieser Probleme haben uns hauptsiichlich vier Fragen beschaftigt: 1. Lagern sich die Silberkorner bei der Perjodat- Silberurotropin-Reaktion nur in den Dunkel- teilen der Fibrillen ab? 2. Ist eine geordnete Ablagerung von Silberkor- nern ein Merkmal gereifter und eine ungeord- nete Ablagerung ein Merkmal ungereiftcr Fi- brillen ? 3. Erfolgt die Ablagerung der Silberkorner im Innern der Fibrille durch ihren ganzen Quer- schnitt hindurch? 4. Kann man mit Hilfc von Enzymcn. wie H>alu- ronidase oder Trypsin, die perjodatempfind- lichen Gruppierungen von den Fibrillen ab- trcnnen? Dazu untersuchten wir zuniichst als Standard- produkt einer ausgereiften Fibrille die Achillesschne eines erwachsenen Menschen. Die Anfiirbung mit der Perjodat-Silberurotropin-Reaktion zeigte das schon von Pahlke beschriebene Bild. Es hebt sich deutlich die normale Pcriode des Kollagens mit bis zu drei Silbcrquerstreifen pro Pcriode heraus. Ahnlich ist es bei Fibrillen aus Kalbshaut: man kann cinen Silbcrquerstreifen pro Periode erkennen. Auch bei embryonaler Achillesschne und bei reti- 222 U. HOFMANN UND K. KUHN ■f^'m.9'1 2*. m ■.•^'^i^' Abb. 3. ReticLiliimfaser aus Katzenmilz, mit Trypsin behan- delt. Perjodat-Silberurotropin-Methode (15 Min. mit Natri- umperjodat bchandelt, 18 Stdn. versilbert). Elektr.-opt.Vergr. : 12000 , Endvergr. 22 000 . Periode ca. 500 A. ^ly 0. Abb. 5. Long-spacing-Fibrillen aus saurer Prokollagen-Lo- sung in Gegenwart von ai-Glycoproteid hergestelit, mit HyaJLironidase behandeit. Perjodat-Siiberurotropin-Methode (15 Min. mit Natriumperjodat behandeit, 8 Stdn. versilbert). Bis zu sechs Silberquerstreifen pro Periode. Elektr.-opt. Vergr.: 11700 , Endvergr. 29 000 , Periode 1500 bis 2000 A. cularen Fibrillen aus Katzenmilz, Abb. 3, konnten wir eine schwache aber doch typisch regelmaBige Versilberung beobachten, nachdem wir die storende silberanfiirbbare Zwischensubstanz mit Trypsin ent- fernt batten. Fibrillen aus Citratlosung weisen eine besonders regelmaBige und sehr scharfe periodische Silberablagerung auf mit drei Silberquerstreifen pro Periode, Abb. 4. Die Versilberung der long-spacing- Fibrillen liiBt eine deutlich periodische Silberablage- rung von mindestens sechs Querstreifen pro Periode erkennen. Die Schwarzung der Querstreifung haben wir beim gleichen Praparat und gleicher Entwicklung verschieden gefunden. Abb. 5, 6a. Die Klarung der Frage, inwieweit sich die Silber- korner in den Dunkelteilen ablagern, ist aus den cben gezeigten Abbildungen nicht eindeutig zu entnehmen, weil die verhiiltnismaBig grobe Silber- ablagerung die Querstreifung verdeckt. Wir haben daher schwach versilberte Fasern nachtriiglich mit PWS angefiirbt. Die Silberkorner lagern sich haupt- Abb. 4. Aus Citratlosung abgeschiedene Kollagen-Eibrillen, mit Trypsin behandeit, Perjodat-Silberurotropin-Methode (15 Min. mit Natriumperjodat behandeit, 8 Stdn. ver- silbert). Bis zu drei Silberquerstreifen pro Periode. Pfeil 1 : Dicke Fibrillen, sehr regelmaBige Versilberung. Pfeil 2: Weniger dicke Fibrillen, periodische, aber schon etwas unregelmaBige Versilberung. Pfeil 3: Diinne Fibrille, un- regelmaBig versilbert. Elektr.-opt. Vergr. : 1 1 600 , End- vergr. 27 000 , Periode ca. 650 A. sachlich in den Dunkelteilen ab, konnen aber auch in den Hellteilen beobachtet werden. In Abb. 6 haben wir Aufnahmen des gleichen long-spacing- Praparates einmal versilbert (Abb. 6o), einmal unbe- handelt (Abb. 6h) einander zugeordnet. Die Zuord- nung kann hier eindeutig erfolgen, da es Aufnahmen gibt, bei denen die Dunkelteile durch die Versilbe- rung hindurch sichtbar sind. Man sieht, daB die intensiver hervortretenden Dunkelteile keineswegs auch bei der Versilberung betont sein miissen. Im Gegenteil treten hier die beiden AuBenstreifen beson- ders hervor. Eine regelmaBige periodische Silberablagerung ist nach unserer Meinung bevorzugt eine Folge der Dicke und Festigkeit der Fibrillen. Dies sieht man am deutlichsten in Abb. 4. Die dicksten Fibrillen zeigen eine sehr regelmaBige Ablagerung. Bei diin- neren Fibrillen fa lit die Versilberung nicht mehr so geordnet aus. wahrend bei ganz diinnen Fibrillen Silberkorner unregelmaBig iiber die Fibrille verstreut sind. Zu der Frage. ob die Silberkorner im Tnnern der Fibrille abgelagert werden, haben wir von alien hier gezeigten Fibrillen Stereoaufnahmen gemacht. Sie zeigen deutlich, daB der groBere Teil der Silberkor- ner in den AuBenbereichen der Fibrillen abgelagert werden, ein Teil befindet sich aber auch im Fibril- leninnern. Diese Ablagerung ist unabhangig davon. Abb. 6. Long-spacing-Fibrillen, wie bei Abb. 5 hergestelit, mit Hyaluronidase bchandelt, a) nach Perjodat-Silberuro- tropin-Behandlung. Elektr.-opt. Vergr.: 12 600 , b) ohne AnfarbungElektr.-opt. Vergr. : 12600 , Endvergr. 38 000 x , Perioden ca. 2000 A. Gleiche Strukturelemente unterein- ander. Neiic Bcfimde zur Stniktur der Sehncnfihiille 223 ob die Silberkorner regelmiiBig odcr unrcgclmiiBig angeordnct sind. Versuche, die perjodatempfindlichen Gruppierun- gen von den Fibrillen durch Hyaluronidase- oder Trypsinbehandlung abzutrennen, waren crfolglos. Im Gegenteil vvurde bei embryonalcn Fascrn sowie bei Reticulin dadurch die storendeZwischensubstanz von den Fibrillen abgelost, und die Versilberung vvurde so deutlicher sichtbar. AbschlieBend neigen wir aiif Grund unserer elelme. HM<^ > ^:. ^ Fig. 1. Damage of the structural organization of collagen fibrils produced by treatment in acid solution. 228 G. LELLI, U. MAROTTA AND A. D AMORE In a second group of experiments in which we looked for possible structural modifications of the collagen fibrils after freezing we divided the tests, 20 in all, into two subgroups. In the first of these we subjected samples of collagen to temperatures ranging from - 50 C to - 70'C for 24 hours, they were then lacerated in double distilled water and examined. Nothing particular regarding the thickness, the periodic structure and the length of the period was found in these fibrils. In the second subgroup, designed to obtain more exact information, the collagen fibrils were lacerated in double distilled water and then examined and repeatedly photo- graphed. They were then subjected for 24 hours to a temperature of - 70 C and then again photographed electron microscopically in the same microscopic field. A comparative study of the electron micrographs thus obtained and an acurate measurement of the length of many periods showed that freezing produced no substantial changes in the periodic structure of the collagen. However it was evident from visual obser- vation and even more so from checks of the electron micrographs with the comparator that there is a constant and uniform thinning of the fibrils after freezing. The degree of thinning is usually moderate and at most one fifth of the original thickness. Conclusions. — (a) Extreme cold applied to collagen produces in the fibrils an increased resistance to the destructive action of acid solutions and to the action of some proteolytic enzymes such as papain. The hardening which may occur in frozen meat may be due to this increase in the resistance of the collagen. (b) Cold does not produce appreciable structural changes in the collagen fibrils when seen in the electron microscope. This is in agreement with earlier observations (I, 2). (c) Cold produces a greater or lesser degree of thinning which is constant in collagen fibrils. References 1. Lelli, G., Bonanome, A., and Sappa, M.,Z. wiss. Mikro- skop. 61, 298 (1953). 2. Nemetschek, Th., Grassmann, W., and Hofmann, U., Z. Natiirforsch. 10 b, 61 (1955). Electron Microscopic Observations on Collagen Exposed to X-Rays G. Lelli, U. Marotta and A. D'Amore Istituto Superiore di Sanita, Roma In view of the ever-increasing application of radium and x-ray therapy in neoplastic diseases, the study of the effects of irradiation on the connective tissue is of great practical interest. By means of the electron microscope we have studied the efTects of x-rays on the collagen fibrils. We have irradiated at varying doses the abdominal skin of 7 pigs aged 3 months, weighing about 15 kg, of 6 guinea pigs weighing about 700 g, and some bits of skin taken from animals of the same species before irradiation. All treated animals showed rather early manifestations of circumscribed radiodermatitis, apart from the general toxic symptoms which in some cases killed the animal in the course of a few days. Specimens for biopsy were taken regularly from the irradiated area of the skin. In four cases the first biopsy was done immediately after the irradiation, and successive ones after 4, 8, 16 and 24 days, provided the animal was still alive. In two cases 6 biopsies were done 24 hours after each irradiation and after 8, 16 and 24 days if the animal was then still alive. The dermis was isolated from the biopsy material and from the pieces of skin removed prior to irradiation. It was then lacerated in double distilled water and examined under the electron microscope after previous chromium shadowing. Careful examination of a very large number of electron microscopic photographs showed that the collagen fibrils had a normal transverse striation, usually without changes of either shape or thickness. We only occasionally encountered a fibril which was swollen, ruptured, flattened, winding and apparently disintegrated in some parts. We attach little importance to such changes, as identical pictures were seen in the controls. More- over, these changes were not in any way related to the x-ray dosage applied. Nemetschek et al. (2) using x-ray doses smaller than ours, found under the electron microscope that the normal transverse striation of the fibrils persisted and that these showed a fine granular appearance. This we could not see in our material which was always examined after chromium shadowing. Also Marotta (I) excludes electron microscopi- cally demonstrable changes in the collagen fibrils after a single dose of 700 r. Our negative results are also in agreement with those recently obtained by Nerli (3) using the method of x-ray diffraction. Indeed this author found no difference whatsoever in the diagrams he obtained using dermis from the skin of rats irradiated with a single dose of 1200 r without filter and non- irradiated derma. According to Nerli this suggests that "x-ray irradiation does not change the basic structure of the polypeptide chains of the collagen". References 1. Marotta, U., Reiul. ist. super. Sanita 18 (1955). 2. Nemetschek, Th., Grassmann, W., and Hofmann, U., Naturwissenschaften 16. 371 (1954). 3. Nerli, A., Rac/ioter. Radiobiol. e Fis. Med. 2, 166 (1956). Studies on the Fibrogenesis of Collagen Sylvia Fitton Jackson Medical Research Council Biophysics Research Unit, Wheat sioiie Laboratory, King's College, London, W.C.2 Collagen protein is widely distributed in the various phyla of the animal kingdom (3; Rudall, 1955). Recognition of its presence may be based in the characteristic fibre diagrams obtained from va- rious materials by means of high angle x-ray ditlYac- tion. Though different sources of collagen give fibre diagrams indicating substantial differences of orien- tation and crystallinity. they are all recognizably belonging to the same class of protein. On the other hand, there are variations, for instance, in the amino- acid composition and in the structure of the con- stituent fibrils as seen by means of electron micros- copy. It is of interest, therefore, that though a typical fibre diagram for collagen is obtained from the cuticle of annelids (I), the characteristic axial perio- dicity of about 640 A of the fibrils is apparently lacking (4). It was thought that an investigation of the fine structure of the cuticle of the annelid, Liimbricus sp., by the use of thin sections in the electron microscope might help to elucidate the Fig. 1. Section through the ctiticlc of Liiinhriciis sp. whicli shows the layers of fibrils which lie adjacent to the epidermal cells. Cytoplasmic processes stretch between the cell surface and the outer membrane of the worm, and form a basket- weave pattern with the fibrils. The microvilli on the outer surface of the worm are seen at the top of the micrograph. Magnification 23,000. method whereby the protein molecules became orien- tated into the distinctive layered structure which has been shown to be a feature of this material. This work is still in a preliminary stage, but it has been found that the cuticle is composed of five regions. The region adjacent to the main body of the epidermal cells, which lie immediately beneath the cuticle, contains many evenly spaced folds of cytoplasm; these folds appear to taper into fine cytoplasmic processes which penetrate through the cuticle and are connected to the exterior membrane of the worm (fig. 1). In sections cut parallel to the surface of the worm, these processes are seen to be arranged in rows; they are usually about 700-1000 A in diameter and about 2000 A apart. Each row appears to be embedded in a ribbon of less dense material about 2500 A wide. It is possible that the processes correspond to the granular layer recorded by Read and Rudall (4) in replicas of the earthworm cuticle. Region 2, immediately above this lowest layer, con- tains fibrils about 200 A in diameter; there are about 4 layers of fibrils in this region. Region 3, which is about 5 // thick, is composed of about 18 layers of apparently unhanded fibrils. Transverse sections show that the fibrils are irregular in outline, in fact some are nearly square, and measure up to about 2000 A across (fig. 2). Longitudinal sections demon- strate that the fibrils are in layers, which confirms previous observations. Measurements have shown that the layers of fibrils are orientated from 74' to 106' to each other; these measurements vary depending on whether the worm was contracted or extended on fixation, and suggest that the layers Fig. 2. Transverse sections of the fibrils of the cuticle, which shows they are irregular in outline. Magnification 30,000. Fig. 3. The fibrous honeycomb of amorphous material which surrounds each fibril. Magnification 23,000. 230 M. K. KEECH AND R. REED may slide over each other to allow for the alteration in the contour of the animal. No banded structure has been observed, but the fibrils are surrounded by an amorphous material which tends to make precise observation difficult. This latter material is formed into a fibrous honeycomb which surrounds each fibril (fig. 3). It is possible that this material may re- present a mucopolysaccharide; hence it may form an underlying lattice system which may be connected with the mechanism of determining the orientation of the fibrils. Region 4 lies just beneath the outermost layer; it sometimes contains a few fibrils but more often consists only of a thin amorphous zone. The outer- most layer, region 5, is composed of a double mem- brane, from which project numerous microvilli about 1500 A long and 500 A in diameter. Reed and Rudall (4) described this region as a corpuscular layer, but observations on transverse sections of the cuticle show clearly that the exterior of the worm is covered by a system of microvilli. This finding may account for the presence of the numerous fine cyto- plasmic processes which penetrate the cuticle, for it is unlikely that the microvilli could exist in an extra- cellular position. Previous work has demonstrated that in avian tendon the collagen fibrils of developing tissue in- crease in diameter v^hile there is a relative decrease in the amount of interfibrillar material which sur- rounds the fibrils; it has been concluded that this material contains collagen molecules which are depo- sited onto the growing fibrils in the form and packing appropriate to the characteristic fibre diagram of the collagen protein (2). In the present work, it is apparent that the layers of fibrils adjacent to the epidermal cells are of smaller diameter than the main fibrils of the cuticle. If it is assumed that the epidermal cells secrete the precursors of the cuticle, it follows that the internal layers may be composed of fibrils that are in the process of growing. Thus, it may be suggested that, as each fibril layer is formed, the next adjacent layer may determine the orienta- tion of this new layer. On the other hand, it is possible that the precisely arranged cytoplasmic processes may take part in the orientation of the fibril layers. It is hoped that future studies on the regeneration of the cuticle will clarify the method of fibril elaboration and layer formation. References 1. AsTBURY, W. T., Essays on Growth and Form. P. 309 (1945). 2. Jackson, S. Fitton, Pioc. Roy. Soc. B. 144, 553 (1956). 3. Marks, M. H., Bear, R. S., and Blake, C. H., /. E.xptl. Zool. Ill, 55 (1949). 4. Reed, R. and Rudall, K. M., Biochini. Biophys. Acta 2 7 (1949). Further Observations on the Transformation of Collagen Fibrils into ''Elastin'' M. K. Keech and R. Reed Departments of Medicine and Leather Industries, Leeds University, Leeds Xhe dermis consists mainly of collagen (70 80 g per 100 g dry tissue) and a small quantity of elastin, all embedded in the jelly-like ground substance. Previous work (7) on the action of collagenase on human abdominal skin from different age-groups, described a breakdown product called "moth-eaten" fibres (MEF). They figured prominently in colla- genase-treated material derived from persons be- tween the ages of 1 and 20 years, but their numbers steadily decreased as the age of the collagen source increased. At first sight it was difficult to understand the appearance of these large and very dense struc- tures during an enzyme degradation process, so further investigations were undertaken in an effort to establish their identity (8). Identical experiments designed to obtain more information about the "alkali-produced elastin" (APE) described by Bur- ton et al. (I) were also reported. Prepared collagen. — Samples of the same substrates used in previous work (5, 6, 7) were employed, i.e., human abdominal skin collagen from persons of different age- groups, purified by the method of Neuman (9, 10) as shortened by Keech (5, 6). This material contained a very small quantity of the three morphological varieties of fully-formed elastin described below. Preparations of "moth-eaten' fibres (MEF). — Prepared collagen was incubated with collagenase in phosphate buffer (pH 7.4) for 24 hours at 37~C. The remaining material was centrifuged, and the pellet treated one of the following ways: — (a) Re-suspended in sterile distilled water, pH 5.6, and heated at 55 C for 1 hour, at 75'C for a further hour and finally at 100 C for 1 hour. (h) Re-suspended in a mixture of 1 "o sodium meta- periodate and 1 % NaCI in phthalate buffer at pH 5.0 and incubated for I h hours at 37''C. ((■) Re-suspended in borate buffer (pH 8.8) containing 0.2 mg elastase and incubated for 6 hours at 37C. (d) Re-suspended in 2 % acetic acid, allowed to stand at room temperature for an hour and then heated at 100 C for 1 hour. Preparations of alkali-produced ''elastin" (APE). — The method used in previous work (1, 3) to effect the apparent transformation of collagen into elastin was employed, i.e., prepared collagen was incubated in borate buffer (pH 8.8) for 24 hours at 3TC. The centrifuged pellet was then treated as in (a) and (d) above. In addition it was Transformation of Collagen Fibrils into "Elastin" 231 incubated in phosphate buffer (pH 7.4) containing 0.2 mg elastase for 6 hours at 37 C. 0.05 ml. of a penicillin and streptomycin mixture was added to each test tube in the above experiments, and this successfully prevented bacterial contamination. CI. histolyticiini collagenase was kindly supplied b\ Dr. J. D. MacLennan; the elastase was prepared by Hall and Gardiner (2). Samples of three of the substrates (aged 2, 9 and 78 years) were heated in sterile distilled water to 55 C for 1 hour, to 75 C for a further hour, and finally to 100 C for an hour. Prior treatment with either collagenase or alkaline buffer was of course omitted. Samples for electron microscope examination were taken at each stage in the experiments, and ground gentl> in a glass tissue-grinder until the material appeared milky; drop preparations were made which were shad- owed with chromium and examined in a Siemens electron microscope, type UM 60 C. Counts of the different kinds of elastic structures were made by carefully scanning two grids from each sample. The grids contained 16 squares, and each part of each square was examined in sequence. As about 20 fields covered a square, a total of 640 fields was scrutinised from each specimen. The wide spectrum of morphological change pro- duced exhibited marked age-differences. 1. Heat controls. Heat alone on the untreated 2-, 9- and 78-year-old prepared collagen samples simply produced progressive gelatinisation as the temperature increased, which was complete after boiling for one hour. There was no age-difference, no increase in elastic structures and none of the new structures described below. The fully-formed elastin present in the starting material was some- what reduced in quantity, whilst the component elastin filaments became particularly well defined. Fully-formed elastin. This is a normal component of human fresh whole dermis and prepared dermal collagen from all age groups. Natural elastin forms about 1-5 "o of the histologically stained dermis, but the amount normally present increases with age. The following variants can be distinguished and counted under the electron microscope: (a) Skin-type elastin (which forms about 95 % of the elastin) denotes dense, irregular, structures without any regular features (fig. 1). (h) Filamenting elastin (which constitutes about 5 "o of the elastin) is usually found in long, twisting, ribbons. (c) Large natural networks (which account for 1 "^o of the elastin) differ from the manufactured networks (MN) described below in being completely free of amorphous material or dense bits and seem to be a network variety of skin-type elastin. They occurred sporadically in all preparations examined. This fully-formed elastin was unaffected by heat, although in the 56- and 78-year-old preparations there was a suggestion of concentration (i.e. higher counts) as the collagen gelatinised. 2. Effect of heat on collagenase-treated prepara- tions. Over the age of 50 years collagenase only produced slight macroscopic digestion of the col- lagen samples without the production of true MEF. Fig. 1. Typical example of skin-lype elastin from llic fresh dermis of an intlividiial aged 66 years. This variant forms ')5"„ of nauiral (fully-formed) elastin and contains a fair amount of dense amorphous material (elastomucin). The other variants (filamenting and large networks) have less elastomucin so their component filaments are better visua- lised. Magnification 12.000. Fig. 2. Typical "moth-eaten" fibre (MEF) from the prepared dermal collagen of a 2-ycar-old child after incubation for 24 hours at 37 C with collagenase (CI. histolyticiini) in phos- phate bufter pH 7.4. Note the beaded fibrils and numerous "beads" in the background. Magnification 10,000. Thus the starting material consisted simply of stri- ated collagen showing slight "standard"" collagenase change (7) and the usual ciuantity of fully-formed elastin. These components responded to heat in identical fashion to the heat controls described above. However, samples from the child and young adult substrates gave a very different picture. 85-90 "o of the collagen was digested, and the deposit consisted mainly of typical MEF (fig. 2). Gentle grinding fragmented these thick, dense fibres into portions of different lengths accompanied by a large number of angular, dense bits. These MEF proved relatively heat-resistant, a considerable degree of breakdown occurring only after boiling. In addi- tion many portions of MEF were found to terminate as "elastic networks" (MEFC) and a number of isolated networks (MN) were also seen. MEF conversions (MEFC) (figs. 3 and 4) denote moth-eaten fibres transforming into "elastin net- works", identical with the manufactured networks (MN) described below. In the two child substrates some of the MEF appeared to be transforming into sheet-like elastin in addition to the usual network type and heat markedly increased the numbers so transformed. The substrate from the 27-year-old was outstanding in consisting almost entirely of large MEF conversions throughout, in fact the starting material resembled the heat-treated prepa- rations from the 2- and 9-year-old substrates. The 28- and 43-year-old starting material showed no definite MEFC or MN but both these structures became definite and easily identifiable after heating. No conversions were seen in material over the age of 50. 232 M. K. KEECH AND R. REED •.^J^ Fig. 3. "Moth-eaten" fibre disintegrating into an "elastin network" (MEFC). Obtained after heating the MEF prepara- tion from a 9-year-old child in water for 1 hour at 55°C and a further hour at 75''C. Magnification 11,000. Fig. 4. Another example of a "moth-eaten" fibre conversion structure (MEFC) from the same preparation as fig. 3. Magnification 1 1 ,000. The manufactured networks (MN) were found only in the presence of MEFC and probably repre- sent the complete conversion or separated portions of these structures. Similar MN occurred after heating "alkali-produced elastin". Collagen was seen only in the adult specimens and exhibited slight "standard" collagenase change. 3. Effect of other reagents on MEF. A dramatic change from 90 % MEF to 85-90 % large "elastic networks" was produced by 1 % sodium periodate (in buffer pH 5.0) and by elastase (acting at its op- timum pH of 8.8). Also numerous large and small dense bits were noted, either converting into "elas- tin" individually or situated in groups linked together by "elastin". A more prolonged incubation with periodate (3 hrs. at 37°C) caused disappearance of the MEF and elastic structures, but numerous small, round, dense balls were found, similar to those described by Hall et al. (4) as a product of the action of elastase on elastin. Incubation of the starting material in borate buffer pH 8.8 alone did not pro- duce any change, but prolongation of incubation with elastase caused rapid digestion of the "elastic networks". Two per cent acetic acid at room temperature produced no macroscopic change but under the electron microscope the dense bits were broken up into smaller pieces and MEFC and MN were seen. After 1 hour at lOO'C the dense bits were markedly decreased, accompanied by a great increase in MEFC, MN and fenestrated sheets. The latter were only seen after treatment with acetic acid, and probably represent MN with a thicker layer of amorphous material than usual. 4. Effect of various agents on ''alkali-produced elastin"" {APE). Collagen and elastin have always been considered as two separate and distinct entities with different chemical and tinctorial properties. But the problem of material apparently intermediate between them, found in senile and pathological skin, has been a repeated source of debate by his- tologists over the past 90 years. Recent work (1,3) has produced biochemical, histological, and electron microscopic evidence that collagen can be converted into "elastin" /// vitro by agents such as periodate and alkaline buffer (pH 8.8). For this reason, the elastase in the present study was incubated at neutral pH in order to avoid the continued production of APE at the optimum pH value of the enzyme 8.8. The "alkali-produced elastin" (APE) proved sensitive to the elastase. The addition of 2 % acetic acid to APE led to immediate dispersal of the collagen suspensions which then appeared gelatinous. This contrasted with the lack of macroscopic effect when acetic acid was added to the (collagen-free) suspensions of MEF. Under the electron microscope the collagen exhibited typical acid disintegration and there was a marked reduction of amorphous material and dense bits together with the appearance of MEFC, MN and fenestrated sheets. The effect of heat on APE produced a dramatic change to numerous MEFC and MN with total disappearance of the collagen. Concurrent biochemical and histological studies on the same material used in this investigation lead us to believe that "moth-eaten" fibres are interme- diate structures, midway between collagen and elas- tin. In the rheumatic group of diseases, the fibrous components of the connective tissue exhibit histo- logical damage. Until the reactivity of the various connective tissue components to a wide range of stimuli (physical, chemical or enzymatic) is eluci- dated and their normal interrelationships established, one cannot interpret the pathological tissue found in disease. References 1. Burton, D., Hail, D. A., Keech, M. K., Reed, R., Saxl, H., Tlinbriejce, R. E., and Wood, M. J., Nature 176, 966 (1955). 2. Hall, D. A. and Gardiner, J. E., Biochem, J. 59, 465 (1955). 3. Hall, D. A., Keech, M. K., Reed, R., Saxl, H., Tun- bridge, R. E., and Wood, M. J., /. Gerontol. 10. 388 (1955). 4. Hall, D. A., Reed, R., and Tunbridge, R. E., E.xptl. Cell Research 8, 35 (1955). 5. Keech, M. K., Anat. Rec. 119, 139 (1954). 6. — Yale J. Biol, and Med. 26, 295 (1954). 7. — Ann. Riwiimatic Diseases 14, 19 (1955). 8. Keech, M. K., Reed, R., and Wood, M. J., /. Pathol. Bacterial. 71, 477 (1956). 9. Neuman, R. E., Ph.D. Thesis, University of Cincinnati. (1949). 10. — Arch. Biochem. 24, 289 (1949). Osteoarthritis of the Hip Joint K. LiTTi.i; and L. H. Pimm The Nuffield Orthopaedic Centre, Oxford, and The Atomic Energy Research Estahlishincnt, Harwell With the increasing longevity of the population as a whole the degenerative wear and tear changes which occur in the joints are becoming one of the greatest problems facing modern orthopaedics. These changes, which we term "Osteoarthritis'", aflfect principally the main weight-bearing joints, and thus the head of the femur in the hip joint is most often involved. The gross and microscopic changes of the established condition have been studied for many years, but little is known of the pathology underlying the origin of the lesion. In an attempt to elucidate the problem a con- siderable amount of work has been carried out at the Nuffield Orthopaedic Centre, Oxford, and studies of the histological changes and vascular patterns have been published by Trueta, Harrison and Schajowicz (3, 9). One of their fundamental discov- eries was that the initial lesion is in the hyaline articular cartilage covering the head of the femur and in the «o«-pressure area, contrary to what might be expected. The light microscope shows disruption of the superficial layer of cartilage, followed by the appear- ance of collagen bundles, lissuration and an altera- tion in staining properties, leading ultimately to ulceration and destruction of the cartilage and its replacement by fibrocartilage. The bony changes occur later and are secondary. This study and the work carried out in Sweden by Ingelmark and Saaf (5), Tngelmark (4), Ekholm (1) and Ekholm and Norbiick (2) suggests that mechanical forces are concerned with the nutrition of the cartilage from vessels lying immediately below it in the bone and probably also from the synovial fluid as there are never any vessels in hya- line cartilage itself. It seems possible that there is some intermittent pumping action as the pressure on the segments of cartilage varies, the cartilage itself behaving something like a sponge. Muller as long ago as 1929 suspected something like this and in support of the hypothesis Matthews (6) has shown that the weight bearing areas do comprise the thickest cartilage with the highest mucopolysaccha- ride content: whereas the non-pressure areas suffer nutritional deficiency changes first. Electron microscopy has made it possible to see that the collagen fibrils are closely surrounded by the polysaccharide containing matrix which appears to be of a gelatinous or spongy nature. Four distinct sizes of collagen fibrils have been noted, and in the normal articular cartilage these fibrils are present in a three dimensional network with no preferred orien- tation and surrounded by the ground substance. In the h\aline cartilage from the arthritic joint two differences are apparent. Firstly the collagen has a very definite orientation (figure I). Secondly the electron microscope shows a decreased proportion of ground substance, so that the material in the section has a rather more open appearance (figure 2). This is in agreement with the chemical analyses performed by Matthews (7), in which it was found that the total quantity of polysaccharide was de- creased considerably in arthritic cartilage. Although the onset of orientation changes can be seen easily in sections viewed in the electron micro- scope, the area of specimen examined, and indeed the whole specimen, is so small, being only about ^ mm across, that finding the direction of this orienta- tion is difficult. Information on this point can be gained by using x-ray diffraction in conjunction with electron microscopy. A survey of the changes in orientation produced both by normal ageing and by the onset of osteoarthritic changes has therefore been carried out using x-ray diffraction. Fig. 1. Hyaline cartilage from an osleoarilinlic hip joml showing the definite orientation of the fibrils. Magnification 13,000. Fig. 2. Hyaline cartilage from an osteoarthritic hip joint showing the open appearance of the ground substance typical in this condition. Magnification 26,000. 234 V. CAGLIOTI, A. ASCENZI AND A. SANTORO 33-week foetus Age 60 Arthricic Fig. 3. Diagram showing direction of orientation of the collagen fibrils in the articular cartilage of the femoral head (a) of a 33-week stillborn foetus, (b) of a normal adult aged 42, (c) of an osteoarthritic subject. In normal ageing, by the time middle age is reached the orientation on the surface is still parallel with that surface with a tendency towards preferred orientation away from the ligamentum teres. The outer two thirds or so of the articular cartilage remains un- orientated (figure 3). but there is a tendency for the inner layers of cartilage adjacent to the bone to show definite orientation at right angles to the sur- face. In arthritic hips the entire thickness of the carti- lage has its collagen fibrils orientated at right angles to the surface of the bone, with the exception still of the fibrils in the surface layer of the cartilage which remain parallel to the surface till the cartilage has begun to split and wear away. It seems that this change in the direction of the collagen fibrils is most easily explained in terms of tensional forces across the joint which is in agreement with the obser- vation that arthritis first appears in the non-weight bearing areas. With regard to the cartilage cells both living and dead cells have been encountered at all ages and in all states of the cartilage, and with the electron microscope it has been so far impossible to establish any relationship between the cartilage cells and the well-being or otherwise of the cartilage. References In the foetus and newborn child diffraction shows that the whole of the cartilage is unorientated except for the immediate surface layer in which the collagen fibrils lie parallel to the surface. This lack of orienta- tion as demonstrated by x-ray diffraction could be due to one of two causes. Either, as in the present case, because the collagen fibrils themselves are randomly orientated, or as in fibrocartilage, because the bundles of collagen are randomly orientated. It requires the electron microscope to distinguish between these two. 1. Ekholm, R., Acta Aiiat., Suppl. 15, 1 (1951). 2. Ekholm, R. and Norback, B., Acta Ortlwpaed. Scand. 21, 81 (1951). 3. Harrison, M. H. M., Schajowicz, F., and Trueta, J., /. Bone ami Joint Surg. 35 B, 598 (1953). 4. Ingelmark, B. E., Acta Ortlwpaed. Scand. 20, 144 (1950). 5. Ingelmark, B. E. and Saaf, J., Acta Ortlwpaed. Scand. 17, 303 (1948). 6. Matthews, B. F., Brit. Med. J. 2, 1295 (1952). 7. — ibid. 2, 660 (1953). 8. MiJLLER, W., Biologic der Gelenke. Johann Ambrosias Barth. Leipzig, 1929. 9. Trueta, J. and Harrison, M. H. M., /. Bone and Joint Surg. 35 B, 442 (1953). Correlation of Electron Microscopy with X-Ray Diffraction anci Optical Birefringence in the Stu(dy of the Bone V. Caglioti, a. Ascenzi and A. Santoro Institute of General and Inorganic Chemistry and Institute of Morbid Anatomy, University of Rome Engstrom and Finean (8, 10, 11), and Carlstrom and Finean (6) demonstrated that in addition to the wide angle x-ray diffraction pattern, bone tissues also give a diffuse low-angle scatter. The same authors assumed that the low-angle scatter could be treated as a particle scatter pertaining to the inor- ganic or mineral fraction. In this way they conclude that the particles are rod-shaped, the long axis of rods being aligned in the direction of the longitudinal axis of the bone, and parallel to the collagen fibres. In the intact human bone these particles appear to have a diameter of about 73 A and a length of about 210 A. Recently Robinson and Watson (15) have criti- The Study of Bone: Electron Microscopy, X-Ray Diffraction, & Optical Birefringence 235 cized the conclusions of Engstrom and Finean, be- cause observations with the electron microscope do not support the view that the inorganic particles are rod-shaped. On the other hand North, Cowan and Randall <14) have been able to show that the collagen gives a low-angle scatter not dissimilar to that obtained by Engstrom and Finean. The array of results referred to show such diver- gent points that they deserve a deeper critical exa- mination. In the hope of contributing to the solLitit>n of the problem, we have taken up Finean and Engstrom's researches on human and cow bone, integrating them with further observations with the polarizing and electron microscope. A low-angle scatter apparatus somewhat similar to that employed by Finean (9) was used. The scatter was lecorded using Ni-filtered CuA'a radiation (A ^ 1.54 A). We did not think it advisable to use monochromatic radiation since with a system of the type existing in ihe bone tissue, the use of Ni-filtered radiation (CuA'a) does not entail more substantial errors than the use of a monochromatic radiation. The intensity variation of the low-angle scatter was measured using a Leeds and North- rup automatic recording microphotometer. The electron microscope preparations were obtained employing both the mechanical dissociation in a Waring blendor and the pseudo-replica technique. The detailed description of these techniques has been reported in previous papers (2, 3). X-ray, polarizing microscope and electron microscope investigations were performed on samples of cow's and human bone. This last pertained to foetus, adult and old man. The samples comprehended untreated bone, decalcified bone and bone from which ossein had been removed by trypsin digestion or according Gabriel's method. In agreement with the results obtained by Eng- strom and Finean, the low-angle scatter of x-rays from sections of bone suggests that the scattering elements are well aligned and symmetrical around their long axes. Therefore the same elements appear as if they were the same as ellipsoids of revolution, the long axes being aligned in the direction of the longitudinal axis of the bone. The short diameter of the ellipsoids is of about 70 A and the long one of about 200 A. Still in agreement with the results of Engstrom and Finean, the low-angle scatter from the longitudinal sections of bone from which ossein has been removed according to Gabriel's method, suggests that the shape and the dimensions of the scattering elements are changed. In fact the short diameter of the ellip- soids is about 80 A, while the long diameter is re- duced to about 130 A. In addition, the intensity of diffraction seems greatly increased by the removal of ossein. These results provide arguments for the following discussion. The low-angle x-ray scatter might be interpreted either in terms of the holes irregularly placed inside a homogeneous body or in terms of a particle scatter, when inter-particle interference is not taken into consideration. In the general equation adopted by Guinier (12) in treating the scatter from such systems, the intensity of the scattered radiation appears as a function of the square of the difTerence of the electron densities pertaining to the single components responsible for the scatter (particles and surrounding medium or homogeneous body and the holes). Such a condition does not enable it to be established a priori to which of the two aforesaid systems a low-angle scatter is related. In this respect the choice of the system is suggested by evidence deriving from the data of other investigations. The identity in the low-angle x-ray diffraction pattern pertaining, respectively, to the bone, collagen tissue ( 14) and calcified collagen ( 10) does not permit the acceptance of the theory that the scattering units pertain to inorganic particles, these latter being completely lacking in collagen tissue. Therefore the scattering units must be considered as related to ellipsoidal entities (200 70 A) pertaining to the col- lagen and oriented in parallel. In bone and calcified collagen the calcium salts, enclosing the organic el- lipsoids, increase the difTerence in electronic density between the same organic particles of collagen and surrounding medium. This interpretation of the low-angle x-ray scatter from bone tissue in terms of holes (containing the organic particles) irregularly placed inside a homo- geneous body (the inorganic bone fraction) is also supported by the following arguments. {a) The form birefringence of the bone tissue from 0.009 0 008 0 007 0 005 -0003 Fig. 1. Graph of the form birefringence obtained from bone deprived of ossein according to Gabriel's method. 236 V. CAGLIOTI, A. ASCENZI AND A. SANTORO Fig. 2. Pseudo-replica of bone, etched with a 1 "o HNO3 solution. Fig. 3 a and b. Pseudo-replicas of bone digested with trypsin, in b, a thin fragment of bone stripped oti" the replicated surface. which ossein has been removed (see fig. 1) follows the Wiener law on the mixed rod-shaped body (1,7, 16). According to Schmidt who made special obser- vations in this connection, such an optical behaviour is due to the penetration of fluids at diff"erent refrac- tion indices in the ultramicroscopical holes already occupied by the micellae of osteomucoid and osteoal- buminoid. {b) A bone system resulting from inorganic parti- cles or crystallites dispersed in the ossein, according to Engstrom and Finean, could not subsist when ossein had been removed, as the particles would break down into an "incoherent" powder. On the contrary, bone from which ossein has been removed, subsists as a coherent structure though its resistance is greatly decreased. In addition, the polarizing microscope and the low- and wide-angle x-ray dif- fraction patterns demonstrate that the structural orientation of the inorganic fraction is unmodified. Finally the curve of form birefringence can be recorded from the bone without ossein (see above). (f) It is very important to note that the change observed in the low-angle scatter of the bone from which ossein has been removed according to Ga- briel's method appears as new evidence of the neces- sity to interpret the low-angle scatter of the bone as organic ellipsoidal particles enclosed in the holes circumscribed by the inorganic fraction. Indeed such a change is neither related to the orientation nor the physical state of the inorganic crystallites, both con- ditions being unmodified by Gabriel's treatment, according to the wide-angle x-ray diffraction pat- terns. Therefore it is more plausible to maintain that in the homogeneous body pertaining to the inorganic bone fraction the removal of the organic substance from the hole results in an arrangement of the material delimiting the same holes. This view finds ready support in the existence of the chemical bonds between ossein and the inorganic bone fraction (5). (d) Our studies in progress suggest that in foetal bone the low-angle scattered radiation shows inter- ference patterns. This finding can be interpreted as the result of the higher concentration of the scatter- ing elements. It may be assumed therefore that organic elements are responsible for the x-ray scatter, as they are less dispersed in foetal bone in consequence to lower content of calcium salts. This array of arguments is supported by the pres- ent electron microscopical investigations. The re- sults shown in figs. 2 and 3 give the most outstanding findings. In fig. 2, corresponding to the pseudo- replica prepared from bone previously etched with a 1 °o HNO., solution, the ground substance is made up of small globules or particles apparently spherical in shape. The same globules are often arranged end to end, building up a network of regularly aligned fibrils. Pseudo-replicas obtained from bone di- gested with trypsin (fig. 3o) show a very fine frame- work, the meshes of which delimit spaces correspond- ing in size to globules which have been removed, belonging to the ground substance. Fig. 3/? shows a thin fragment of digested bone removed from the sample in stripping the pseudo-replica. The struc- ture appears as a very fine framework delimiting round spaces. The maximum diameter ranges from 200 to 250 A. According to the electron microscope investiga- tions of Barbour (4) the diameter of the holes per- taining to the inorganic bone fraction ranges from a minimum of 62 A to a maximum of 225 A. The inorganic bone fraction may be regarded as a homogeneous body with holes occupied by the The Study of Bone: Electron Microscopy, X-Ray Diffraction, & Optical Birefringence IZl micellae of the organic substance. The holes corres- pond to ellipsoids of revolution and are aligned with the axis of the bone. The present interpretation ap- pears to be in agreement with the properties of the bone and not in contradiction to the knowledge derived from the wide-angle diflYaction pattern. The latter corresponds to the crystalline units. From their aggregation derives the homogeneous body with holes responsible for the low-angle x-ray dif- fraction pattern (13, pp. 164-165). References 1. AscENZi, A., Science 112, 84 (1950). 2. — Sci. Med. Ilalica 3. 701 (1955). 3. AscENZi, A. and Chiozzotto, A., Expeiiciiiiii II. 140 (1955). 4. Barbour, E. P.. Am. J. Phys. Aniliropol. 8, 315 (1950). Nature, 171,564(1953). 60 (1953). A., Biochim. et Biophys. 5. Caglioti, v.. AsrrNZi, a., and Scrocco, E.vpe'nV/;/;« 10, 371 (1954). 6. Carlstrom, D. and Finean, J. B., Biochim. et Biophvs. Acta, 13, 183 (1954). 7. Dallrmacinf, M. J. and Melon, J., J. Washington Acad. Sci. 36, 181 (1946). 8. Engstrom, a. and Finean, J. B. 9. Finean, J. B.. J. Sci. Instr. 30, 10. Finean, J. B. and Encisirom, Acta 11, 178 (1953). 11. - Experientia 10, 63 (1954). 12. GiiNiiR, A., X-Ray Crystallographic Technology. (Eng- lish trans!, by T. L. Tii'pfi.) F.d. by K. Londsdale. Ililgcr and Walts Ltd. London, 1952. 13. GuiNiER, A. and Fournet, C, Small-Angle Scattering ofX-Rays. J. Wiley & Sons. New York, 1955. 14. North, A. C. T., Cowan, P. M., and Randall, J. T., Nature 174. 1 142 (1954). 15. Robinson, R. A. and Watson, M. L., Ann. New York Acad. Sci. 60, 596 (1955). 16. Schmidt, W. J., Ber. Oherhess. Ges. Natui- u. Heilk., Naturw. Aht., Giesaen, 15, 219 (1933). X PATHOLOGY Elektronenmikroskopische Untersuchungen des experimentellen Lungenodems H. SCHULZ Pathologisches Institiit der Mediziiiischen Akadeniie Diisseldoij iind Lahoratoiiimi fur biologische Feinstruktur-Forsc/iiing der Anatomischen Abteilung des Karolinska Institutes, Stockholm In der Entstehung des Lungenodems ist die Trans- sudation von Flussigkeit aus den Lungenkapillaren von besonderer Bedeutung fiir die Erschwerung des Gasaustausches zwischen Kapillarblut und Alveolar- luft. Die aus den Lungenkapillaren filtrierte Fliissig- keitsmenge wird bestimmt von der Hohe des Kapil- lardrucks. von der Oberflache des Kapillarbettes, vom kapillaren Blutstrom je Zeiteinheit, vom onko- tischen Druck des Plasmas, sowie von der Kapillar- permeabilitiit. AuBerdem gehort zu diesen Faktoren der Lungengewebsdruck, der entsprechend der Lungenstruktur niedrig ist. Zusammenfassende Darstellungen zur Pathoge- nese des Lungenodems geben Altschule ( 1 ) und Hayward (6). Uber die pathologische Anatomic des Lungenodems nach Befunden der Lichtmikroskopie wurde durch Ceelen (3) und v. Hayek (5) be- richtet. Im submikroskopischen Bereich sind bisher keine Befunde iiber das experimentelle Lungenodem mitgeteilt worden. Die Ultrastruktur der normalen menschlichen Lunge und der verschiedener Tierarten wurde inzwischen untersucht. Swigart und Kane (16), Low (9), Policard (10) sowie Schlipkoter (13) befaBten sich vor allem mit der Lungenalveole der Ratte; Kisch (8), Karrer (7) und Bargmann (2) beriicksichtigen vorwiegend die Ultrastruktur der Lungenkapillaren von Vogein, Amphibien und Saugern. Die normale menschliche Lunge konnte in eigenen elektronenoptischen Untersuchungen stu- diert werden (15), sowie von Gieseking (4). Zur Erzeugung eines experimentellen Lungen- odems wurden Ratten desselben Stammes bei norma- lem Luftdruck einem Gasgemisch von 3 "„ Kohlen- saure mit Luft und konzentriertem O., ausgesetzt. Das Lungenodem ist hierbei auf die unmittelbare Einwirkung der Gase auf die atmende Oberflache zuriickzufiihren. In einer zweiten Versuchsgruppe wurden Ratten intraperitoneal mit 30 mg/kg Korper- gewicht Thiosemikarbazid (Tennekoon (17)) bzw. mit 50 mg/kg Korpergewicht alpha-Naphtylthio- harnstoflf (ANTU, Richter (11)) behandelt, um das Lungenodem vom Blutwege her zu bewirken. Die Tiere wurden jeweils 10, 20 und 60 Minuten nach der Injektion getotet. In einer dritten Versuchsgruppe erzeugten wir an Ratten ein mechanisches Stauungs- odem durch operative Abschniirung des Hauptstam- mes der Vena pulmonalis eines Lungenlappens. Die Lungenlappen wurden einmal nach fiinf, ein anderes Mai nach 15 Minuten entfernt. Die Lungenstuckchen der ersten Versuchsgruppe wurden nach Palade, die der beiden anderen nach Sjostrand tixiert Die Schnitte fertigten wir an mit dem Ultramikrotom nach Sjostrand. Fiir die Aufnahmen benutzten wir das RCA EMU 2c Elektronenmikroskop. Unsere elektronenmikroskopischen Untersuchun- gen haben wir auf die Fragen gerichtet, die licht- mikroskopisch nicht geklart werden konnten. Be- sonders beriicksichtigten wir den ProzeB des Proto- plasmaodems, die Veranderungen an der Basal- membran der Lungenkapillaren sowie die Veran- derungen des Kapillarendothels. Die elektronenmikroskopischen Aufnahmen zeigen pathologische Veranderungen an alien Strukturen des Blut-Luftweges. In den Anfangsstadien des Lungenodems besteht ein intraepitheliales Odem (Abb. 1). Die Alveolarepithelfortsatze iiber den Kapillaren, die in Normalfallen bis zu 0,1 /< schmal sein konnen, schwellen bis zu einer Breite von 2,5 /<. ijber den Kapillarvorwolbungen ist dem Fliissig- keitsaustritt der geringste Gewebswiderstand entge- gen gesetzt. Daher kommt es zuerst an diesen Stellen zur Entwicklung eines Lungenodems. Die schmalen Abb. 1. Intraepitheliales Odem mit Schwellung des Alveo- larepithels. Keine Abhebung derZelle von der Kapillarwand. Intakte Membrancn. Elektronenoptisch: 7900 . Abb. 16000 . Untersiichtmgen dcs cxpcrimcntcllcn Liiiii^'cnodcnis 241 OS^ Abb. 2. Beginnende Schwellung und Aufhellung des Alveo- larepithels ^m Bereich des Blut-Luftweges der Rattenlungc. 210-315 A weite schleusenartige Offnungen der basalen Membran der Alveolarzelle (s. Pfeil). Elektronenoptisch: 17000 , Abb.: 49000 x. Epithelfortsatze runden sich mit zunehmender Auf- quellung ab. Die seitlichen Zellgrenzen der Alveo- larepithelien verlieren dadurch den engen Kontakt zu den Nachbarzellen. Die cytoplasmatische Matrix des Epithels im Bereich des Blut-Luftweges ist elelc7;5c7//-. 65, 740-743 (1953). Untersiichimgen des experimentellcn Limi^enodems 243 MOR/VV\L AlvaolQrlumen . AlveolQrepiHne-l ■0.1 Z/>- ' ,,bQSQle-MembrQn d Alv epiVh 80 A -■■■•■■■■■■■■■■ . • ;.^p=s=^fS5^ ■ ■ ■ • •■-■■. ^^^ ■.-.■ ■.:■..■.■■ ■~----.t,asQie Membt-an d tnaokiets 80 A^ ^"~ EndoHneJ ^ 850A - 0,5H^ ■---KQpillQrli^rnen Alveolar iumen __-:nH-Qep:HneiiQliM Odem 0,8S/>( - i,»)/x 0 QoJl 0 - Schleusenarhge Offnungen zio-575/A we<> c( Dosalen Membran d Alvep/ki periendokneJiQler Sh-e^ffen Vokuolen d Endofhele — EndohndblQsen Kopiilorluinen ■ 650A 0,55-0,55/W Z.erre<'5SunQ der ^ur Alveole hingeridni^Jen EpiWieJmembran inh-QepiHncJiales Odem bis -> 2,15 /« -; ]nl-t-Q-QlveolQre5 Odem Auposung des Membron- Syslems der KapillarwQnd mil- Odem des periendol+iel SH-e"* ^'^^i^ Y- 1 \ 1 ^•fl 11 ; I ^ / i / u t- ste Breite deli J E-i* i" / ^ - -^Fln ze/prozenie |- V t / p J -- ^TKT —J L /•N 1' i'H - M r 1 — -J---H — -r~r — ' — J. . ± 'i^rossfe Lsnge c / 1 \i \ ' 'Mill \ / J > V \ , ,! illll r-\-i- Mitochondrien Rattenlunge sih in '~' ,f 1 ko- ' II 1 1 i 1 1 1 1 1 X""" 1 1 1 1 1 1 1 1 1 1 1 "t 4995 99 95 50 10 0.1 02 Oil OiOiio 1.2 rfi.iii/i 0.2 ail OiO.il 0 u m un^ Abb. 4. Wahrscheinlichkeitsnetz. Anzahl der ausgesuchten Mitochondrien bei Rattenlunge normal 288, silikotisch 354. tualen Beteiligungen der einzelnen Gruppen zu ein- ander verhalten. Da wir bei der Durchsicht der elektronenoptischen Bilder den Eindruck hatten, daB die mit Sicherheit als Mitochondrien anzusprechenden Zytoplasmapar- tikel im Lungengewebe der Ratten, die fraktioniert feinkdrnige amorphe Kieselsaure erhalten hatten, im Vergleich zu den normalen Rattenlungen vergroBert waren, haben wir die den Gruppen A, B und C zugehorigen Formen ausgemessen. Hierzu wurde jeweils die groBte Lange und die groBte Breite be- stimmt und die Werte getrennt in zwei Kurven eingetragen. Die Haufigkeitsverteilung der Mito- chondrien in Rattenlungen nach intratrachealer Injektion von amorpher, feinkorniger Kieselsaure zeigt eine deutliche Verschiebung nach rechts, und auch die Eintragung der Summenhiiufigkeitspro- zente im Wahrscheinlichkeitsnetz (Abb. 4) liiBt eindeutig erkennen, daB die Mitochondrien in den Zellen der Kieselsiuiregranulome vergroBert sind, wobei zwei Teilkollektive uberlagert sind. Aus den Beobachtungen, Auszahlungen und Aus- messungen, die noch durch ein groBeres Zahlen- material ergiinzt werden miissen, geht hervor, daB die Veranderungen, die beim Quarzgranulom gefun- den worden sind, in den Lungen der Versuchstiere nach fraktionierter Kieselsaureapplikation in ganz ahnlicher Form auftreten. The Lung Tissue in Mice Infected by Tubercle Bacilli B. Cedergren Department of Bacteriology ami tlie Laboratory for Biological Ultrastnicture Research of the Department of Anatoim\ Karolinslca Institutet, Stockholm Mice of the same sex and age weighing about 20 g have been infected through the intravenous route with a known dosage (0.1 or 1.0 mg tbb) of a bovin strain of tubercle bacilli. After different intervals — 1 to 8 weeks — the lungs have been prepared for light and electron microscopy. The specimen for light microscopy have been stained according to Ziehl-Neelsen. The specimens for electron microscopy have been essentially prepared after a scheme used at the Department of Anatomy. Very soon after infection — one or two weeks — pathological alterations are seen with the light micro- scope. The most obvious ones are circumscribed areas — granuloms — containing a great number of tubercle bacilli and different kinds of tissue cells. The Lung Tissue in Mice Infected by Tubercle Bacilli 249 Most of the material presented is from such an area in an animal killed eight weeks after infection. In the electron microscopic picture most obvious are the great number of cells containing lots of black homogeneous, rounded granules about 0.5 to 1 // in diameter. Some of them have been recorded with surrounding concentrically layered membranes, each with a thickness of about 100 A. In the neighbourhood of these cells are others filled with vacuolar structures of about the same size. In most of the cells containing these black granules a rather high number of tubercle bacilli and often some other characteristic structures described below, have been found. They are supposed to be stages in the degeneration of cells, in this experiment caused by the tubercle bacilli. One type of these granules (A) is rounded with a fine-granulated ground substance, rather poor in contrast and of varying size about 0.2 0.5 //. Within these granules are sometimes seen rounded clusters of strong osmiophilic granules (B), 50-1 00 A in size. A more complicated type of granules (C) about 0.5 to 1 // in width and often of a charac- teristic shape is also observed. This structure seems to consist of a great number of peripheral concentri- cally layered membranes, approximately 100 A in width and a medulla consisting of small osmiophilic granules, the most peripheral of which seem to be Fig. 2. Tubercle bacilli in a cell containing black granules within vacuoles. The picture is supposed to demonstrate a later stage of cell destruction. Magnification 20.000. Fig. 1. Clusters of small B-granules in an A-granule. The very characteristic C-granule with its many concentric membranes is also seen. Two tubercle bacilli (tbb) are seen. The picture is supposed to demonstrate an early stage of cell destruction. Magnification 46,000. arranged in parallel rows. The C-granules are some- times more or less covered by a black osmiophilic homogeneous substance. The different types of granules just described have been recorded in the same section (fig. 1) as well as some others which according to their ultrastructure (number of concentric membranes and granulated centers etc.) very possibly could be intermediate stages. The different granules and vacuoles are supposed to be stages in the degeneration of cells, where "A-, B-, and C-granules" are found in the early stages. The black granules could be intermediate products, which later on are dissolved leaving vacu- oles in the eel I. (The eel Is containing the black granules are In this case supposed to be macrophages.) The degeneration is believed not to be absolutely specific for the tuberculous inflammatory reaction. The reason for this conclusion is that a very few opaque granules of about the same size can be seen in normal lung tissue cells and \o a cerlaiii extent have also been observed in lung tissue from pneumo- coccus-infccted mice. This experiment, howe\er. has shown a distinct increase in the number of the characteristic black granules. My interpretation is, that normally cells are to a certain extent degenerating and dying. In the infected tissue, however, where the infecting agent is localized in the cell proper, this degeneration is proceeding rapidly and intensively thus giving very obvious alterations. The Importance of an Accurate Size Determination of Fine Particles when Investigating Their Biological Effects G. Bloom, J. Glomme and A. Swensson Department of Occupational Medicine, Karolinska sjiikhiiset; King Gitstaf V Research Institute and the Department of Histology, Karolinska Institiitet. Stockholm When studying the biological reactions to different particles it has become evident that the size of the latter is of importance. With the aid of the micro- scope it is possible to obtain a particle size distribu- tion of a given sample, but this method is naturally limited by the resolving power of the microscope used. As the biologically most active particles often are of such a size that a closer examination of them is not possible in the ordinary light microscope, indirect methods have been used to determine their 100% 80 60 P:1 TOXICITY 1,1 1 0,06 EXCRETION 6,55% 40 A = 0,I5/J 20 0 4^- 02 0.4 0.6 0.8 1.0 1.2^ 00% P:2 80 TOXICITY 1,8 i 0.17 60 EXCRETION — 4U A=0,40ju 20 ^ 0 t'JMV^^^ I00%| 80 60 40 20 0 Q2 0.4 Q6 0,8 1,0 1.2 ^ P--3 TOXICITY 3.4*0.19 EXCRETION 2.6 % A =0.75^ 0.2 04 06 0.8 1.0 1.2/J sizes. These methods are all suffering from con- siderable errors. A great need for new possibilities of examining very fine particles has arisen. In modern dust combat- ing rather effective methods have been worked out for eliminating particles of light microscopical size, but these methods are less effective with regard to submicroscopical particles. Thus recirculation of air in dustladen workrooms may cause a concentration in the air of submicroscopical particles which pass 100% 80 60 P:61 TOXICITY 0,2 i 0.01 EXCRETION 54% 40 A = 0.015^ 20 0 100% 80 ■ 60 • 40 ■ 20 0 100% 80 60 40 20 0 OZ 0.4 Q6 0.8 1,0 1.2 jU P:5 5 TOXICITY 0,5^0,02 EXCRETION 23% A=0,IOyU 02 0.4 Q6 0,6 1.0 l,2yU P--71 TOXICITY 2,1 ±0,06 EXCRETION 21% A = O.I5jU 02 Q4 0,6 0,8 1,0 12 /J Fig. 1 (left). P:l, P;2 and P:3 — diflerent size fractions of the same sample of 99 "o a-quartz. Toxicity: Acute toxicity in mg per 30 g mouse by fractioned intravenous injection ad modum Dale and King (1953). Excretion: Urinary excretion after intraperitoneal injection during first five days in per cent of injected amount. A: Average particle size. Histograms: Particle size distribution. Electron micrographs: The different particle samples, see text. Fig. 2 (right). P:61, Aerosil, Gold- und Silberwerke. P:55, amorphous silica prepared by combustion of SiClj rtf/ /?;o(/«/» Flemmert. P:71, ground, fused silica. Other items as in fig. 1. Stauhkorngrdfien in Staiihlungen 251 through the dust fiher. These fine particles do not easily lend themselves to ordinary industrial hygienic control measures. Little is known of their biological effects, and it is difficult to discuss their practical importance, adequate protection against them, etc. The electron microscope has made possible closer examinations of these extremely small particles both with regard to general appearance and to si/e. Several methods have been proposed for preparing samples for electron microscopical examination. Special devices have been made and standardized for collecting particle samples from air directly or indirectly via millipore filters onto specimen grids, and in such a way as to make the sample representa- tive of what is contained in the atmosphere. Such investigations of the atmosphere at workplaces where a risk for silicosis exists have shown that the percentage of particles in the submicroscopical range is very high. It has also been shown that dust-com- bating methods, such as wet drilling in mines, chiefly influence the somewhat larger particles. In our experimental investigations we have been interested chiefly in determining the particle size distribution of particle samples used for biological investigations. For the preparation of samples we have applied different methods but chiefly we have dispersed the particles in water with the aid of a colloidal mill. The same method has also been used for preparing the particle suspensions used for the biological experiments. The particle size distribution was determined by measurements performed on electron micrographs As the irregular particles of ground silica have no definite geometric diameter we have consequently measured in a horizontal direction the distance between two extreme points on the particle. It has been known for some time that the bio- logical reaction depends on the particle size, but mostly only gross differences of size have been registered. In the following some examples will be given of the relationship between particle size and biological reaction of particles of silica of different kinds. The importance of the particle size is seen in fig. 1 , which shows three fractions from the same ground sample, of almost pure (99 "„) a-quartz. There is a close relationship between particle size and toxicity in such a way that the toxicity decreases with in- creasing particle size. The excretion values indicate a similar relationship with an increased excretion with decreasing particle si/e. in tig. 2 corresponding investigations on samples of amorphous silica prepared in different ways have been grouped together. The relationship be- tween particle size on the one hand and toxicity and excretion on the other is evident also here. It is, however, obvious that also factors other than particle size must he of importance, which is natural with regard to the differences in mode of preparation and structure of the various samples. Regarding the fibrogenetic efTect, it is well estab- lished that this effect is marked in all the different crystalline samples. As for the fine amorphous samples of silica the situation is somewhat more complicated. This is exemplified by the reactions to two different samples prepared in the same way but having different particle sizes (P:5I, average particle size 0.01 microns, toxicity 0.2 r 0.01, and P:55, cf. fig. 2). Sample P:55 one month after intraperitoneal injection in the rat gives a marked cellular reaction with moderate fibrosis, which after 7 months is poorer in cells and slightly more fibrotic than after I month. With sample P:51, the reaction after I month is chiefly a relatively mild cellular reaction with some reticulin reaction. After 7 months the cellular reaction is almost gone and reticulation is minimal. A definite regress has taken place. Macro- scopically it is not possible with this sample to see any changes at all in the peritoneum after 7 months. According to these investigations the biological effect of silica particles seems to be highly dependent on the particle size. Detailed knowledge about the particle size distri- bution is a necessary basis for a scientific discussion of these problems, and more important than hitherto supposed. Reports on results obtained will be published in Aich. Ind. Health, Acta physiol. Scaiul.. and Acta phaniuicol. et toxical. Elektronenoptische Untersuchungen von StaubkorngroBen in Staiibkingen H. W. ScHLiPKOTER und A. Colli Imtitut fiir Hygiene unci Milirobiologie der Medizinischen Akademie, Diisseldoif Aus dem groBen Problemenkreis der Staublungener- krankungen, vor den sich die medizinische wie auch die physikalische und die chemische Forschung ge- stellt sieht, ist die Frage nach der Staubkorngrof3e von wesentlicher Bedeutung. AusmaB und Starke der Gewebsreaktion sind weitgehend von ihr ab- hiingig. Besonders fur Abv\ehrmaf3nahmen, vor allem fiir die Staubbekampfung. ist es wichtig zu wissen, wie groB die Teilchen sind, die die Staub- schiidigungen in den Lungen hervorrufen. 252 H. E. SCHLIPKOTER UND A. COLLI Die Bestimmung der StaubkorngroBe ist daher seit langererZeit Gegenstand vieler Untersuchungen. Bisher konnten aber keine einheitlichen Ergebnisse, vor allem iiber die untere Grenze der gefahrlichen Kornfraktionen erzielt werden. Dabei bediente man sich der verschiedensten Methoden, auf die hier nicht naher eingegangen werden kann. Es sei nur hingewiesen auf die theoretischen Arbeiten von Findeisen und Davies und auf die Retentionsmes- sungen von Brown. Ney. Hatch und Cook, wie auch von Worth und Schiller und auf die chemischen AufschluBmethoden staubhaltiger Lungen von King, Pfefferkorn. Gessner, Riittner und Biihler. Thomas und Stegemann. Die hierbei gefundenen Korngioficn- wertc zwischen 0,7 und 5 // wurden Hchtoptisch er- mittelt. Dabei bUeb aber immer die Frage offen, inwieweit auf diese Weise die feineren Kornfrak- tionen erfaBt werden konnen. Mit Hilfe des Elek- tronenmikroskops wurde es nun moghch, auch feinste Partikel sichtbar zu machen. Allerdings ent- standen dabei methodische Schwierigkeiten. PoUcard und Mitarb. versuchten dieses Problem zu losen, indem sie silikotisches Lungengewebe mit Mikro- nadeln zerzupften. Zebel untersuchte Lungenstaub, der nach der Formamid-Methode von Thomas und Stegemann eliminiert worden war. Dabei zeigte sich, daB die Hauptmenge der Staubteilchen zwischen 0,1 und 1 /< lag. In eigenen Untersuchungen wurden nun, um even- tuelle, durch chemische Einfliisse hervorgerufene Veriinderungen des Lungenstaubes auszuschlieBen, von silikotischen Sektionslungen Diinnschnitte her- gestellt. Das Lungengewebe wurde in Methacrylat eingebettet. Auf eine OsO^-Fixierung muBte verzichtet Kurve 1. Haufigkeitsverteilung der KorngroBen von Staub- teilchen in einer Sandstrahlbliiserlunge. Obere Kurve Breite und untere Kurve Lange der Teilchen. werden, da es sich nicht um frisches Gewebe han- delte. Morphologische Strukturen des Lungenge- webes konnten daher nicht sichtbar gemacht werden. Dieser Mangel brachte aber den Vorteil mit sich, daB man den Staub klar und ohne storende Ge- websteile sehen konnte. Es wurden fiinf Staublungen untersucht und zwar handelte es sich dabei um die Lungen von einem Sandstrahlbliiser, einem Steinmiiller und drei Koh- lenhauern. Bei der Sandstrahlblaserlunge wurden 1025 Teil- chen ausgemessen. Das elektronenoptische Bild liiBt nadelformige Teilchen erkennen (Abb. I). Die groBte gemessene Liinge betrug 1,3! //, die kleinste 0,0125 //. Die meisten Teilchen wurden bei einer Lange von 0,15 // im Mittel gefunden (Maximum). Zwischen 0.6 und 1,31 // waren nur noch ganz vereinzelt Teilchen zu beobachten, die deswegen nicht in der Kurve eingezeichnet sind (Kurve 1). 250 K 1 \ If \ 1 \ Kfl 1 \ IM 1 \ 1 \ -^ V 50 \ , ^ r\\ 1 1 0. V 2 0 ^ H 0 i OS It 7 /. ? /, / /.fc Kurve 3. Haufigkeitsverteilung der KorngroBcn von Staiib- tcilchcn in einer Kohlenhauerlunge (Nr. I). Oberc Kurve Brcitc und untcre Kurve Liinge der Teilclien. Im Gegensatz zu der verhaltnismiissig lang ausge- zogenen Kurve fur die Teilchenliinge ist das Kurven- bild fiir die Breite der Teilchen ein ganz anderes. Diese Kurve steigt sehr schnell und steil bis zum Maximum, das hier bei 0,045 // liegt, an, um dann bis auf 45 Teilchen bei 0,1 // ebenso steil wieder abzufallen. Bei der Steinmullerlunge zeigt die elek- tronenoptische Aufnahme im Gegensatz zu der Sandstrahlbliiserlunge, daB die Form der Partikel quadratisch ist. Die Kurven, die sich bei der Auszahlung von 993 Teilchen in der Steinmiillerlunge fur die Lange und Breite ergaben, ahneln einander sehr. Sie weisen beide in ihrem Verlauf einen steilen Anstieg auf und erreichen beide das Maximum bei 0,15 //. Die Werte fallen dann bis zu zwei bzw. 4 Teilchen bei 1,35 // ab. In der GroBenordnung zwischen 1,35 // und 2,45 // fanden sich nur wenige Teilchen (Kurve 2). ZOO too \ 1 A i N \ \ \ \ '-^ V X ^ , , O.'t 0% 1.2 /,6 2.0 21 2.8 Kurve 4. Haufigkeitsvcrteilungskurvcn der KorngroRen von Staubteilehen in einer Kohlenhauerlunge (Nr. II). Spii/ige Kurve Breite und stumpte Kurve Liinge der Teilchen. Kurve 5. Hiiufigkeitsverteilungskursen der KorngroRen von Staubteilehen in einer Kohlenhauerlunge (Nr. III). Die Kurve mit dem hochsten Maximum zeigt die Breite und die andere Kurve die Liinge der Teilchen. Bei der ersten Kohlenhauerlunge wurden 1012 Teilchen gemessen (Kurve 3). Die Hiiufigkeitsvertei- lungskurve fiir die Breite der Teilchen zeigt auch hier wieder das charaktcristische Bild, wic cs schon von vorher gezeigten Kurven bekannt ist: steiler Anstieg und fast ebenso steiler Abfall der Kurve. Das Maximum tindet sich bei 0,075 //. Das Kurvcn- bild fiir die Teilchenlange zeigt dagegen einen ver- zogerten Anstieg und einen noch langsameren Abfall. Der Grund dafiir ist wohl in der besonders unein- heitlichen Liinge der Partikel zu suchcn. Das liingste Teilchen, das hierbei gemessen wurde, war 5,52 // groB. Das Maximum liegt bei 0,23 /< (Kurve 3). Die zweite Kohlenhauerlunge weist ein ganz ahn- liches Kurvenbild wie Nr. I auf (Kurve 4). Die Anzahl der ausgeziihlten Teilchen bctrug 7(S7. Fiir die Breite liegt die Hauptmenge der Partikel bei 0,06 //, wiihrend das Maximum fiir die Liinge bei 0,2 fi liegt. Einc leichtc Vcrschicbung nach rechts. d. h. zu hoheren TeilchengroBe hin, zeigt die Kurve der dritten Kohlenhauerlunge. Das Maximum fiir die Breite der Teilchen liegt bei 0,25 /< und das der Liinge bei 0,35 // (Kurve 5). Alierdings war die Anzahl der gemessenen Partikel wcscntlich geringer als bei den beiden anderen Kohlenhauorlungen. Sie bctrug nur 375 Teilchen. Das groBte hier beobachtete Teilchen hatte eine Liinge von 4,7 //. Zur idcntifizicrung dcs Staubes wurde die MothtHie der Llektronenbcugung angcuandt. Da cs sich bei den vorliegenden Priiparaten nicht um isolierte Stiiube, sondern um staubhaltiges Gewebe handelt, konnten nur von einzelnen Partikcln mit Hilfe der Fcinbereichsbeugung Beugungsbildcr cr/ielt werdcn. Nach den ausgemessencn Nct/ebencnabstandcn diirfte es sich bei den in der Sandstrahlbliiserlunge gefundenen Teilchen vorwiegcnd um das Silikat 254 F. MILLER AND A. BOHLE Hektorit oder um das ebenfalls zur Montmorillonit- reihe gehorende Nontronit handeln. Die bei der Steinmiillerlunge gemessenen Netz- ebenenabstande entsprechen dem Bild des Montmo- rillonits. Das Beugungsbild, das bei der Kohlenhauerlunge erhalten wurde, Iief3 sich nur schwer einordnen. In der Mehrzahl stimmen die Netzebenenabstande und die Intensitiit der einzelnen Beugungsringe mit denen des Quarzes Uberein. Bei einem Vergleich der fianf in dieser Arbeit auf- gezeigten Haufigkeitsverteiiungskurven ergibt sich, daB die Maxima fUr die Lange der Teilchen zwischen 0,15 // und 0,35 /< bzw. fiir die Breite zwischen 0,045 und 0,15 // liegen. Das bedeutet, daB ein erheblicher GroBenunterschied zwischen friiheren lichtoptischen und den vorliegenden elektronenmikroskopischen StaubkorngroBenbestimmungen besteht. Die Frage, wie es zu diesen unterschiedlichen Ergebnissen kommt, laBt sich leicht an Hand eines hoher ver- groBerten Staublungenschnittbildes klaren. Es zeigt sich hier namlich, daB Partikel, die bei geringerer VergroBerung als wenige groBe Teilchen erscheinen miissen, in Wirklichkeit viele zusammengelagerte feinste Teilchen sind. Die Konsequenzen, die sich aus diesen Ergebnissen fiir die Staubbekampfung und damit fiir die Verhii- tung der Silikose ergeben, sind leicht einzusehen. Bei den praktischen Staubmessungen und den bishe- rigen AbwehrmaBnahmen gegen die schiidigenden Lungenstaube wurden namlich vorwiegend Partikel hoherer GroBenordnungen beriicksichtigt. Nach die- sen und anderen elektronenoptischen Untersuchun- gen diirfte aber nur eine solche Prophylaxe Aus- sicht auf Erfolg haben, die es versteht, auch — und man darf wohl sagen: vor allem — die Feinststaube unter 0,5 n aus der Lunge fernzuhalten. Electron Microscopy of the Glomerular Basement Membrane in Experimental Amyloidosis of the Mouse F. Miller and A. Bohle Department of Pathology and Laboratory of Electron Microscopy, University of Innsbruck, Austria, and Department of Pathology, University of Heidelberg. Germany Recent investigations of the renal glomerulus (13, 14, 18, 21) with improved technique have placed on a firmer basis our concepts of this complicated structure. The continuous basement membrane of the capillary tuft is covered on the inside by the attenuated and porous sheet of the endothelium and on the outside by the interdigitating foot processes of the visceral epithelial cells. The basement mem- brane is built up of three layers. A dense osmio- philic middle layer (lamina densa, Yamada (21)) is lined on either side by a less osmiophilic inner and outer layer (lamina rara interna and externa, Yamada (21); inner and outer cement layer. Pease (13)). The middle layer in the mouse glomerulus is about 600 A thick (18, 21). The inner and outer layers are about 300 A thick (18). Hall (4) apparently called the entire basement membrane lamina densa and did not further comment upon the less osmiophilic layers. Policard et al. (16) conceived of the outer layer as an intermediate space. Hall (3) described pores in the basement membrane after fixation in buffered formalin and formalin-alcolhol mixtures but could not find them after osmium fixation. Rhodin (18) found a lamellated, spongy structure of the osmiophilic middle layer, and Yamada (21) observed a dense feltwork of fine filaments about 30 A thick in the lamina densa. Piel ct al. (15) noted a thickening of the basement membrane in the early stage of the Masugi nephritis of the rat. The present work gives a preliminary report on the basement membrane of the mouse glomerulus in experimental amyloidosis. Amyloidosis was produced by the method of Letterer (7) as modified by Latvalahti (6). White mice were given 20 injections of 0.5cc Natrium-Casein (Merck) suspended in n/10 NaOH (pH 10.0) subcutaneously together with 0.5 lU ACTH (Hoechst) over a period of 4 weeks. The kidneys were exposed under light ether anaesthesia and pieces of the cortex of about I mm^ were fixed for 4 hours in I "o osmium tetroxide buflfered at pH 7.2 in the manner of Palade (10). The tissue was embedded in butyl- methylmethacrylate (95:5) and polymerized at 47^ using ] ",, dichlorobenzoyl peroxide as a catalyst. A large area of the tissue was sectioned very superficially on the Sitte (19) ultramicrotome and controlled in a phase contrast microscope. When a glomerulus was cut the tissue block was trimmed to an area of about 0.5 : 0.5 mm under a reflected light microscope in such a way that the glomer- ulus was located in the center of the cut surface. Thin sectioning was then done with glass knives. The sections were mounted on Athene specimen grids by a method developed by H. Sitte (20) under a phase contrast micro- scope. Using this method it was possible to move the glomerulus into one of the central meshes of the specimen grid. This was necessary because of the small mobility of the stage of the microscope used. Furthermore, the entire glomerulus could be observed without being par- tially covered by the bars of the grid. Sections were studied with a Siemens microscope (UM 29) without sublimation of the methacrylate at a low beam intensity. The objective aperture was approx. 50 //. Paraffin-embedded sections were stained for light microscopy with various methods suitable for the detection of amyloid. The Glomerular Basement Membrane in Amyloidosis of the Mouse 255 Fig. 1. Low power view of several glomerular capillaries in experimental amyloidosis. Arrows indicate nodular protru- sions of the osmiophilic middle layer of the basement mem- brane. i\ capillary lumen; b, cavity of Bowman's capsule. Magnification 5000. Light microscopy showed an amyloidosis of almost all glomcruh. With the periodic acid-silver reaction the capillary wall stained black. The amyloid seemed to be deposited within the space enclosed by the capillary wall. Electron microscopy revealed hitherto unknown changes of the basement membrane, in particular of the osmiophilic middle layer. Measurements of the basement membrane in normal mice confirm the findings of Rhodin (18) and Yamada (21). The total thickness of the basement membrane is 1200 A ± 117. The osmiophilic middle layer is 632 A ± 73 thick. The less osmiophilic inner and outer layers are 273 A ±31 thick. In mice with amyloidosis the basement membrane is thickened either continuously or in wavy form (fig. 1). The total thickness is 2200 A ± 116. This increase in thickness is exclusively due to a broadening of the osmiophilic middle layer measuring 1557 A ± 107. The inner and outer less osmiophilic layers are of the same thickness as in normal animals. The difference of the means of the osmiophilic middle layer in sick and normal mice is significant on the 99 % level (Student's r-test). The variance of the osmiophilic middle layer in mice with amyloidosis is so much greater than in normal animals (95 "o level; Fisher's F-test) that it cannot be explained as caused by chance alone.' This seems to indicate that the increase in thickness of the middle layer in mice with amyloidosis is caused by a patiiological condition and therefore is varying within relatively broad limits. Apart from the general increase in thickness local protrusions of the middle laser in the shape of knots or mushrooms were observed (fig. 2). These nodular, sometimes hernia-like protrusions have a height of 0.5-1.5 // and a width of 0.6 2.0 /<. They were detected later also with the light microscope on sections stained with the periodic acid silver reaction (9). All the protrusions arc bulging exclu- sively towards the cavity of Bowman's capsule. Fhis could be conceived of as a morphological expression of the decrease in pressure towards the cavity of Bowman's capsule and of the direction of the filtra- tion gradient. Sometimes several protrusions were observed on the same capillary loop spaced closely together. The inner and outer less osmiophilic layers line the bulges of the middle layer and are of the same thickness as in normal mice. The foot processes are always in contact with the outer layer lining the ^ For greatly appreciated help with the statistical evalua- tion of the material we are much obliged to Dr. E. Olbrich, Department of Histology and Embryology, Univ. of Inns- bruck. Fig. 2. Glomerular capillary wall in experimental amyloid- osis. C, capillary lumen \\ith fixed plasma protein particles; P, protrusion of the osmiophilic middle layer in the shape of a mushroom; O, oblique section through capillary wall. Magnification 26,000. 256 A. BERGSTRAND AND H. BUCHT protrusions. Thus, no disruption between basement membrane and foot processes has taken place even in the vicinity of large protrusions of the middle layer, and the outer layer at least may indeed have a cementing function as suggested by Pease (13). Tangential or oblique sections through the capillary wall can be distinguished easily from the protrusions by the aspect of the endothelium and of the foot processes (fig. 2). Within the protrusions of the middle layer a finely spongy or felt-like structure with interwoven filaments of about 30-40 A is observed. Pores, however, were not found. A disrup- tion between endothelium and inner layer of the basement membrane by deposition of amyloid was not yet detected. Sometimes the endothelium seemed to be swollen. The parietal basement membrane measuring 1200- 1400 A in normal mice is thickened in animals with amyloidosis. The mean width is 7400 A. The mem- brane seems to be split into fine filaments but may also appear homogeneous. The question arises whether the thickening and the local protrusions of the osmiophilic middle layer are due to the deposition of amyloid. Recent in- vestigations (1, 2, 5) agree that amyloid is deposited between endothelium and basement membrane. On account of a comparative study with the light and electron microscope (9) it is felt that the thickening of the osmiophilic middle layer is not (or not exclu- sively) due to an infiltration with amyloid. Since the animals had a severe proteinuria it is probable that the passage of pathologic proteins through the base- ment membrane has resulted in a swelling of the osmiophilic middle layer. This view is confirmed by investigations of Mellors and Ortega (8). These authors found by use of a microfluorescence method that in human secondary amyloidosis globulins were localized in the thickened glomerular capillary walls before amyloid deposition was detected micro- scopically. Randerath (17) also observed a swelling of the glomerular capillary walls in human glomer- ulonephrosis probably consecutive to a passage of proteins. The observations reported in this study seem to have some bearing, however, on the problem of structure and function of the glomerular basement membrane. The basement membrane in mice with amyloidosis has the same triple-layered structure as in normal mice. The thickening of the osmiophilic middle layer seems to indicate that the definite ultra- filter is formed by this layer alone. Pores were not observed in the lamina densa although it could be expected that they would become visible in the swollen or bulging middle layer of mice with amyloidosis rather than in the unaltered membrane of normal animals. That pores do not exist cannot be excluded with absolute certainty on the strength of present information. The observations of Hall (3), however, do not form a morphological basis for the thesis of Pappenheimer (II, 12). References 1. BoHLE, A. and Krecke, H.-J., Vlrchow's Arch. 327, 663 (1955). 2. Churg, J. and Grishman, E., Adi. J. Pathol. 29, 199 (1953). 3. Hall, B. V., Proc. V. Ann. Conf. Neplirotic Syndrome, New York. The National Nephrosis Foundation, Inc., 1954, 1. 4. — Proc. VI. Ann. Conf. Nephrotic Syndrome, New York. The National Nephrosis Foundation, Inc., 1955, 1. 5. Jones, D. B.. Am. J. Pathol. 27, 991 (1951). 6. Latvalahtl J., Experimental Studies on the Influence of Certain Hormones on the Development of Amy- loidosis. Thesis. Helsinki, 1953. 7. Letterer, E., Beitr. pathol. Aiiat. 75, 486 (1926). 8. Mellors, R. C. and Ortega , L. G., Am. J. Pathol. 32, 455 (1956). 9. Miller, F. and Bohle, A., A7///. Wochschr. 34, 1204 (1956). 10. Palade, G. E., J. Exptl. Med. 95, 285 (1952). 11. Pappenheimer, J. R., Physiol. Reviews 33. 387 (1953). 12. — Klin. Wochschr. 32, 362 (1955). 13. Pease, D. C, Anat. Rec. 121, 701 (1955). 14. — /. Histochem. Cytochem. 3, 295 (1955). 15. PiEL, C. F., Dong, L., Modern, F. W. S., Goodman, J. R., and Moore, R., /. Exptl. Med. 102, 573 (1955). 16. PoLicARD, A., Collet, A., and Giltaire-Ralyte, L., Arch. anat. microscop. 44, 1 (1955). 17. Randerath, E., Klin. Wochschr. 20, 281, 305 (1941). 18. Rhodin, J., Exptl. Cell Research 8, 572 (1955). 19. SiTTE, H., Mikroskopie (Wien) 10, 365 (1956). 20. — personal communication (1956). 21. Yamada, E., J. Biophys. Biochem. Cytol. 1, 551 (1955). Electron Microscope Investigation on Biopsy Material from Patients with Renal Diseases: A Case of Subacute Glomerulonephritis A. Bfrgstrand an the routine methods of this laboratory. The other half was fixed in 1 per cent osmium tetroxide solution, dehydrated and embedded in methacrylates according to the method described by Palade (4), Newman, Borysko and Swerdlow (3) and Rhodin (5). The blocks were sectioned on a Sjostrand ultramicrotomeand studied in a RCA electron microscope model 2d. Results. — In the light microscope no changes could be observed in the renal glomeruli. This does not exclude the presence of changes in the glomeruli in other parts of the kidneys or that there are changes, which are so slight that they are not observable with this technique. In our opinion the clinical signs are so significant, however, that the diagnosis may anyhow be regarded as very probable. The electron microscope investigation showed the presence of red blood cells in the space between the capillaries, which is most probably the source of the haematuria. Furthermore, changes were observed both in the endothelial and epithelial cells of the capillary walls. Fig. 1. Part of a capillary lumen in a glomcruUis with "spherical bodies" in the lumen. Biopsy material from patient suffering from mild subacute glomerulonephritis. Magnification x 19,000. 1 7 — 568204 Electron Microscopy Fig. 2. Part of endothelial cell in a gioincrulu^. Same case as fig. 1. Large vesicle bordered by a double membrane at A. Magnification < 30,000. The endothelial cells: Fig. 1 shows part of a capil- lary lumen in a glomerulus. In the lumen a number of "spherical bodies"" are observed. They have a diameter of about 1 micron and are bordered by a single membrane. Inside the bodies small rounded or elongated organelles may be discerned. They are bordered by a very delicate membrane and may be regarded as very small vesicles. Similar organelles may be observed in the cytoplasm of the endothelial cells and also in many other kinds of cells outside the renal glomeruli. The "spherical bodies"" cannot be regarded as pathological phenomena. Similar bodies have been observed by the present authors in normal animals (rats, rabbits and dogs). The number and size of the "spherical bodies"" is marked 1\ in- creased in the glomeruli of this patient as compared to what has been observed in animals, however. In the laboratory animals the small organelles inside the bodies are very few or entirely lacking, whereas they are numerous and very distinct in our patient. Thus it is possible that the process of formation of "spherical bodies"" is increased in the diseased kid- ney. The function and the exact nature of these "spher- ical bodies"" is not known. In our opinion they are formed inside the endothelial cells and are parts of the cell protoplasm, v\hich are ejected into the blood stream. Some observations both from this patient and from animals support this assumption. Fig. 2 shows a part of an endothelial cell in the glomeruli of our patient. Inside the c> toplasm a very large vesicle (A) is observed. It is bordered by a double membrane. There are no cell organelles in- side it. Fig. 3 shows a similar vesicle close to the nucleus of the endothelial cell. Here the small cell organelles 258 A. BERGSTRAND AND H. BUCHT Fig. 3. Large vesicle in glomerular endothelial cell located close to the cell nucleus. Magnification ■ 20,000. are very distinct and the vesicle closely resembles a "spherical body". In both pictures "spherical bodies" are observed in the lumen. In a previous paper (1) a large polypous protru- sion from the surface of the endothelium in the efferent artery of a rat was described. There were no cell organelles inside it but the wall consisted of three distinct membranes. We believe this is a "spherical body" protruding from the endothelial cell. When the cell membrane bursts, the "spherical body'" with its single membrane (the inner of the two membranes, bordering it in the cell) is ejected. The epithelial cells: The nuclei of the epithelial cells were well preserved. There was no increase in the number or size of the epithelial cells and there were no adhesions between the epithelial cells of the capillary walls and those of Bowman's capsule as commonly seen in chronic glomerulonephritis. In the cytoplasm of the epithelial cells changes were observed, which may be regarded as degenerative. A large number of vacuoles could be seen in most cells. Many cells were entirely vacuolised, in others only a few, irregular, were visible (Fig. 4). This can imply an increased uptake of fluid from the glomerular filtrate. The mitochondria were markedly enlarged. The inner structure was irregular or completely destroyed Fig. 4. Left. Mitochondrion in epithelial cell of a glomerulus with irregular abnormal inner structure. Magnification X 50,000. Right. Part of epithelial cell of a glomerulus with a large vacuole. Magnification 12,000. as described by Rhodin (5) in the tubular cells during resorption of proteins. It is probable that the glomer- ular filtrate in this patient contains an increased amount of protein, since there is proteinuria. Thus a resorption of protein from this fluid to the capillary epithelial cells may take place. The capillary basement membrane: It has been pointed out in a previous paper (1) that the finest structures of this membrane cannot be demonstrated with this technique since they have very little affinity to osmium tetroxide. No certain changes of the basement membrane have been demonstrated in this material. It is possible, however, that it is thic- kened, and that there are structural changes in it, responsible for the increased permeability. References 1. BERGSTRAND, A., Laboratory Invest. (1957, in press). 2. Kark, R. M. and Muerhcke, R. C: Lancet 1, 1047. (1954). 3. Newman, S. B., Borysko, E., and Swerdlow, M., /. Res. Natl. Bur. Stand. 43, 183 (1949). 4. Palade, G. E., J. E.xptl. Med. 95, 285 (1952). 5. Rhodin, J., Correlation of Ultrastructural Organization and Function in Normal and Experimentally Changed Proximal Convoluted Tubule Cells of the Mouse Kidney. Stockholm, 1954. XI MICROBIOLOGY Some Observations on the Structure of Tobacco Mosaic Virus H. E. Huxley Medical Research Council, Department of Biophysics, University College, London X-RAY diffraction studies of oriented preparations of tobacco mosaic virus have made it possible to elucidate the internal structure of the molecule with a very high degree of detail (1, 2, 3, 6). Electron microscope observations, on the other hand, have, until recently, revealed only that the molecules are rod shaped, about 150 A in diameter and 3000 A long, and have shown nothing of any internal struc- ture in these molecules. During the initial period of operation of the Siemens Elmiskop I at University College, the question was considered of whether the very high resolution now available with that instru- ment would make it immediately possible to see structural regularities within the TMV molecules, and indeed, within other large biological molecules and filaments which previously had appeared struc- tureless. A very dilute solution of tobacco mosaic virus was placed on carbon-filmed electron-microscope grids, and stained with 40 % phosphotungstic acid in the manner described by Hall (4). The results at first seemed disappointing, for even when the resolu- tion was better than 10 A, no regular internal struc- ture was visible in the particles. However, a curious effect came to light which did reveal one feature of the internal structure. In some areas of the grids it was sometimes appa- rent that the excess stain had not been completely removed by washing before the preparation was allowed to dry. In these areas, the particles became outlined by stain in a very distinctive manner. A similar effect has previously been noted by Hall (4) with bushy stunt virus. In addition to the outline, however, many of the particles now showed a dark line running along their long axes (fig. I). In the longer particles, this line was frequently much better defined at the end of the particle than near the centre (fig. I). The line was indistinct or invisible in particles which had not been outlined by the stain. It was found that this outlining effect could be produced not only by phosphotungstic acid, but also by 0.1 Af potassium chloride, or 0.1 M bicar- bonate buffer; it seemed to be sufficient merely to have a little salt of some sort present when the prepa- ration was allowed to dry. The extent of the effect was found to vary very considerably from area to area on the grids, and from one batch of grids to the next, and is presumably related to the wetting properties of the supporting film. Preparations of virus degraded into short lengths in bicarbonate buffer at pH 10.3 (5) also showed particles with a dark line down their centre when treated by the above techniques. Very short rods, standing on end, are often observed in such prepara- tions; these rods appear to be accurately cylindrical in section, and the central core is readily seen (figs. 2 and 3). Preparations of virus which have been very thor- oughly washed (fig. 4) do not appear "outlined" in this way and it seems that both the PTA or the salt is easily removable. The most likely explanation of this effect is in terms of the structure, proposed by Franklin on the basis of x-ray studies, in which the virus rods have a hollow core. Thus, if a TMV particle is allowed to dry in a pool of dilute salt solution, the salt will tend Fig. 1. Tobacco mosaic virus, outlined by drying in very dilute solution of potassium chloride (see text). Magnification 140,000. Fig. 2. TMV degraded in alkaline solution and outlined in phosphotungstic acid. Magnification 140,000. The Periodicity in Tobacco Mosaic Virus 261 Fig. 3. TMV degraded in alkaline solution and outlined by same solution. Magnification 140,000. Fig. 4. TMV stained with 40 ',',', phosphotungstic acid, and thoroughly washed before allowed to dry. Magnification X 140.000. to be deposited around the surface of the particle, and if the particle has an inner surface which is accessible to salt, the salt will be deposited there too. The diameter of the core appears to be about 20-30 A in the E M, which is in agreement with the x-ray values. These outlined particles show one or two other features of interest, whose significance is unevaluated. They do not seem to show any obvious external grooving of the particle, and the short lengths of degraded virus do not seem to be showing any material protruding beyond the ends of the particle. Degradation always seems to take place by almost accurately transverse cleavage. The "outlining" technique would appear to be quite a useful one for this type of specimen, particu- larly as it is so simple and gives excellent contrast and resolution. J am indebted to Dr. J. D. Watson and Dr. R. Franklin for interesting me in this material. References 1. Bernal, J. D. and Fankuchen, I., /. Gen. Physiol. 25, 111 (1941). 2. Franklin, R. E., Nature 175, 379 (1955). 3. Franklin, R. E. and Klug, A., Biochini. Biophys. Acta 19, 403 (1956). 4. Hall, C. E., /. Biophys. Biochem. Cytol. 1, 1 (1955). 5. Schramm, G., Schumacher, G., and Zillig, W., Z. Natiirforsch. 10, 481 (1955). 6. V/atson, J. D., Biochini. Biophys. Ada. 13, 10 (1954). Electron Microscope Studies on the Periodicity in Tobacco Mosaic Virus R. E. F. Matthews, R. W. Horne and E. M. Green Moltciio Iiulitiac, Cambridge, ami Cavendish Laboratory, Cambridge Investigations by x-rays on the study of the sub- unit structure of tobacco mosaic virus (1, 3, 4, 5, 8) have suggested that such a structure is arranged in the form of a helix with a pitch of 23 A and a repeat distance of 69 A. This paper is concerned with an attempt to resolve such small surface detail on the surface of tobacco mosaic virus rods using a single stage carbon replica technique (2). Pre shadowed replicas applied to the study of crystal virus structures (6) indicated that small biological struc- tures could be resolved employing this technique. The work described here was carried out on purified suspen- sions of TMV dried down on carefully cleaned glass surfaces and carbon evaporated onto the preparation at an angle of about 30 from the horizontal. The carbon replica is then stripped from the glass surface and placed in 2 A' KOH at 60 C for approximately 20 minutes to remove any remaining rods attached to the replica. After a final washing the replicas were examined in the electron microscope at instrumental mangifications of ■' 40.000 and 80.000. Preshadowed replicas have pro\cd to be of little \alue as much of the structure is obscured by the deposited metal tending to confuse rather than reveal much of the fine structure. Three features have been resolved: (a) a structure repeating along the rod axis spaced at 46 A in the 262 A. FELTYNOWSKI Fig. 1. Single-stage carbon replica of tobacco mosaic virus rods showing periodic structure along the rod axis, longi- tudinal structure and hexagonal cross-section. Instrumental mangification 40,000. Final magnification 160,000. form of a "herring bone"' arrangement, (b) a longitu- dinal structure running the full length of the rod, (c) evidence that the cross-section of the rod is angular. Many of the rods have been resolved having a definite hexagonal cross-section which fit into the arrangment of longitudinal structures. The periodic "herring bone" structures are possibly due to carbon piling up on one side of a helix when the rod is lying along the direction of incidence of the evapo- rated carbon. A striking feature is the regular angle of the periodicity relative to the longitudinal struc- tures. In certain instances some of the longitudinal features appear as hollow grooves at the corners of the hexagonal cross-section. A through focal series of electron micrographs have shown that these ridges and grooves are not due to fringe effects caused by focusing. Preparations containing both tobacco mosaic virus and turnip yellow virus have been examined under identical conditions to confirm that the structures described are confined to the TMV. A serious limit- ing factor is the presence of very small material in the virus suspensions and considerable care must be taken in the preparation prior to replication in order to minimise these effects. From the preliminary work carried out in this laboratory employing carbon replicas, there is some evidence that the structures in TMV predicted by x-ray methods may be observed in the electron micro- scope. We are indebted to Dr. Roy Markham of the Molteno Institute for supplying the purified suspensions of the TMV, to Dr. V. E. Cosslett for many valuable discussions and also to the Agricultural Research Council for finan- cial aid. References 1. Bernal, J. D. and Fankuchen, I., /. Gen. Physiol. 25, 111 (1941). 2. Bradley, D. E., Brit. J. Appl. Phys. 5, 96 (1954). 3. Cochran, W., and Crick, F. H. C, Nature 169. 234 (1952). 4. Cochran, W., Crick, F. H. C, and Vand, V., Acta Crystal. 5, 581 (1952). 5. Franklin, R. E., Nature 175, 379 (1955). 6. Price, W. C. and Wyckoff, R. W. G., Nature 157, 764 (1946). 7. Sennett, R. S. and Scott, G. D., /. Opt. Soc. Amer. 40, No. 4 (1950). 8. Watson, J. D., Biochim. Biophys. Acta 13, 10 (1954). Filamentous Forms of Influenza Viruses A. Feltynowski State Institute of Hygiene, Department of Virology, Warsaw riLAMENTOUs fomis of infiucnza viruses were first observed by Mosley & Wyckoft'(8). Chu, Dawson & Elford (3) found that recently-isolated strains of the A-type frequently showed filamentous forms. They demonstrated that these forms were characterised by several properties of the elementary bodies of the influenza virus. Hoyle (7) found in the dark field of the light miscroscope protrusions in the infected allantois membrane and concluded that they contain virus material. It was Wyckoflf (11) who first exam- ined the membrane of the chicken embryo infected by influenza virus on thin sections and found that the cells apparently excreted influenza virus filaments. In a review of Angulo in 1951, the argument was presented that the filamentous forms were no viruses but breakdown products of the cell cytoplasm. In a previous paper (6) this argument was proven to be non-valid Most investigators concerned with the morphology of the influenza virus have described either the ele- mentary round bodies or the elongated forms from Filamentous Forms of Influenza Viruses 263 oval through short rods to filaments. It is only Archetti (2) and Draganow (5) who give a more comphcated description of the influenza virus, called by Draganow N.W. -bodies. In this paper we are concerned with the observa- tions on the influenza virus strains isolated in Poland (in the Virology Department of the State Institute of Hygiene headed by Prof. F. Przcsmycki) during the influenza epidemics in the years 1953, 1954 and 1955. The virus strains were as follows: — The 1953 epidemic: 23 strains were isolated, from which 14 were examined under the electron miscroscope. The isolated strains belonged to the A-type and only 2 were established as belonging to A-prime-type (see (9) where all methods are given). The 1954 epidemic: 12 strains were isolated, ail ex- amined morphologically and found to belong to the A- prime-type. The 1955 epidemic: 6 strains were isolated, all belong- ing to B-type. The laboratory characteristics of the last two epidemics are given in (10). The preparation of the specimen for the electron miscroscope is described in the paper previously men- tioned (6). One strain isolated from the 1954 epidemic (so-called "LAK"), belonging to the A-prime type was examined morphologically in more detail. Twenty passages were performed and after each of them the strain was observed under the electron microscope. The optimal passages were 10 ^ or 5 x lO"*. The main result was that the isolated strains were different morphologically. Some of them were fila- mentous, other had round forms. The well-known strain PR8 (6) showed long and filamentous forms but it ought to be considered as a strain of round forms because the elementary round bodies are prevalent here. One of the strains isolated in 1953 showed exceptional concentration of long filaments. This strain was a mixed one (obtained from several washings), and it may be interesting to note that the individual strains did not show any or very little filamentous forms. The filaments have thus become prevalent in the mixed strain. Another strain of the same epidemic (so-called "Z.Z."), be- longing to the A-prime type, can also be considered as "filamentous'". Most of the strains from this epidemic belonged to the A-type and they are of round forms. The long filamentous forms were not present, but the charac- teristic short beaded forms consisting of 3-4 beads were also observed (fig. 1 ). All the 12 strains from the 1954 epidemic belonged to the A-prime type and all should be considered as filamentous. They showed a great variety of forms (fig. 2). After several passages of the "LAK" strain the concentration of the filaments decreased. All the strains isolated in the year 1955 belong to the B-type. They consisted of elementary round bodies only. Fig. 1. Strain K.M. Type A- 1953. Magnification 18,000. hig. 2. Strain TM 1. Type A'- 1954. Magnification 1 1,000. In the "filamentous" strains the hemagglutination titer of the infected allantois fluid was very low and amounted from 1:40 up to 1:160. Although the morphological pattern of the "LAK" strain after the 1st and 1 5th passages was different, no diflerence was established in the antigenic structure. The "filamentous" strains of the virus show strong polymorphism. In addition to the round bodies, there are also "transitory" forms (from spheres to filaments) and complicated filamentous forms which were called by Draganow N.W. -forms. These poly- morphic bodies are characteristic only for the re- cently-isolated strains and were observed on the A-prime-type. All the B-type strains turned out to be those of round forms. In my view the filamentous forms present an ear- lier stage of the development of the virus which then pass into elementary round bodies. The present work was carried out under the guidance of Professor Przesmycki to whom I express my thanks. I should like to acknowledge the assistance of Mrs. K. Zgorzelska and Dr. B. Malczewski in the preparation of the virological samples. References 1. Angulo, J. J., Arcli. Virusforscli. 4, 199 (1951). 2. Archetti, L., Arch. Vinisforsch. 6, 29 (1955). 3. Chu, C. M., Dawson, I., and Elford, W. J., Lancet 256, 602 (1949). 4. Donald, H. B., Isaacs, A., J. Gen. Microbiol. 11, 325 (1954). 5. Draganow, K. I., J. Gen. Biol. {U.S.S.R.) 17, 13 (1956). 6. Feltynowski, a. and Zych, Z., Med. Doswiadc:alna i Mil>1 ^Q' >♦ ^ ♦-»^ Fig. 6. The polymorphism of phage particles of Bac. iiiyco- ides at the place of lysis of bacterial cell. The heads of phage particles with ring and spiral form (enclosed in squares), with solid granules (enclosed in rings) and formless (shown by pointers). or spirals. The filament twisted more tightly forms the head of a phage particle with granules, which optically reflects the places of bending of the fila- ment that forms the head. The filament twisted very tightly gives an impression of a homogeneous head of a phage particle (fig. 6). More and more data are accumulated testifying to the fact that a phage particle is a spiral or clew twisted filamentous aggregate of protein macro- molecules with a free end in the shape of a tail. References 1. Chapman, G., Hillier, J., and Johnson, F., J. Bacterial . 61, 261 (1951). 2. Freeman, V., J. Bacteriol. 61, 675 (1951). 3. Hercik, F., Casopis Lekdrii Ceskych 89, 91 (1950). 4. — ibid. 90, 925 (1951). 5. Jesaitis, M. and Goebel, W., /. Exptl. Med. 102, 733 (1955). 6. Kriss, a. E. and Tikhonenko, A. S., DAN 96, 42\ (1952). 7. — DAN 93, 353 (1953). 8. NoDA, H. and Wyckoff, R., Biochiin. Biophys. Acta 8, 381 (1952). 9. Pockotinsky, I. S., Krivisky, A. S., and Luzyanina, T. J., Zhunial mikrobiologii, epidemiologii i immutio- biologii 5, 19 (1951). 10. Williams, R. and Fraser, D., J. Bacteriol. 66, 45S {1953). Comparative Studies on Sections of Intact Cells, Protoplasts and "Ghosts" of a Bacillus species C. Weibull and K. G. Thorsson Central Bacteriological Laboratory of Stockholm City, Stockholm In 1952 it was shown (8, 12) that the eff"ect of the enzyme lysozyme on gram-positive bacteria is due to a dissolution of the bacterial cell wall. Soon afterwards it was demonstrated (13) that if the lysozyme treatment is carried out in a medium containing a protective agent such as sucrose or polyethylene glycol, the cell wall can be digested away without ensuing lysis of the protoplasm. Free bacterial protoplasts are thus formed. This indicates that the far-reaching lysis ordinarily effected by the enzyme, resulting in the formation of membraneous structures or "ghosts", is due to secondary phe- nomena, probably to a great extent of osmotic nature. The protoplasts were prepared as usual by lysozyme treatment of the cells in a sucrose containing medium. "Ghosts" were obtained by suspending protoplasts in dilute phosphate buffer. Fixation was effected by treating the structures with 1 per cent osmic tetroxide at 37°C for two hours. The fixed specimens were then dehydrated in alcohol, embedded in a mixture of 1 part of methyl and 9 parts of butyl methacrylate and sectioned. The organism investigated was a Bacillus strain named Bacillus M by Tomcsik (12). It is closely related to Bacillus mcgaterium and has even been regarded as a variant of this species (11). Electron micrographs of sections of Bacillus M cells reveal the same organisation as found in all other bacteria investigated so far with the same technique (fig. 1). The protoplasts show essentially the same struc- tural organisation as the intact cells, except for the absence of the cell wall in the former bodies (fig. 2). Intact Cells, Protoplasts and ""Ghosts" of a Bacillus Ibl The cytoplasmic areas and the nuclear equivalents are seen. Sometimes the outer border of the cyto- plasm appears slightly darker than the rest of it, suggesting the presence of a cytoplasmic membrane. These findings are in accordance with the fact that the protoplasts very much show the same biochemi- cal and physiological capabilities as the intact cells. The protoplasts thus respire at the same rate as the cells (13), they synthesize protein and nucleic acid from low molecular weight compounds (4) and they are able to produce adaptive enzymes (3, 5, 16). Moreover, bacteriophage particles and spores de- velop within them (I, 7, 9, 10). Under certain conditions protoplasts even exhibit growth and divi- sion phenomena (6). The absence of the cell wall structure in proto- plasts has earlier been indicated by analytical data (13), by light microscopical investigations (12, 14) and by the fact that they do not absorb phage particles (13). The "ghosts** consists of shells enclosing some protoplasmic material more or less in the form of granules but otherwise empty, as may be apparent from figs. 3 and 4. Thus earlier observations by means of light microscopy (12, 15), strongly suggesting that the "ghosts*" represent cytoplasmic membranes, are confirmed. On the other hand it should be empha- sized that some protoplasmic material evidently re- mains enclosed within this membrane even after the osmotic lysis of the protoplasts. Staining experiments performed on whole cells and "ghosts** using Sudan Black seem to indicate that the granules of Bacillus M. are not of lipid nature. In all events they should not be identified with the granules of polymerized /j-hydroxy-butyric Fig. 1. Section of intact cells of Bacillus M. Magnification >: 28,000. Fig. 2. Section of protoplasts of Bacillus M. Magnification X 38,000. Fig. 3. Section of "ghosts" of Bacillus M. Magnification 25,000. Fig. 4. Electron micrograph of drop preparation of "ghost" of Bacillus M. acid found in genuine B. nwi^atcrium strains. They seem instead to have more in common with the non-lipid granules that have recently been isolated by Georgi, Militzer and Decker (2) from the ther- mophilic bacterium Bacillus stcarothcrniophilus. These workers report that the granules contain desoxyribonucleic acid among other things and, from the enzymological point of view, cytochrome c and phosphatases. It should, however, be borne in mind that the granules found in the bacterial "ghosts" may be artefacts. The osmotic lysis must be regarded as a rather violent disintegration process, readily causing rearrangements between the protoplasmic constitu- ents of the bacterial cell. Such an interpretation is strengthened by the fact that no granules of the size and the electron density found in the "ghosts" of Bacillus M, are seen in sections of intact cells or protoplasts. References Brenner, S. and Stent, G., Bioc/iim. Bioplivs. Acta 17, 473 (1955). Georgi, C. E., Militzer, W. E., and Decker, T. S., J. Bactoriol. 70, 716 (1955). Landman, O. E. and Spiegelman, S., Pioc. Natl. Acad. Sci. Wash. 41, 698 (1955). McQuillen, K., Biochim. Biophys. Acta 17, 382 (1955). ^ /. Gen. Microbiol. 13, iv (1955). ^ Biochim. Biophys. Acta 18, 458 (1955). Mutsaars, W., Ann. Inst. Pasteur 89, 166 (1955). Sai.ion, M. R. J., Nature 170, 746 (I952J. — /. Gen. Microbiol. 13, iv (1955). Salton, M. R. J. and McQuillen, K., Biochim. Biophvs. Acta 17, 465 (1955). ToMCSiK, J., Symp. Soc. Gen. Microbiol. 6, 53 (1955). ToMcsiK, J. and Guex-Holzer, S., Schwei:. Z. allgcm. Pathol, u. Bakteriol. 15, 517 (1952). Weibull, C, J. Bacterial. 66, 688 (1953). — E.xptl. Cell Research 9, 139 (1955). — ibid. 10, 214 (1956). WiAME, J. M., SroRCK, R., and Vanderwinkel, E., Biochim. Biophys. Ada 18, 353 (1955). 12. 13. 14. 15. 16. A Study of the Division of Saccharomyces cerevisiae Using Carbon Replicas D. E. Bradley Research Laboratory, Associated Electrical Iiuhistries Limited, Aldermaston, Berkshire, England DECAUSE bacteria and similar organisms are dense to electrons, their surfaces cannot be satisfactorily studied by direct examination in the electron micro- scope. If the surface detail of the organisms is to be revealed, a replica technique must be employed. The most suitable method found so far uses evaporated carbon as a replicating material, (3), and is carried out as follows: A clean aqueous suspension of yeast cells is first pre- pared. This is most satisfactorily obtained from a liquid rather than a solid nutrient medium. Cells grown in 6°o malt extract solution were found to be free from con- taminating matter. Cultures were incubated for varying periods up to 72 hours at 33"C, and fixed for 30 minutes in 4 % formalin before being finally washed prior to replication. A single-stage replica technique was used in which the cell is coated with a layer of carbon and then dissolved away. The yeast cells were first dried down onto a thick formvar film mounted on an electron microscope speci- men support grid. It was found that a good dispersion could be obtained by spraying techniques, though there was a risk of damage to the yeast cells. The cells were next coated with a layer of carbon 150 to 200 A thick. It was then necessary to wash away the formvar substrate so that the yeast cells could be dissolved from the carbon replica. This was carried out by flowing a few ml of chloroform over the grid from a burette at the rate of 1 ml per minute. After the chloroform had evaporated. the grid was immersed in a solution of 3 g of a mixture of potassium permanganate and dichromate in concen- trated sulphuric acid for fifteen minutes. After removal, it was washed in water, then in concentrated hydro- chloric acid for a few seconds (this removed the manganese dioxide formed by the decomposition of the acid mixture), and finally in water. After drying, it was ready for shad- owing and examination in the electron microscope. Care must be exercised in the interpretation of electron micrographs of replicas produced by this technique, since a number of artifacts, which are easily recognisable, are liable to occur. The most obvious of these is distortion during the dehydration of the cell. This causes a very gross effect which can be clearly seen in figure 1. It is also likely that, in the case of large organisms such as yeast, the replica will become distorted and a similar effect produced. The rim at A in figure 2 is caused by cell dehydration and gives the appearance of the cell having been flattened against the formvar film. It was found that these artifacts were by no means a serious hindrance and undistorted cells could easily be found by systematic scanning of the grid. Two kinds of budding scar, illustrated in figure 1, were found on yeast cells. The "birth" scar (marked C), which is the scar formed at the point where a cell was attached to its parent, and the "bud"* scar (marked A and B), which is formed on the parent cell when the daughter becomes detached. These scars are quite different in form as can be seen from the figure. These observations are in agreement with those of Barton (2), who carried out his work opti- cally. The morphology of these scars is of great im- portance in a study of the mechanism of division. Electron micrographs of cells prior to division help in proposing a sequence of events for the process. Figure 2 shows the '"neck" connecting two cells Fig. 1. Shadowed carbon replica of single yeast cell showing "bud" scars at A and B and "birth" scar at C. (By courtesy oi Research.) Ma.gn\f\ca.Uo:\ 14,000. Fig. 2. Internal rims showing in the "neck" joining two cells before division. Shadowed carbon replica. (By courtesy of the /. Roy. Microscop. Soc.) Magnification 11,000. The Division of Saccharomyces cerevisiae Using Carbon Replicas 269 Fig. 3. Two cells broken apart before ready to divide. Shad- owed carbon replica. (By courtesy of tiie /. Roy. Microscop. Soc.) Magnification 14,000. before division. Two lines can be seen crossing the "neck" indicating that a rim has formed within the cell wall. In figure 3, a cell has been mechanically broken away from its parent before it was ready to divide, and an internal rim can be clearly seen, together with the broken cytoplasmic connection. It is thus possible to say that in the division process, two internal rims form in the neck connecting the two cells before the formation of transverse mem- branes which must be present when the cells are ready to divide. Now if a hypothetical longitudinal section is drawn of a birth scar and a bud scar and these are inverted over each other, it will be found that they key together. Thus it is possible to propose the sequence of events shown in figure 4. (a) shows Oouqhher Daughher Porenh Fig. 4. Diagrams of hypothetical longitudinal sections through the "neck" joining two cells illustrating structural changes during the division process. (By courtesy of the J. Roy. Microscop. Soc.) Fig. 5. Electron micrograph of ihin section through the "neck" illustrating the key mechanism. (After Agar and DoLiglas, by courtesy of the J. Bacleriol.) Magnification X 21,000. the first stage in the formation of the internal rim and in ih) the rims are nearly completely formed be- fore the transverse membranes, in (c) the interlocking scars are shown in section. These sections arc easily deduced by studying the electron micrograph in figure 1. id) shows the last stage in the process as proposed by Barton (2) in which the birth scar on the daughter cell grows outwards, thus releasing the key mechanism. An independant investigation by Agar and Doug- las (I), who examined sections of yeast concurrently with the present work, showed that the above con- clusions are in agreement with information provided by sections. The key mechanism is well illustrated in the micrograph of the section shown in figure 5. This work is more fully described elsewhere by Bradley (4). in addition to the study of yeast cells, the carbon replica has been applied to the study of the surface structure of some bacillus spores. These are e\en Fig. 6. Carbon replica of spores of B. brevis. Magnification X 25,000. 270 D. E. BRADLEY Figs. 7-8. Carbon replicas of spores of i?. 5///>r//M'. Magnifi- cation 27,000. more dense to electrons than the bacteria themselves and only a silhouette is seen. In a preliminary ex- amination, spores of Bacillus snhtilis and Bacillus hrevis were chosen and replicas prepared by the method outlined above. Figure 6 shows some typical spores of B. brevis. The surface seems to be relatively smooth save for a single rib which runs longitudinally down the cell. In the case of B. subtilis however, a larger num- ber of ribs are present, and though these tend to run longitudinally down the cell (figure 7), they fre- quently intersect and form a network such as that shown in figure 8. It seems extremely unlikely that the ribbing is due to shrinkage, if only because of its form. Further- more, the water content of bacillus spores is very small and in addition, it is hardly possible that shrink- age would cause one rib running lengthwise down every cell as is the case with B. hrevis. It therefore seems certain that the structure is genuine. At this stage it is not possible to explain the ribs or their function, but it is hoped that a combined sectioning and replica study together with studies of spores after various treatments will provide much information on their structure. The author is grateful to Prof. B. C. J. G. Knight of the Department of Microbiology, University of Reading, for his valuable advice on the study of yeast, and also to Mr. D. J. Williams of The National Institute for Research in Dairying, who is collaborating with the author in the study of bacterial spores. Dr. T.E. Allibone, F.R.S. Director of the A.E.I. Research Laboratory, has kindly given permission to publish this paper. References 1. Agar, H. D. and Douglas, H. C, /. Bacteiiol. 70, 427 (1955). 2. Barton, A. A., /. Gen. Microbiol. 4, 84 (1950). 3. Bradley, D. E., Brit. J. Appl. Pins. 5, 65 (1954). 4. — /. Roy. Microscop. Sac. 75, 254 (1956). xu BOTANY Die Ontogenese der Chloroplasten von Chlorophytwu comosum E. S. Perner Botanisches Institiit der Universifcit Miinsteij Westf. Uber die Ontogenese der Chloroplasten hoherer Pfianzen liegen heute drei Hypothesen vor: die von Strugger (7, 8) auf lichtmikroskopischen, die von Miihlethaler (4) und Leyon (3) allein auf elektronen- mikroskopischen Befunden aufbauend. Sie unter- scheiden sich in hezug auf den Bau des Proplastids — dem primaren Entwicklungstadium des Chloropla- sten — die Natur des Primargranums und dessen Bedeutung fur die Entstehung der Sekundiirgrana und Tragerlamellensysteme im fertigen Chloroplasten. Die Analyse von Chlorophytuni coniosuiu, einer Liliiflore, die nach bisherigen licht- und elektronen- mikroskopischen Erfahrungen gut geeignet erscheint, diirfte zur Klarung der bestehenden Widerspruche beitragen. Auch im Urmeristem des SproBvegetationskegels und in den jijngsten Blattanlagen lassen sich in Ubereinstimmung zu lichtmikroskopischen Kon- trollen Proplastiden nachweisen,die im Stroma einen stark elektronenstreuenden Komplex enthalten, der nach Lage und GroBe dem stark farbbaren Primar- granum lichtmikroskopischer Untersuchungen ent- spricht (Abb. 1). Bei ausreichender Vergleichsmog- lichkeit ist eine Verwechslung der Proplastiden mit Chondriosomen oder Spharosomen im elektronen- mikroskopischen Bild auszuschlieBen. Die Form und GroBe von Proplastid und Primargranum be- dingen aber imZusammenhang mit der verschiedenen Lage des Proplastids innerhalb der Zelle, daB in einer groBen Anzahl von Schnitten entweder nur das Stroma bzw. vorwiegend das Primargranum getroflfen wird. Im ersten Fall kommt man zu den granum- freien ,, dense bodies" im Sinne von Leyon, im zweiten zum ,,osmiophilen Granulum" nach Miihle- thaler, die im Gegensatz zu Strugger als die primaren Entwicklungstadien von Chloroplasten angesehen werden. Wenn Miihlethaler und Leyon erst in etwas alteren Bliittchen Proplastiden typischer Ausbildung auffin- den konnen, so liegt das an der groBeren Wahr- scheinlichkeit, hier Proplastiden in charakteristi- scher Schnittrichtung zu erfassen, denn es hat 1 ) eine lebhafte Vermehrung der Proplastiden durch Teilung stattgefunden, und die Proplastiden sind 2) durch Wachstum von Stroma und Primargranum groBer geworden. Dies steht in absoluter Uberein- stimmung zu lichtmikroskopischen Erfahrungen (vgl. 9). Ebenso wie Miihlethaler und Leyon haben auch wir gefunden, daB das Primargranum in den jiing- sten Blattanlagen nicht die von Heitz (1) und Leyon (2) zuerst aufgefundene kristallartige Ordnung auf- weist. Sie ist regelmiiBig erst bei Blattchen von etwa 0,5 cm Lange ab zu finden. Dieser scheinbare Unter- schied diirfte auf die schlechte Fixierbarkeit und die starkere traumatische Schadigung des SproBvege- tationskegels im Vergleich zu den leicht isolierbaren und gut fixierten Blattchen zuriickzufiihren sein. Dafiir spricht der schlechte Erhaltungszustand des Cytoplasmas im Urmeristem und der einwandfreie bei alteren Blattchen. Nach Heitz soil das Primargranum ein Chloro- phyllkristall sein, der erst sekundar aus dem Stroma neu entsteht und der demnach kein persistierender Bestandteil des Proplastids sein kann. DaB Chloro- phyll im Primargranum vorhanden ist, haben fluo- reszenzmikroskopische Beobachtungen gezeigt. DaB aber auch Nucleinsauren enthalten sind zeigen cyto- chemische Befunde von Spiekermann (6) und Ruch (5). ]m elektronenmikroskopischen Bild ist die inten- sive Osmierung der entscheidende Faktor fiir die Darstellung des Primargranums. Fiir die Reduktion von OsO^ kommen aber verschiedene StofTgruppen in Frage: ausser Fetten und Lipoiden auch EiweiB- ■P .\bb. J. Proplastid aus dem SproBvegetationskegel mit Pri- margranum (17 000 < ). Abb. 2. Proplastid aus etwa 1 cm langem Biatt. Das Primar- granum zeigt einen Auf bau aus Lamellen (40000 > ). Die OntogcHCSc dcr Chloruplastcii von Cliluropliyiiiiu coniosiini 273 Abb. 3. Aiisschnitt aus dem Primargrainim (96000 >' )• Abb. 4. Jiingcliloroplast mit bcginncnder Slromalamcllierung (46000 - ). korper, aber nicht Nucleinsauren. Wie die braun- schvvarze Fiirbung des Mediums zeigt, fiihrt eine 23 Tage lange Behandlung des osmiumsaurefixierten Materials zu einer Extraktion alkoholloslicher und zur Reduktion von Osmiumsaure befahigter Sub- stanzen, in erster Linie Fette und Lipoide. Die Starke Elektronenstreuung des Primargranums bleibt aber fast unveriindert erhalten. Am Aufbau sind demnach auch alkoholunlosliche und zur Reduktion von Osmiumsaure befahigte Stoffe beteiligt, wahr- scheinlich EiweiBkorper. Das Primargranum ist da- her stoftlich heterogen und diirfte einen Chromo- Lipo-Nucleo-Proteid-Komplex darstellen. Nach Heitz wird das Primargranum weiterhin als ein einfacher bzw. zusammengesetzter Kristall be- trachtet, wobei ,,kreisrunde, dunkle Flachen" als Gitterpunkte des dreidimensionalen Systems ange- sehen werden. Bei geeigneter Schnittrichtung zeigt das Primargranum aber parallele Lamellen und dijrfte demnach eher ein komplexer Schichtenkorper sein (Abb. 2). An extrem diinnen Schnitten war die Anwendung einer stiirkeren PrimiirvergroBerung (bis zu 44000) moglich und es gelang, die angeblichen Gitterpunkte weiter aufzulosen. Es sind in den Bil- dern etwa 170-190 A groBe Gebilde zu erkennen, die trotz verschieden langer Fixation nur aussen osmiert sind, im Inneren aber hell erscheinen. Sie durften demnach stofflich heterogen aufgebaut sein. Gegen die Kristallnatur spricht ferner, daB diese Elemen- tareinheiten in spezitischer Weise miteinander ver- bunden sind, wobei der Abstand etwa 400-500 A betriigt (Abb. 3). Die groBe biologische Bedeutung des Primargra- nums ergibt sich aus dem Verhalten bei der weiteren Ontogenese, was bereits seit Leyon und Miihlethaler bekannt ist. Die im Jungchloroplasten entstchen- den Stromalamellen gehen von den Elementarein- heiten des Primargranums aus, die dabei anscheinend unter Mitwirkung des Stromas zu Lamellen werden (Abb. 4). Der Jungchloroplast wird schlieBlich voil- kommen von Stromalamellen erfijllt, die in Richtung 18-568204 Electron Microscopy der Lamellen des Primargranums verlaufen, wenn in der richtigen Ebene geschnitten wurde. Die typische Struktur des Primargranums verschwindet zugunsten dichter Lamellenlagen, wobei bereits die ersten Sekundargrana sichtbar werden. Mit der Persistenz des Primargranums steht die Frage nach der bereits lichtmikroskopisch postulier- ten Teilungsfiihigkeit im Zusammenhang mit der Teilung des Plastids zur Diskussion. Drei Schnitte durch einen noch amoboiden Jungchloroplasten im Stadium beginnender Stromalamellierung zeigen, daB hier ein bigranulares Plastid vorliegt. Nachdem eine spontane Entstehung von Primiirgrana niemals beobachtet worden ist, konnen sie nur durch Teilung entstanden sein. Die hantelformige Einschniirung bei einem anderen Jungchloroplasten deutet auf ein Teilungstadium hin. Jedes Tochterplastid wird je ein Primargranum enthalten, die entweder direkt sichtbar sind oder an den bereits vorhandenen Stromalamellen erkannt werden konnen. Diese Befunde bei Chlorophytiini comosum fordern eine Revision des Entwicklungsschcmas von Miihle- thaler und Leyon in bezug auf den Bau des Propla- stids und die Natur des Primargranums im Sinne der Auffassung von Strugger. Das lichtmikroskopische Schema von Strugger muB — jedenfalls fiir dieses Objekt — den neueren submikroskopischen Befun- den angepaBt werden. Das Primargranum geht in ein System von Stromalamellen ein, aus denen sich die Sekundargrana und Triigerlamellen diflferenzieren. LiTF.RATUR HEirz, E., E.xptl. Cell Research 7, 606 (1954). Leyon, H., E.xptl. Cell Research 7, 609 (1954). — Svensk kern, ticlskr. 68, 70 (1956). MiJULETHALr.R, K., Protopl. 45, 264 (1955). RucH, p. D., Vorlr. Deutsch. Bot. Gcs. Tagung in Hann. Munden. (1956). 6. Spiekermann, R., Protopl. (1957, im Driick). 7. Strugger, S., Satiirwiss. 37, 166 (1950). 8. — Ber. dent, botan. Ges. 66, 439 (1953). 9. Strugger, S. und Ferner, E., Protopl. 46, 711 (1956). Some Botanical Applications of the Carbon Replica Technique D. E. Bradley Research Lahoratory, Associated Electrical Industries Ltd., Aldermastoii, Berkshire It is likely that a knowledge of the sub-microstruc- ture of the sporoderm would be of some value in a study of post-glacial flora. The electron microscopy of the surface structure of pollen grains has been made possible by the use of carbon replicas: the technique was first applied in this field by Muhlethaler (4). In the present work, over sixty different species were studied in order to ascertain firstly, the reliability of the replica method developed for the purpose, and secondly, the range of structures which might be encountered in a wide study of pollen. The replica method was identical with that described elsewhere (1), save for one modification. It was found more satisfactory to dust pollen onto the initial Formvar substrate and then disperse it with acetone or alcohol, rather than to suspend the grains in water and dry down a drop of the suspension. It was found that the efficiency of the technique depended on the size of the pollen grain. In the case of large grains, the method was only 10 % efficient, i.e. only 10 out of 100 grains were replicated, the remainder breaking up and leaving a hole in the replica film. It is, of course, only necessary to take micrographs from a few grains so that this 10% efficiency was found to be quite adequate. In the case of small grains, the efficiency was at least 50 ';'o. It was found that the use of acetone or alcohol to disperse the grains caused the absorption of suffi- cient moisture during the evaporation of the solvent to cause the expansion of some grains. Many pollen grains are deeply divided by grooves, and the swell- ing causes the grooves to be pushed outwards so that a grain in the form of an ovoid becomes a sphere. This is a well known phenomenon. It was found that the fine detail of the grain was not dis- torted and that it was possible to study the structure within the groove. An example is shown in figure I, which is a micrograph of part of a grain of Lawium alhiim. The wedge shape represents the groove. Structures were found to vary very greatly. In some cases, there was much fine detail such as that shown in figure 2 which is the surface of a grain of Rhododendron; most of this detail is of the order of a quarter of a micron in size, and not properly re- solved by the optical microscope. In many cases, however, the structure was much coarser as in figure 3 (Bryonia dioica) . Most of this detail which is two to three microns in size, could be easily resolved optically, but the electron microscope provides a different viewpoint, and of course, a very much clearer picture. It is certainly likely that electron micrographs such as these will be valuable in a study of pollen. The examination of fungus spores. — Figure 4 shows a shadowed carbon replica of a spore of Russiila venosa. This spore is believed to be covered with a thin surface film known as the amyloid layer. The evidence for its presence is that the spore gives a very marked iodine reaction, but this reaction does not occur if the spore is first treated with dilute so- dium hydroxide. Some optical evidence of this film has also been described. The matter is still contro- versial. In the replica shown, there is certainly a surface film present as it has been broken at A and B. This film appears to be from 200 to 500 A in thickness, according to the position on the spore. The film does not extend over the plage (the rather rough region immediately behind the apicle). This Fig. 1. Shadowed carbon replica of pollen grain of Laniiinii album showing detail within the groove. Fig. 2. Shadowed carbon replica of part of pollen grain of Rhododcudron. Fig. 3. Shadowed carbon replica of part of pollen grain of Bryonia dioica. Fig. 4. Carbon replica of a spore of Russula venosa. {A) and (B) mark places where the suspected amyloid layer has broken. Shadowed. Botanical Applications of the Carbon Replica Technique 275 Fig. 5. Shadowed carbon replica of a scale of Mallonioiuis leboiinii. Fig. 7. Direct unsiiadowed electron micrograph of a scale of Syniiiii I'cliiniilata. Fig. 8. Shadowed carbon replica of a scale of S. echinulani. is a significant point since the plage does not show the iodine reaction. On the basis of this evidence, it seems Hkely that this surface film is, in fact, the amyloid layer detected by the iodine reaction. The electron microscopy of the scales of Malloinonas and Syniira. — One of the most interesting botanical applications of the carbon replica technique has been the study of the scales of Mollonionas and Syniira. These organisms are classed in the Algae and lie on the borderline between plants and ani- mals. They are unicellular and covered with a flexible armour of silica scales which are much used in the identification of species and exhibit elaborate characteristic ornamentation. Mallomonas has fairly large scales and frequently long bristles. Electron micrographs of replicas provide an en- hanced three-dimensional effect, differentiate be- tween the inner and outer surfaces of the scales, and permit the examination of complete organisms. The replica technique used was a single-stage method, the scales being covered with a film of carbon and then dissolved away with hydrofluoric acid. Figure 5 shows a replica of a scale of M. leboimii. The array of scales on the surface of an Fig. 6. Shadowed carbon replica of part of a cell of Mallo- monas ptiniilio. (By courtesy of the J. Roy. Microscop. Soc.) organism is shown by replicas of whole cells, as in figure 6 which shows one end of a cell of M. piiniilio. This is a new species with scales differing only slightly from another closely allied species. The difterence can only be seen in the electron micro- scope. The scales of Synura are much smaller and more delicate than those of Mallomonas, and the case for using replicas is not quite so strong. However, it has been found that a complete picture of the structure of the scale can only be obtained both by examining scales directly and using replicas. For example, in figure 7, which shows a scale of S. echiniilata photo- graphed directly, the structure at the base of the spike is clearly shown. In the case of the replica shown in figure 8, this structure is absent, indicating that it is actually internal. This fact is not revealed by shadowing the scale itself. The inner surface of the scale of S. echiniilata is not illustrated here; it is, in fact, the other side of the punctate region shown in both micrographs. The holes do not extend over the whole of the inner surface. It is hoped that the electron microscope will clear up many of the difficulties encountered in identifying the species of these two genera. This work is de- scribed in greater detail elsewhere (2, 3). The author is grateful to Prof. T. Harris of the Botany Department, University of Reading, for his help with the work on pollen, to Mrs. K. Harris with whom the work on Malloiuoiuis and Synura has been carried out, and also to Dr. P. B. Hora of the same Department with whom the work on fungus spores is being carried out. Permission to publish was given by Dr. T. E. Allibone, F. R. S., Director of the A.E.I. Research Laboratory. References 1. Bradi KY, D. F., These Proceedings, p. 268. 2. Harris, K. and Bradlfy, D. E., Discovery, 17, .^29 (1956). 3. — /. Roy. Microscop. Soc. (1957, in press). 4. MUHLFTHAITR. K., Plaiita 46. 1 (1955). On the Ultrastructure of a Fungus: The Gametes of A/Ioinvces G. TuRiAN and E. Kellenberger Botanical Institute and Biophysical Laboratory, Institute of Physics, University of Geneva IVloTiLE cells (gametes, zygotes, zoospores) of Allo- myces sp. (aquatic Phycomycetes) exhibit a nuclear cap giving positive cytochemical tests for RNA (4). Ultrathin sections of the gametes of Allomyces ma- croi^ynus Emers. (OsO, fixation, methacrylate embed- ding medium, according to Turian and Kellenberger (6)) show that this basophilic cytoplasmic formation is surrounded with mitochondria. The lamellar struc- ture of these fungal mitochondria is similar to that already described for animal mitochondria (1,2, 3). The nuclear cap is made of a granular substance, devoid of membranous structure and denser than that of the nucleus. Peripheral, non-basophilic cytoplasm does not contain any granular constituent but seems to con- tain some membranous structures. In fact, no typi- cal ergastoplasm is formed either in the basophilic nuclear cap or in the peripheral non basophilic cyto- plasm. The nuclear membrane is a double structure. Pores are visible in transverse sections as well as an alveolar structure in tangential and oblique sections through this membrane. In the germinated zygotes, the nuclear cap is dis- sociated and basophily extends to the whole cyto- plasm of the seedling (5). Our latest results reveal that this extension of basophily corresponds, on an ultrastructural basis, to a generalized presence of the granular constituent in the cytoplasm of the seedling. References 1. Palade, G. E., Anat. Rec. 114, 427 (1952). 2. Rhodin, J., Correlation of Ultrastructural Organization and Function in Normal and Experimentally Changed Proximal Convolute Tubule Cells of the Mouse Kidney. Stockholm, 1954. 3. Sjostrand, F. S., Nature 171, 30 (1953). 4. Turian, G., Compt. rend. acad. sci., Paris, 240, 2343 (1955). 5. — E.xperientia 12, 24 (1956). 6. Turian, G. and Kellenberger, E., E.xptl. Cell Research 11, 417 (1956). Elektronenmikroskopische Beobachtungen liber die Warzenstruktur bei (den Koniferen W. LlESE Forsthotanisches Institut der Univcrsitdt Freiburg i. Br. nnd Rheinisch- Westfdlisches Institut fiir Vbermikroskopie, Diisseldorf L)er Einsatz des Elektronenmikroskopes auf dem Gebiet der Holzanatomie hat in zunehmendem MaBe nicht nur die bisherigen Vorstellungen Liber den Feinbau der Zellwande weitgehend bestatigen kon- nen, sondern auch zu ganz neuen Erkenntnissen gefijhrt. So konnte fiir die Tertiarwand, die als innereZell- wandschicht das Lumen der Langstracheiden ausklei- det, die teilweise recht unterschiedliche Anordnung der zellulosischen Elementarfibrillen weiter aufge- klart und ihre Einbettung in die akzessorischen Begleitsubstanzen sichtbar gemacht werden. AuBer- dem war es mit dem Elektronenmikroskop erstmals moglich, eine bis dahin noch unbekannte Erschei- nungsform der Tertiarwand, die sog. Warzenstruk- tur, festzustellen. Hierunter versteht man kleine, warzenahnliche Gebilde, die als zellwandeigene Bestandteile den Wanden der Tracheiden und der Hoftupfelkammern aufgelagert sind (Bild I, 2). Je nach dem Grad der Inkrustierung sind zwischen den Warzen die Elementarfibrillen der Tertiarwand mehr oder weniger deutlich sichtbar. Beim partiellen AbreiBen der Tertiarwand von der darunter liegen- den Sekundarwand sind die Warzen stets nur auf jener Zellwandschicht vorhanden (Bild 3). Eine Ver- wechslung mit ilhnlich aussehenden Praparations- artefakten kann daher ausgeschlossen werden. Ha- rada (5) iiberpriifte ihre Originalitiit auBerdem an Diinnschnitten von nativem Holz. Nach der ersten Entdeckung dieser submikro- skopischen Warzenstruktur (8, 9) wurden an zahl- reichen Holzarten verschiedener Gattungen einge- hende systematische Untersuchungen durchgefiihrt, um niihere Kenntnis fiber Vorkommen und Erschei- nungsformen zu erlangen (1, 2, 4-6, 10-12). Hierbei ergab sich, daB nur bestimmte Holzarten eine War- Die Warzcnstruktw 111 Abb. 1. Warzenstruktur auf der Traclieidenwand von Pinus patiila: elektr.-opl. 7900 x. Abb. 2. Warzenstruktur auf der inncrcn Hoftiipfelwand von Pinus silvestris; elektr.-opt. 7900 x. Abb. 3. Tracheidenwand von Pinus Jeffreyi; die Tertiarwand ist im untercn Bildlcil abgerissen, so daB die Sekundiirwand erscheint: elektr.-opt. 2600 Abb. 4. Innere Hoftiipfelwand von Picea e.xcelsa mit vereinzelten Warzen: elektr.-opt. 7900 Abb. 5. Warzenstruktur auf der Tracheidenwand von Widdringtonia drcicomoniiuur. lichtopt. 2200 x . Abb. 6. Warzenstruktur auf der Tracheidenwand von Widdringtonia dracomontcinci; elektr.-opt. 7900 zenstruktur besitzen, andere hingegen stets glatte Tracheiden- und Tiipfelwande aufweisen. Unsere neuen Beobachtungen zeigen jedoch, daB auch bei dieser Gruppe vereinzelt Warzen besonders auf den Hoftupfelwanden vorkommen konnen (Bild 4). Die GroBe dieser submikroskopischen Struktur- elemente schwankt in einem recht weiten Rahmen und kann innerhalb einer Art, wie z. B. bei Pinus silvestris, von 20 bis 280 m// reichen. Die groBeren Warzen befindcn sich dabei schon im lichtoptischen Bereich; durch die Kontrastarmut wird jedoch ihre Sichtbar- keit sehr erschwert. Unsere Messungen an zahl- reichen Koniferenarten ergaben einen arithmeti- schen Mittelwert, der meist zwischen 100 und 140 m// liegt. Der allgemein recht weite Streuungsbereich macht es vorerst noch schwer, die vorhandenen GroBenunterschiede statistisch abzusichern, um sie als artdiagnostiches Hilfsmittel verwerten zu konnen. Die WarzengroBe ist auf den Tracheidenwiinden im Durchschnitt starkeren Schwankungen unterworfen als auf den Tupfelhofen; auBerdem sind die Warzen auf den Tracheidenwiinden vergleichsweise etwas groBer. Die Hohe der Warzen betriigt durchschnitt- lich etwa 50-75 "„ des jeweiligen Durchmessers. Ebenso wie die GroBe variiert auch erhebiich die Verteilungsdichte, d. h. die Anzahl Warzen je //'-. Obwohl die Verteilung selbst meist regelmaBig ist und nur der Abstand schwankt, konnen die Warzen mitunter auch aufTallend gehauft vorkommen (vgl. Bild 1 und 3). Die Warzen der Hoftiipfel zeigen in der Verteilungsdichte ebenfalls wieder geringere Unterschiede als diejenigen auf den Zellwiinden. Wenn auch fijr die meisten Holzarten eine bcstimmte Dichte charakteristisch ist, so sind doch gelegenthch, eng nebeneinander, alle Ubergiinge von warzenfreien bis zu dicht bewarzten Flachen aufzufinden. GroBe und Verteilungsdichte korrespondieren nicht mitei- nander. Die Untersuchungsergebnisse vom Vorhandensein oder Fehlen der Warzenstruktur sind bei den bislang untersuchten Arten jevveils fiJr die iibergcordnete Gattung cinheitlich. Lediglich die Kicfcrn bilden eine deutliche Ausnahme(l,2, 10). Klarlasscn sich hicr die beiden systematischen Untergattungen voneinander trennen; die Haploxylon-Kiefern mit fehlender oder nur spiiriicher Warzenstruktur und die Dipioxylon- Kiefern mit einer meist auffailigen, dichten Struktu- rierung. Die Befunde fiir die einzelncn Arten konnen starker als bei anderen Gattungen von einem jewei- ligen Normzustand der GroBe und Verteilungsdichte abweichen. Innerhalb der Gattung Pinus ist die Warzenstruktur daher als sehr variabel anzusehen. Wahrend die Existenz dieser submikroskopischen Warzenstruktur erst durch die elektronenoptische 278 E. LIESE Tabelle 1 . Streiiimgsbreite der Warzenstruktur hei einigen der imtersuchten Koniferenarten. Umfang des Kollektivs Maximum der Hdufigkeit gewog. arithmetrisches Mittel mit a FamiLie Cuptvssa- ceoB. Pinacc?(JB Gcdtu ns Callitris Widdrinqtonia Abies Cedrus Pl nu5 krt rebuild dfa co»**On^Qna i/bd f,r Hofw Zeitw 2ellw Hofw Zeilw En 100 100 10O 100 100 Lasiocatpa pirisapo deodara attantica kor^t-ensi- i %irok>ui par/L^lora bun ^ecxri oL monia n a. ^a ir\ ai a SiivestriS Hofw Zedw Zedv Hofw Mofi. Hofw 2e/lw Hofw Hofw Hofw leilw Hofw ZeUw 50 100 100 100 100 ■foo 300 SO 50 50 100 100 50 SO Hofw ZeUw nxun caia. fO^asi^a. clfifjualjuanct pond e'osa. £nqe(.w>anni. c po-tuLla. Sa t>Ki la no. Hofw lellw Hofw He,fv leilu Hofw Hofw IcUw Hofw Hofw Zt-Uw Hofw ZtUw 50 SO hOO 300 SO 50 50 100 50 50 SO SO 50 50 50 50 'too 150 100 ISO 300 350 400 mfj I I • 1 I 1 I < JJW- JUX ill. -J- JJ. JL-J. JUL JL ilL J— L JL JL JL iLL JO. JD. 41— t JL JL Jl. X. JIL ■r 960 — *i 50 100 ISO 200 250 300 350 400 mp Untersuchung festgestellt werden konnte, ist von einigen weniger verbreiteten Koniferen eine bereits lichtmikroskopisch sichtbare feinkornige Strukturie- rung der Zellwande bekannt (Bild 5) (3, 7). Es war von besonderem Interesse, diese vereinzelten Beobachtungen durch eine elektronenmikroskopi- sche Analyse zu erganzen, um festzustellen, ob hier ein Zusammenhang mit der submikroskopischen Warzenstruktur vorliegt. Die Untersuchung mehrerer Arten verschiedener Gattungen (Callitris, Widdringtonia, Pilgerodendron, Pherosphaerd)^ ergab auf den Zellwanden auf- fallend groBe und unterschiedlich geformte Gebilde, welche die Dimensionen der bisher beobachteten Warzen um ein Vielfaches iibertreffen (Bild 6). Ihre AusmaBe reichen weit in den lichtoptisch erfaBbaren Bereich hinein und konnen vereinzelt fast bis zu 1000 m// betragen. Es besteht wohl kein Zweifel, daB die beschriebene feine Kornung der Zellwande durch diese ,,Makro-Warzen" hervorgerufen wird. Zwischen ihnen befinden sich in regelloser Verteilung zahlreiche kleinere und kleinste Warzen bis zu etwa 20 m/<. Eingehende Messungen ergaben fast kontinuierlich alle tJbergange von den kleinsten bis zu den groBten Formen. Ahnlich wie bei der bereits bekannten, eingangs erorterten Warzenstruktur ist die Verteilungsdichte ^ Fiir die Uberlassung der Holzproben danken wir herz- lich Herrn Prof. Dr. P. Greguss, Szeged, und der Bundesfor- schungsanstalt fiir Forst- und Holzwirtschaft, Reinbek. Die Warzenstruktur 279 25 so 100 ISO 200 2S0 300 350 600 950 1000 m/J. ^rzergrosse Abb. 7. GroBenverteilung von je 300 Warzen auf denTrache- idenwiinden von W'iddringlonia chacomontnna ( ) und Piniis siheslris ( ). hier ebenfalls recht unterschiedlich und reicht von groBerer riiumlicher Trennung bis zu einem schol- lenformigen Muster. Mit der GroBe wiichst meist auch die Hohe der Warzen, die vereinzelt bis zu 1 // betragen kann. Ein ganz anderes Erscheinungsbild zeigt sich je- doch auf den inneren Hoftiipfelwanden. Hier sind meist nur kleine Warzen vorhanden, und die groBen Formen fehlen. Wahrend sich daher die Trache- idenwande z. B. von Finns silvestris und Widdringto- nia dracomontana elektronen- und sogar auch licht- optisch eindeutig voneinander unterscheiden, besteht bei einem Vergleich der inneren Tupfelwande kaum ein wesentlicher Unterschied. Es ergibt sich somit wieder, daB die Tupfelwande geringere Schwankun- gen hinsichtlich GroBe und Verteilungsdichte der Warzen zeigen, als die tracheidalen Zellwande. Um weitere Kenntnisse iiber die Warzenstruktur dieser Holzarten zu erlangen, wurde die GroBen- verteilung der Zellwandwarzen naher untersucht. Bild 7 zeigt als graphische Darstellung die Ergeb- nisse von Widdringtonia dracomontana, wobei die groBe Anzahl von kleineren Warzen besonders auf- fallt. Zwischen 25 und 130 m/^ liegen 53 "o aller gemessenen Werte. Etwa ab 200 m// bleibt ihre Anzahl trotz zunehmender GroBe ziemlich konstant und sinkt erst bei 400 m/< starker ab. In das Koordi- natensystem wurde zum Vergleich auch eine entspre- chende Kurve fur Pinus silvestris, einer Holzart mit submikroskopischen Warzen, eingetragen. Es er- gibt sich hierbei eine sehr deutliche Ubereinstim- mung des Kurvenverlaufes; die Zahlenwerte fiir den Kurvenansticg und sogar fiir die Lage des Hiiufig- keitsmaximums liegen dicht beisammen. Lediglich der rechte Kurvenast fiillt bei Widdringtonia etwa von 200 mn an infolge der ,,Makro-Warzen" bedeu- tend flacher ab. Ahnliche Ergebnisse zeigen sich auch bei einer Gegeniiberstellung andcrcr Artcn (Tab. 1). Hieraus kann gefolgert werden, daB die so unter- schiedlichen Warzenformen der verschiedenen Koni- feren einer einzigcn Elementarstruktur angehoren, wobei lediglich WarzengroBe und -form sowie die Verteilungsdichte von der Art beinfluBt sein konncn. Als zusammcnfassendes Ergebnis liiBt sich aus den Untersuchungsbefunden von iiber 120 verschiede- nen Nadelholzartcn entnehmen, daB die Warzen- struktur der Tertiiirwand aufgelagert ist und bei den Koniferen hiiufig vorkommt. ihre Ausbil- dung kann jedoch sehr unterschiedlich sein. Wahrend einige Arten und Gattungen nur recht selten Warzengebilde zeigen, sind sie bei anderen stets regelmiiBig vorhanden. Ihre GroBe reicht von sub- mikroskopischen Dimensionen bis zu lichtopti- schen AusmaBen. Die Klarung der hiermit zusam- menhiingenden entwicklungsgeschichtlichen und che- mischen Probleme steht noch aus. Fur die Unterstiitzung bei den elektronenmikrosko- pischen Beobachtungen danke ich Herrn W. Steinmetz, Diisseldorf, herzlich. Mit Unterstiitzung der Deutschen Forschungsgemein- schaft. LiTERATUR 1. Frey-Wyssling, a., MOhlethaler, K., und Bosshard, H. H., Holz (lis Roll- II. ^Verkstoff 13,245-249 i\955). 2. — ibid. 14, 161-162 (1956). 3. Greguss, p., Xylotomische Bestimmung der heule leben- den Gymnospermen. Budapest, 1955. 4. Harada, H., J. Japan. Forestry Soc. 35, 393-396 (1953). 5. — /. Japan. Wood Research Soc. 1, 85-89 (1955). 6. Harada, H. und Miyazaki, Y., J. Japan. Forestry Soc. 34, 350-352 (1952). 7. HuBER, B. und Rouschal, Ch., Mikrophotographischer Atlas mediterraner Holzer. Verlag F. Hallcr, Berlin, 1954. 8. KoBAYASHi, K. und Utusumi, N., 1951, zit. bei H. Harada (6). 9. LiESE, W.: Ber. dent, hotan. Ges. 64, 31-32 (1951). 10. — Holz als Roll- u. Werksloff U, 417-424 (1956). 1 1. LiESE, W. und Harimann/Fahnenbrock., M., Biocliitn. et Biopltys. Acta 11, 190-198 (1953). 12. Liese, W. und Johann, J., Planta 44, 269-285 (1954). XTII PAPER AND TEXTILE RESEARCH X-Ray Microscopy of Paper J. IsiNGS, Ong Sin Poen, J. B. Le Poole and G. van Nederveen Central Laboratory, TNO, Delft 1 HE examination of the structure of paper is on the whole a laborious and time-consuming work. The normal light microscopical techniques are not suited for the investigation of intact papers as most types of paper do not transmit enough light of the visible range of the spectrum. Moreover most papers have a large thickness so that the depth of focus of the objectives will be a limiting factor. Without preliminary treatment it is only possible to investi- gate the surface of the paper with the aid of a micro- scope with vertical illumination. If it is necessary to get an insight into the structure and the mutual connexion of the fibres in the paper, cross sections are required. In practice they are mostly made with a Hardy microtome, a little hand instrument of simple device. By this method sections having a thickness of 10-15 // can readily be made and in some cases even thinner sections can be obtained. It is, however, very difficult to adjust the section thickness which in most cases is rather large. Hence these sections can only be used for investiga- tions at a low magnification. Even in that case the section thickness is limiting the investigations to such an extent that little can be seen of the mutual connexion of the fibres in the three dimensional network. For closer examination it is preferable to make sections with the aid of the freezing microtome. After embedding the paper in ice or frozen gelatin sections 3-5 // thick can be made which can be studied in the phase contrast microscope (1). In order to recon- struct the three-dimensional network of the fibres in the paper a large number of successive sections must be examined. We can, however, not prevent disturbance of the paper structure by a fault of the microtome knife or by the many manipulations required to fix the sections on the slides. Even very small deflections of the knife may have such a serious effect that it is not possible to get a clear insight into the inner structure of the paper in this way. Although with these methods many good results have been obtained, it is clear that they are far from ideal. As in both techniques the cutting of the paper gives rise to disturbance of the structure, it is worth while to think of a method which makes possible to study the paper structure without any alteration of the initial situation. Sometimes a suspension of fibres is dried on a microscopical slide and the fibre conglomeration is studied. With this method some results may be obtained. It remains, however, doubtful as to how far the pictures are representative for the original structure of the paper. The connexion of the fibres in the suspension may differ from the structural connexion of the fibres in the paper. The method may be useful for the investigation of the fibrillation. Some years ago Pelgroms (2) published an article on the microradiography of paper by the contact method. In this way it is possible to get some im- pressions of the paper but one needs a thin specimen to get a good picture. If the specimen is a thick one sectioning of the paper is still necessary. In our laboratory we studied suspensions of beaten paper fibres in a permanent water containing mountant (aqua mount or glycerogel). Just as is the case with the previous method, only the fibrillation of each individual fibre can be studied, but it is impossible to get an impression as to how these fibres are linked in the original paper. None of these methods giving satisfactory results for our purpose, we proceeded to use the X-ray projection microscope, and a large number of dif- ferent papers was examined. We got interesting data on the structure of various kinds of paper, viz. the felting of the fibres and fibrils, the place of the fillers in a paper and the structure of the upper layer of coated papers. In the course of the investigation it proved necessary, however, to make stereomicrographs in order to get the most valuable results. When using non stereo- scopic micrographs it was very difficult to get a clear insight into the image, but by examining stereomicro- graphs of the same object the three-dimensional structure was clearly seen as well as the loading material in it. If paper is studied in the X-ray microscope without any pretreatment, little absorption of the X-ray takes place. Consequently the cellulose fibres give a slight contrast with the environment. Fillers on the con- trary, in particular calcium carbonate and titanium oxide, having more X-ray absorption, give a very good contrast. Hence it was possible to study very easily their position, size and shape in this way. To increase the contrast of the cellulose fibres the paper was pretreated with an alcoholic iodine solution. The heavy iodine atom of this solution was adsorbed on the fibres and absorbed the X-rays to such an extent that the cellulose fibres were clearly visible in the pictures. The alcoholic iodine solution does not swell the cellulose fibres, consequently the structure of the paper is not influenced by this treatment. It was, however, impossible to get a good impres- sion of the fibrils brought about by the beating of the fibres, unless the fibrils were stuck together forming Partial Embedding Technique for Replication of Fibres 283 a bundle with a diameter of at least half of the size of a fibre, but even then we could not clearly visua- lize the size and shape. By shadowing the paper with metals such as gold it was possible to see the fibrils themselves and the membrane-like fibrils which have already been re- vealed by the phase contrast microscope in previous investigations. With the aid of the X-ray microscope it is also possible to study coated papers and metal lamina- tions. Coated papers, art papers and machine- coated papers are provided on one or both sides with a layer composed of a pigment, e.g. CaCOg or kaolin, and a binder, e.g. starch or casein. The filler of the layer can be distinguished very easily from the cellulose fibres and the place of the pig- ments in the layer is clearly seen. In particular titanium oxide and clay gave con- trasts better than those obtained with calcium carbo- nate. There is no possibility of removing fillers by cutting cross sections with the aid of a microtome knife. This is often the case with the techniques used so far. In the microscopical slides frequently much loading substance is present outside the sections. In case of laminations the punctures in the layer appearing when the metal at that spot is too thin were seen as bright pits in a grey field. The magnifications used in these investigations are low, mostly ranging from 17 to 20 times. Higher magnifications may be applied with success e.g. when the size, shape and distribution of fillers must be examined. In that case a high resolution of the microscope is necessary. When studying the network of the fibres a low magnification is sufficient. On account of the results obtained in our investi- gations we consider the X-ray microscope a useful instrument for the study of the structural details of paper. In comparison with the methods used up to now in this field the X-ray microscopical technique shows some advantages which we can summarize as follows. (i) The three-dimensional paper structure can be examined without any disturbance of the links between fibres and fillers. (ii) The paper can be studied in its whole thickness without any refocusing of the microscope. It is possible to get an impression of the fibres and fillers situated in front as well as in the backside of the paper in one picture. The reconstruction of the structure of the paper from serial sections in the normal light microscopical techniques is not neces- sary. (iii) Stereomicrographs can be obtained in a simple way. Consequently it is easy to study the whole depth of the paper. This results from the great depth of focus of this technique as compared with the focus of the normal microscope. On account of these advantages it will be useful to improve this method. In the first place the resolution of the microscope must be at least as high as the resolution of the normal light microscope. In the second place it will be necessary to raise contrast in such a way that even the finer fibrillations are clearly visible in the image. It is recommendable to use for this purpose a chlorine zinc iodine solu- tion instead of the alcoholic iodine solution used so far. It is possible that also some use of contrast may be obtained by using different targets and adopting the target to the chemical nature of substance under investigation. This will be valuable when examining the loading substances. In the third place it is necessary that the exposure time is fairly low. References 1. VAN Nederveen, G. and Isings, J., Tappi 37, 103 (1954). 2. Pelgroms, J. D., Paper Trade J. 134, No. 1, 25 (1952). Partial Embedding Technique for Replication of Fibres J. Dlugosz British Rayon Research Association, Manchester When the surfaces of textile fibres are studied with the transmission electron microscope there is fre- quently a need for a routine method for their repli- cation since the number of different fibres submitted for examination may be considerable. As the thickness of a textile fibre is usually greater than 10 // it is difficult to make a replica which will retain, after the necessary manipulation, the geomet- rical shape of the fibre even supposing that the replica film can be freed from the fibre. For these reasons it is convenient to restrict the surface of the fibre to be replicated to such an extent that the resulting replica will be fairly Hat. This can be achieved by the partial embedding of the fibre in some medium to a controllable depth. Less than half of the fibre should emerge from the medium. A method designed to fulfil these two requirements i.e. the ability to control the depth of embedding and to embed in one preparation a large number of 284 :jr— "s^T ■ ^^L-J, Fig. I. Cellulose acetate fibre at the place where it was cut with scissors. Maenitication 3000. Fig. 2. Diatomaceous powder sprinkled on to swollen gelatin. Magnification 3000. fibres with random orientation will be described. The method has been designed so that no skilful or patient manipulation of individual fibres is re- quired. Short pieces of the fibre are dropped on to the swollen gelatin of a fixed and washed photographic plate. This may be done most conveniently by cutting lengths from the fibre or yarn while this is held above the photographic plate and allowing the pieces to fall on the gelatin. Surface tension will cause the swollen gelatin to creep up the sides of the fibres. After the gelatin has dried the fibres will be partially embedded in it. Marco Resin' to which a catalyst and an accele- rator have been added, according to the makers" formula for cold curing, is spread on a microscope slide, which is then laid on the gelatin surface con- taining the fibres, so as to form a sandwich: glass- gelatin-fibres-Marco Resin-glass. This is then left until the resin sets. At room temperature the setting time varies from 40 minutes to 3 hours depending on the accelerator content. Roughening one side of the microscope slide by grinding promotes the adhe- sion of the resin to the glass. A photographic plate slightly larger than the microscope slide is used and the slide is placed so that part of it overlaps the edge of the photographic plate. By pressing the photographic plate with thumbs and the projecting end of the slide with forefingers the assembly is mechanically separated along the interface gelatin-Marco Resin. A similar method of separating Marco Resin from gelatin has been used by Dew (1) in the production of optical replicas of diffraction gratings. It is now found that all fibres have been transferred to the resin. The flat surface of the Marco Resin — the cast of the photo- graphic gelatin — from which in numerous places fibres emerge is now replicated using a modification of the technique described earlier (2). A layer of silver about 0.1 // thick is deposited by vacuum evaporation on the resin matrix and is subsequently freed by immersion of the whole for a few minutes in chloroform. This intermediate replica is then coated with the final replicating mate- rial again by vacuum evaporation. The final replica may be of silicon monoxide or carbon but a better contrast in the image is obtained if the replica is made of gold-palladium alloy or one of the other metals used for shadow-casting which will not be attacked by nitric acid. To prevent the replica film from being damaged during the succeeding operations it is backed, while still on silver, with a thin film of nitrocellulose. The silver is then dissolved in nitric acid and the backed replica film, after washing in distilled water, is eventually collected on specimen grids in the usual way. Finally the backing film is removed by immer- sion of the specimen grid in acetone for a minute or so. References 1 . Dew, G. D., The Nat. Phys. Lab., private communication. 2. Ramanathan, N., Sikorski, J., and Woods, H. J., Biochim. Biophys. Acta 18, 323 (1955). 1 Marco Resin SB.28C obtainable from Scott Bader & Co. Ltd. A Method for the Carbon Replication of Extensive Areas of Very Irregular SinTaces, with Particular Application to the Study of Pulp Fibres, Wood, and Paper D. H. Page British Paper ami Board Inc/iisiry Research Association, St. Winifred's Laboratories, Kenley, Surrey The purpose of the research of which this work forms a part is the study of the effect on pulp fibres of the process known to the paper industry as beating. Fig. 1 shows diagramatically a typical cross- section of a beaten fibre that has been dried down onto a glass slide. During the beating process, which is one of controlled mechanical disintegration, the outermost layers of the cell wall are teased out to give the so-called fibrillation seen spilling away from the fibre onto the substrate. Our research involves an investigation of the structure of this fibrillation and the surface of the fibre from which it originated. It follows that replicas are required that are intact over large areas. Hitherto this requirement has been difficult to satisfy because of the great irregularity of the fibre surface. The scale in fig. 1 gives some indication of the height of the upper surface of the fibre above the substrate. It commonly lies between 5 and 10 microns. The thickness of a conventional replica is much smaller than this, plastic replicas being about 800 A and carbon replicas 100 A thick. For this reason a replica of such a surface is extremely fragile and usually disrupts at some point in its production. This problem is common to all workers in the field of fibre replication and has been solved by some (2, 6, 7) who partially embed their fibres in a plastic. The protruding surface is then compara- tively smooth and replicas are easily made. This technique is not permissible in our case however for much of the fine fibrillation, which may range in thickness from a few hundred to a few thousand Angstroms, would certainly be completely embedded and not subsequently replicated. It was decided therefore that a replica method was needed which, while accepting the fragility of a replica made without the aid of a partial embedding technique, would still give consistently replicas intact over large areas of the fibre and its associated fibrillation. This necessitated the development of a special tech- nique for handling these very fragile replicas to ensure a satisfactory result. Other requirements of the method for our purposes were: (a) It should be capable of resolving cellulosic microfibrils (100- 200 A thick), (b) It should be a multistage method in which the first cast could be kept intact. This cast could then be used for light microscopy and for the production of metal solid replicas for reUcction electron microscopy, a point that is considered in more detail elsewhere (3). The stages of the method can be followed from the diagram in hg. 2. A few drops of a very dilute aqueous suspension of the fibres to be replicated are placed on a clean glass microscope slide and allowed to dry at about 60 C. During the drying process the fibres are pulled into close contact with the glass by surface tension forces (a). An initial cast of the fibres is made in polymethyl methacrylate. Two methods have been found effective. One uses metha- crylate softened by chloroform, the other metha- crylate softened by its monomer. The latter method, which has been suggested by Heidenreich (4) for the replication of metal surfaces, is the more practicable and will be described here. A piece of commercial Perspex 1.5 mm thick and cut to the size of a microscope slide is allowed to stand in destabilised methacrylate monomer for one to two hours. The methacrylate sheet remains on the whole unaffected except for a thin surface layer that is softened by this treatment to a gelatinous consistency. The sheet is removed from the liquid and the surplus monomer is allowed to drain from it. It is then put down onto the microscope slide and fibres and held to them under light pressure, just sufficient to ease out any entrapped air bubbles (/?). The residual catalyst in the commercial metha- crylate polymerises the monomer rapidly at 70°C. On splitting away the microscope slide the fibres K ™™^ ^NVS 0 Cfltrmfnin^ , b CltaniiiiMinD e Fig. I. Diagram of cross-section of beaten pulp fibre dried down onto glass slide. t^^i^^^i^'>:\'c::■iy:^■^/,!■i Fig. 2. Diagram showing stages of the method. 286 D. H. PAGE Fig. 3. Carbon replica, metal-shadowed, of a beaten spruce tracheid and its associated fibrillation. Magnification 1500, inset 13,000. Fig. 4. Carbon replica, metal-shadowed, of a surface of wood tissue. The wood — Scots pine — was cut in a radial plane with a chisel prior to replication. Magnification < 1700. remain embedded in the methacrylate ic). The stripping of the fibres is sometimes dil'Hcult, cellulose tape being by no means always effective. A more successful method employs a 10 "„ solution of poly- vinyl alcohol (PVA) in water. This solution is poured onto the methacrylate cast and gradually sets to a hard film which when stripped, brings with it the embedded fibre id). In particularly stubborn cases this process can be repeated. From the negative cast so formed a robust positive replica is made in PVA. A 10 % solution of PVA in water, sufficient to form a film about 0.5 mm thick, is poured onto the cast and allowed to set (e). This replica, which is a reproduction of the original fibre lying on its glass substrate, can be separated easily from the methacrylate (/). It is preshadowed with gold, palladium, and carbon is deposited on it in the usual manner. At this point a handling technique is adopted (5) that ensures that the fragile carbon replica is mount- ed intact on its grid. Briefly this technique consists of embedding the carbon replica together with the support grid in thick plastic films which are washed away after the composite arrangement has been inserted in the specimen cap and holder. Thus the delicate replica surface is at no time exposed during the manipulation of the grid and accidental damage to the replica is avoided. The only chance of damage is during the final flow-wash (a very gentle one) with chloroform. Very thin replicas of rough surfaces can be made and mounted on coarse mesh grids with confidence. For the replication of fibres, carbon films 70 A thick (estimated by a method based on optical density measurement due to Agar (1)) have been used. "New 100" grids made by Smethurst Highlight Ltd., which have apertures about 200 microns square and whose grid bars obscure only about 13 "o of the total grid area, are suitable. Under such conditions replicas have invariably been intact over the whole area of the grid. Some of the results obtained are shown in the micrographs of figs. 3 and 4. Fig. 3 shows at a low magnification part of a spruce sulphite tracheid that has been mechanically beaten. The ribbonlike piece of fibrillation coming away from it is of interest and the indicated region is shown inset at a higher magnification. The compactness and uniformity of direction of the microfibrils strongly suggest that Pulp Fibers and Paper 287 this fibrillation originated from the middle secondary wall of the tracheid. It is evident too from this micro- graph that repeated replication from the methacry- late cast is possible without perceptible loss of resolu- tion, for the carbon replica shown was deposited on the fifr/i PVA replica to be taken from it. When the method is applied to rough surfaces in general, the procedure is exactly the same as that described for fibres. Fig. 4 shows the application of the method to the study of wood tissue. The wood — Scots pine — was cut in a radial plane with a very sharp chisel prior to replication. Parts of three cells are seen, the boundaries between which are apparent. The micrograph shows the way in which the layers of the cell wall around the bordered pits have been torn away by this rather crude cutting process expos- ing the underlying layers. Wholly intact replicas of surfaces of paper and of metal fractures have been obtained by this method but space does not permit the publication of the results here. Carbon replicas prepared in this way can be ex- amined with advantage by stereomicroscopy. There is always a tendency for the carbon replica of a rough surface to be pulled flat by surface tension forces during the final flow-wash with chloroform. However, replicas 70-100 A thick appear to retain approximately their overall form provided that height variations in them are no greater than 5 microns. (For height variations greater than this the replica collapses to some extent but without disrupting.) Stereomicrographs then are capable of yielding in- formation on both the overall form and detailed structure of the surfaces. To conclude, this method has been of value in the investigation of the structures of pulp fibres, wood and paper. It could be used with advantage for the reliable replication of large selected areas of rough surfaces in general. References 1. Agar, A. W. (1957, in press). 2. Dlugosz, J., These Proceedings, page 283. 3. Emerton, H. W., Page, D. H., and Watts, J., These Proceedings, page 287. 4. Heidenreich, R. D., Rev. Sci. Instnim. 23, 583-594 (1952). 5. Page, D. H., Research (Loiul.) 9 S, 10 11 (1956). 6. Ramanathan, N., SiKORSKi, J., and Woods, H. J., Biochim. Biophys. Acta 18, 323-340 (1955). 7. Takahashi, N. and Asaeda, T., Bull. hist, textile Fiance 35, 75-82 (1952). Further Reflection Electron Microscopy of Pulp Fibres and Paper H. W. Emerton, D. H. Page, and J. Watts British Paper and Board Industry Research Association, St. Winifred's Laboratories, Kenlcy, Surrey roLLOwiNG the revival of interest in the reflection electron microscope in the early years of this decade. Chapman and Menter (3) used it in the study of fibre surfaces. Shortly afterwards Emerton (4) at this laboratory proposed its use as a contribution to the study of the effect on pulp fibres of the paper- making process known as beating. The early work in this field has been reported by Amboss, Emerton & Watts (1). It will be recalled from this latter paper that the reflection electron microscope as used for the study of pulp fibres suffers from two main disadvantages. Firstly the specimen is viewed at glancing incidence giving rise to a severely foreshort- ened image. Secondly under the impact of the elec- tron beam the fibre tends to decompose into gases which cause distension of the cell walls. This "bubble artefact" tended to mask the true structure of the fibre surface and was an undesirable feature of most of the early reflection electron microscopy of pulp fibres, in spite of all efforts to keep the beam inten- sity low during examination. It was in an attempt to overcome this difficulty for biological specimens in general that Bradley (2) devised a method of producing solid metal replicas of specimens. Such replicas will withstand high beam intensities without damage. The present paper describes our experience of the application of this technique to the study of pulp fibres and paper. Apart from the initial preparation of the fibres our technique follows basically that described by Bradley. The fibres are dried down from an aqueous suspen- sion onto a glass slide and replicated in thick plastic from which the robust metal replica is made. Figs. I and 2 illustrate the application of the technique to spruce tracheids that have undergone a fairly heavy beating treatment. Both these micrographs, but more particularly tig. I. which is a micrographic montage, have been reduced far below their useful magnifica- tion to be accommodated on the page. The striking three-dimensional appearance given by the reflection method is vividly brought out. It is this three- dimensional aspect, arising from the great depth of field, the oblique viewpoint and the shadows pro- duced, that permits the form of the fibre surface to be inferred. Both tracheids have been subjected to strong surface tension forces while they were dried to equilibrium with room humidity and this has led to the collapse of the lumen and an almost complete flattening of the cell. This effect, which is common in fibres of this type (i.e. spruce sulphite tracheids). 288 H. W. EMERTON, D. H. PAGE AND J. WATTS All the micrographs are of solid replicas. The scale ellipses are 25 microns in diameter. The angles of illumination and viewing are 5" and 12 respectively. Fig. 1. Reflection electron micrographic montage (considerably reduced) of a beaten spruce sulphite tracheid. Fig. 2. Reflection electron micrograph of a thin-walled spruce sulphite tracheid, heavily beaten. Fig. 3. Reflection electron micrograph of the surface of a tissue paper made from hemp and flax fibres. is not directly apparent from examination in the light microscope which gives only a plan view. The form of the twist in the middle of the tracheid in fig. 1 would also be difficult to appreciate from examination in the light microscope. In both micrographs there is evidence of promi- nent, more or less transverse, fibrils on the surface of the tracheids. Care is needed in interpreting these. In the first place the image is foreshortened. (An impression of this foreshortening is given by the scale ellipse (5) which must be thought of as a circle lying in the plane of the substrate.) A linear feature inclined to the plane containing the line of sight and the normal to the substrate is imaged at an angle exceeding the true value (8) and hence a fibril lying in this plane appears more nearly transverse. Further- more, the width of the fibrils cannot be readily as- sessed for the contrast scheme does not permit meas- urements in this direction to be made with accuracy because of the foreshortening and loss of informa- tion in shadow. The apparently transverse fibrils in these micrographs are, however, thought to be part of the outer secondary wall that has been modified by the sulphite digestion to which these cells have been subjected but further work is required to establish this. The only common artefact arising from the use of solid replicas for the examination of pulp fibres is the incomplete replication, in some cases, of fibrils that come away taut from the upper surface of the fibre. Such fibrils, examples of which may be seen in fig. 1, are evidently completely embedded in the plastic matrix that constitutes the first stage of the replica process, break during the stripping of the fibre and are consequently not entirely replicated. We have also investigated the possibility of ap- plying reflection electron microscopy to the study of paper surfaces. Fig. 3 shows a replica of the surface of a tissue paper made for use as a dielectric in con- densers. Notwithstanding the very heavy and pro- longed beating to which the fibres are subjected to produce this type of paper — a treatment which is said to reduce the pulp to a gel — it is clear that some of the fibres have retained their general integ- rity. This micrograph in particular brings out one of the Pulp Fibers and Paper 189 main disadvantages of reflection microscopy as used by us at present. Much of the image is lost either in shadow or in ground hidden from view, a situation that is aggravated when rougher surfaces are studied. This difficuUy and the disadvantage of a severely foreshortened image could be overcome by the use of higher angles of illumination and viewing. Two factors however have limited our use of the reflection method to low angles. In the first place, the greater the angle of deviation of the beam the poorer the resolution. This appears to be due to the greater spread of velocities of electrons scattered at high angles. It has been shown (7) that this velocity spread is due to the effect of contamination and that provided that the contamination is suppressed by subjecting the specimen to ionic bombardment while the electron beam impinges on it, quite a high reso- lution can be obtained when metal specimens are examined using angles of deviation as high as 25 \ Secondly we were restricted in our early work to relatively low angles of deviation by the rapid fall ofT of intensity of the scattered beam at high angles. The increased intensity of the incident beam required to meet this aggravated the problem of beam dam- age. The angle of deviation of 17 that we finally adopted (and which appreciably exceeds that gener- ally used) was a compromise between the conflicting requirements of high resolution and minimum beam damage on the one hand and small shadow areas and a low foreshortening on the other. However, with the advent of inert solid replicas that are capable of withstanding high beam intensities, it appeared worthwhile investigating whether a useful image could be obtained at high angles even though we have no facilities at the moment for ion bombard- ment. With an angle of deviation of 28 ' the image obtained was rather poor. This may be due in part to the excessive chromatic aberration when the objec- tive lens is operated at long focal length, a necessary condition for high angle reflection work with the Met-Vick EM 3 microscope without modifying it considerably. Further work is needed to determine whether a resolution acceptable for our purpose can be obtained by modifying the microscope so that the focal length can be reduced or whether the resolu- tion is severely limited by the effect of contamination. It remains to be seen whether the scanning electron microscope (9, II) will give a satisfactory solution. It is now possible to assess the advantages and disadvantages of the reflection method more clearly. We believe the method is of some value in this field if used as a method complementary to other techniques of microscopy. It has been shown that the plastic matrix from which the metal solid replicas are made can be used indestructibly both for light microscopy (6), and for the production of carbon replicas for transmission electron microscopy (10). It is therefore possible to obtain micrographs of the same ret^ion of an individual fibre as follows: — (a) a light micrograph showing a large area of fibre but with limited resolution; (/)) a reflection electron micrograph with its oblique viewpoint and remarkable three di- mensional appearance; and ((■) transmission electron micrographs of selected areas at a magnification sufficient to resolve ccllulosic microfibrils (100 200 A thick). When three such images are considered the com- plementary information from each should enable the finest structural detail to be related to the fibre as a whole in a way not previously possible. The chief drawbacks of the reflection method as described here then are the severe foreshortening and the loss of information in shadow and "dead ground"". The reflection method does not permit angular or linear measurements to be made accu- rately with the exception of heights which can some- times be determined with a fair degree of precision. On the other hand when used in conjunction with complementary methods of microscopy the three- dimensional aspect of its image, its sensitivity to height, and its great depth of field which permits montages of considerable lengths of a fibre to be made economically are all of value in revealing the external form of the specimen. References 1. Amboss, K., Emerton, H. W., and Watts, J., Proceed- ings of the International Conference on Electron Microscopy, London 1954, Publ. by Roy. Micro- scop. Soc, London. 560-564 (1956). 2. Bradley, D. E., Brit. J. Appl. Phys. 6. 191-195 (1955). 3. Chapman, J. A. and Menter, J. W., Proc. Roy. Soc. A 116, 400-407 (1954). 4. Emerton, H. W., /. Roy. Micro.scop. Soc, Ser. 3, 74, 35- 41 (1954). 5. — Re.searcli {LoiuL), 7, S56 (1954). 6. Emerton, H. W., Page, D. H., and Watts, J., Brit. Paper and Board Maker.s' Assoc. Proc. Tech. Sect. 37, 105-125 (1956). 7. Fert, C, Marii, B.. and Saporte, R., Compt. rend. accul. sci. (Paris) 240, 1975-1978 (1955). 8. Menter, J. W., /. Inst. Metals 81, 163-167 (1952). 9. McMui lEN, D., Proc. Inst. Elec. En^rs II 100. 245-259 (1953). 10. Page, D. H., These Proceedings, page 285. I 1. Smieh, K. C. a. and Oateey, C. W., Brit. J. Appl. Phys. 6, 391-399 (1955). 19 - 568204 Electron Microscopy Application of Ultra-Microtomy to the Fine Structure Study of Rayon Viscose Fibre' P. Kassenbeck Iiistitut Textile de Fiance, Paris Several papers have already been delivered on the subject of electron microscopy studies of the fine structure of regenerated cellulose fibres (15, 17, 18). All these papers were on the interpretation of E.M. obtained by replica methods. May we recall briefly that we have been able to show with this method, modifications of the surface of fibres accord- ing to the treatment they were subjected to and also the presence on the surface of rayon viscose fibres of an extremely thin layer. Thestructurecharacteristics of this layer differ greatly from the remaining mass of the fibre. The application of ultra-microtomy to the study of the sub-microscopic architecture of these fibres confirms the results obtained by previous studies. Furthermore it shows details of internal structure which could not be seen by replica methods. The problem of embedding textile fibres for ultra- microtomy applications can be summarized as follows: 1. It is necessary to obtain penetration and fixation of the embedding medium into the fibre without any change of micro-structure. 2. The cutting angle must be perfectly known if sche- matic reconstitution of the fibre structure is required. For this reason fibres must be oriented in the embedding medium. Experience shows that simple immersion of dehydrated rayon viscose fibres in methacrylate monomer is not sufficient to obtain a proper embedding. We overcame this difficulty by the use of an embedding method which we call "double cut technic". Our starting material is about 0.5 cm^ of fibre sections which are cut on an ordinary microtome. The thickness of Fig. 2. Thermal advance microtome. the cross-sections varies from 30 to 60 /< for fibres 20 /< in diameter. These cross-sections are swollen in water and swelling is fixed by the zinc chloroiodine reagent. They are dehydrated in alcohol and immersed two days in a first mixture of destabilized methyl methacrylate and butyl methacrylate 50/50. Polymerization is carried out in a second mixture of the same composition in a special cell which allows the orientation of the cross- sections (3, 10, II 12). A perspex tube, 10 mm long, containing the monomer and the cross-sections is mounted between the two electrodes of the cell (fig. 1 ). A 50 cycles alternative current of 3 to 4 KV is applied between the electrodes. The whole cell is rocked one cycle per second with an 180 angle. There cannot be any decantation of the cross-sections towards the bottom of the perspex tube. Rocking insures good dispersion of these cross-sections into the embedding medium. Under the elTect of the electric field, the fibre sections are oriented while the polymerization takes place at 40"C for 6 hours. Cross sectioning on a thermal advance microtome. — After polymerization is completed, we have at our dispo- sal a cylindrical perspex block in which all cross-sections are oriented parallel to the cylinder axis. This block is cut with a thermal advance microtome (fig. 2) (2, 5). The cross-sections are cut again in a much thinner section (100 to 300 A thick). A small brass container (7, 8) is fixed with the help of paraffin to a glass knife (13). The brass container is connected to a glass tube by rubber tubing. By these means the level of the liquid (water-acetone mixture) is ajusted correctly. The ribbon of sections is picked up on formvar film coated grids. Fig. 3 shows a result obtained when the embedding medium has not been removed. Contrast in this micro- Fig. 1. Perspex cell for the embedding of sections of fibres under tension (scale 1:1). ^ A more detailed report lias appeared in Bulletin Institiit Textile de France 61, 7-15 (1956). The Fine Structure Study of Rayon Viscose Fibre 291 Fig. 3. Ulira-ihiii section of rayon viscose fibre without dissolution ortiie embedding medium. Part of a cross-section. Magnification 11,000. graphs is poor but the micro-structure of the fibre is preserved unchanged. It is only the elements of structure which differs sensibly in density from the embedding medium which are visible. The perspex which constitutes the embedding of the cross-sections is dissolved by chloroform in a distillating apparatus (1). The cross-sections are then shadowed with chromium by vacuum evaporation. This method gives micrographs with better contrast, but the flow of the solvent, during dissolution, may dislocate the fine structure and modify certain morpho- logical details which may disappear even completely. It is good practice to use simultaneously both methods and to judge the results by comparison of these difierent micrographs. Before any interpretation of the micrographs, we must have an idea of the scale at which the ultra- thin sections are examined under the electron micro- scope and also of the geometry of the sections. Fig. 3 is the ultra-thin transversal section of a fragment of rayon viscose fibre. The separation be- tween the "skin" and "core" is characterized by a crown of microvacuoles. Their mean diameter is between 2000 and 3000 A. The skin of the fibre is made of several concentric successive layers of dif- ferent densities. The first layer or cuticle which is located on the fibre surface possesses the highest density. Its thickness varies from 500 to 1000 A. Then comes an apparently empty layer, its density does not seem different from the density of the embedding medium. For lack of contrast details of structure cannot therefore be distinguished in this layer. A third layer of higher density appears before the crown of vacuoles is reached. The total thickness of the skin is about 2 //. The core of the fibre shows a spongy structure. Its density is homogeneously distributed. Large vacuoles can be located in the core. It is these large vacuoles which give the well-known milky appear- ance of some types of rayon viscose libres; they are several microns in diameter and can be seen under an optical microscope. The longitudinal sections show a succession of 7 layers of dillcrent densities in the skin of this par- ticular rayon viscose fibre. In oblique sections of rayon viscose fibres con- centric layers can be clearly distinguished. Elod (4) and Horio. Kobayashi and Kondo (9) have already mentioned the presence of such layers. After dissolution of the embedding medium and shadowing, the spongy net-work composed of a very great number of microvacuoles becomes visible in the regions of the skin which appear empty before dissolution of the embedding medium. The difference of density between skin and core is clearly visible in both phase contrast microscopy and in electron microscopy, it is therefore possible to get information on density gradient by simple examination in phase contrast microscopy without the use of electron microscopy. For this examination thick sections made by hand with a razor blade may be thin enough. References 1. Barnfs, R. B., Borton, C. J., and Scorr, R. C, /. Appl. Pliys. 16, 730-739 (1945). 2. Bernhard, W., Compt. rend, du coiioque du C.N.R.S. Toulouse, 1955. 3. Cotton and Mouton, /. Chim. Phys. 4, 365 (1906). 4. Elod, E. and Rauch, H., MelllamI Textilhcr. 1, 1-11 (1953). 5. Gautier, a.. Bull, inicroscop. appl. (1) 5, 1-2 (1955). 6. Hermans, P. H., Textile Research J. XX, 8, 553-569 (1950). 7. HiLLiER, J. and Gettner, M. E., J. Appl. Phys. 21, 889- 895 (1950). 8. — Science 112, 520-523 (1950). 9. HoRio, M., Kobayashi, K., and Kondo, T., Textile Research J. XVll, 5, 264-280(1947). 10. IsARD, J. O., Brit. J. Appl. Phys. 6, 5, 176 (1955). 11. IwANOW, N. and Schneider, R., Bull. I.T.F. 56, 45 (1955). 12. Larose, p., J. Textile Inst. 42, 104 (1951). 13. Latta, H. and Hartmann, J. F., Proc. Soc. Exptl. Biol. Med. 74, 436-439 (1950). 14. MoREHEAD, F. F. and Si.ssoN, W. A., Textile Research J. 15,443 (1945). 15. Kassenbeck, p., Bull. I.T.F. 43, 43-52 (1953). 16. — Teintex, 531-540 (1955). 17. — Ann. Scient. Text. Beiges 1, 176-194 (1956). 18. — Bull. I.T.F. 59, 7 (1956). Structural Details of Natural Fibers as Observed with the Electron Microscope C. Maertens, G. Raes, and G. Vandermeerssche Laboiatoire de Technologie des Matieres Textiles de /' Uiiiversite de Gaiid and Centre de Microscopie Electionique, Medical, Industrie! et Agricole. Brii.xelles-Uccle Most studies with the electron microscope on the fine structure of natural fibers have been limited to the suspension and to the replica techniques. This is mainly due to the fact that cellulose fibers are very hard to cut with the usual glass knives and that no quite appropriate staining techniques for cellulose have been found. At the Conference of Barcelona in 1954 (4) and at the Conference of Brussels in 1955 (2) we have mentioned already our first attempts of making ultra-thin sections of cotton fibers. Several staining methods had been tried and three of those had given good results: osmic acid, ferri- alum and iodine zinc chloride. A still better specific electron staining has been found by Saara Asunmaa ( 1 ) who has made ultra- thin sections of cellulose fibers after treatment with thallous ethylate. The contrast obtained by this method is excellent but unfortunately the method of staining is quite long and elaborate. We have been able to obtain equally well stained sections of cellulose fibers by a very simple treatment with silver nitrate. The fibers under investigation were dipped into a 20 % silver nitrate solution for a couple of hours and then washed, dehydrated and embedded in the usual way as described in our previous work (2). A mixture of 6/4 butyl-methyl-methacrylate has been found most suitable. The microtome used was the Porter Blum model and the sections obtained were on the average about 250 A thick. The electron microscope used was the Philips type EM 75. Cotton fibers of different types and different origin * k«>.*#» , . ■■■» have been tried. Also flax fibers of different types and at different stages of its treatment in the hackling machine have been investigated. The reason for trying these many types of cellulose fibers is to check in how far a particular structural aspect can be considered as a possible artefact or as a structure due to the drying out of the fresh fibers. When our first cotton sections were shown (2) the question arose as to what extent certain struc- tural details could have been considered as artefacts. Several of the cotton sections showed indeed a kind of "detachment"' of some of the layers of which the secondary wall is built up. This "detachment" which had never been observed before, was called a "kind of secondary lumen" or "artificial lumen". This particular aspect of certain cotton fibers is of the greatest importance because it might help in explaining certain discrepancies which exist in meas- urements done in different laboratories with the arealometer (of Dr. Hertel). Fig. 1 shows such "detachments" in cotton fibers which have been fixed and embedded in the same way as described earlier (2). We have therefore made comparisons of a whole series of fixation and staining methods in order to improve our previous method. From these tests we may conclude that almost all the fixatives used gave almost identical results in so far as the shape of the fibers is concerned. As regards the contrast, we have found that the treat- ment with a silver nitrate solution gave by far the best result. '♦* '* m . •>. , - •• y Fig. 1. Transverse section of cotton fibers, fixed in OsOj. Magnification 4000. Fig. 2. Transverse section of flax-fiber. AgNOg fixation. Magnification 4000. Fig. 3. Longitudinal section of flax fiber. AgNOj fixation. Magnification 12,000. Contact Region hcrwccn Two Fibres 293 In some sections obtained from fibers which were fixed with iodium zinc chloride a swelhng elTect was observed and there the detachment could have been explained as being due to a compression by the glass- knife during sectioning. A whole series of different cotton fibers and llax fibers has been examined and from our observations it seems that the above-mentioned detachment is not due to the preparation or sectioning techniques but corresponds to the actual state of the fibers before examination. This kind of detachment described for cotton fibers can be found in some rare cases also in flax fibers. Our new fixation methods have allowed us to make higher magnification micrographs. The hazy dots representing isolated cellulose fibrils have made place for rather sharp and well-defined dots, as can be seen in fig. 2. Fig. 3 is an example of higher magnification of a longitudinal section. It confirms our idea regarding the existence of a detachment of a certain number of layers in the secondary wall. This detachment is probably a result of the drying out of the natural fibers, long before the actual examination in the laboratory. REFERENCtS 1. AsuNMAA, Saaka, Svt'ii.sk I'lippcisticbi. 10, 1 (1954). 2. Castiaux, p., Raes, G., and Vandermeerssche, G., Ann. Textiles 2, 29 (1956). 3. Di: MtuiEMEESTER, D., Rais, G., and Franzfn, T., Ann. Textiles 2,, 260 (1955). 4. VANDFRMErRSSCHF, G. and Raes, G., Conf. Int. Teen. Textil. Barcelona, 1954. Contact Region between Two Fibres Saara Asunmaa Paper Technology Department, Swedish Forest Products Research Laboratory, Stockholm Since new sectioning methods have been developed (9, 10) the fine structure of cellulose fibres has been investigated in ultrathin sections. Chemical reactions and a metal impregnation have been used as contrast treatments. Thallation of the fibre sample gives a sufficient contrast (3, 4, 5). The reaction of thallous ethylate with the fibre material furthermore only applies to the accessible hydroxyl groups of the cellulose fibres. In sections of fibres of holocellulose of Swedish spruce (Picea excelsa), for example, a parallel fibrillation was observed in the main sec- ondary wall, but no parallel striation in the same direction in its outermost part. The fibrils observed have small dimensions, a width of down to 50 A. They presumably correspond to the "fibril strings" observed in hydrolyzed materials (7, 8, 11). Another arrangement of similar "fibril strings" was observed in different kinds of cell walls (4, 5). These "fibril strings" are different from micro fibrils, which earlier were observed in swollen fibres and in many other botanical materials and which are several hundred A wide. Electron micrographs of thallated fibres depict the reacted thallous cellulosate and the excess reagent as an impregnation medium, if present. The reactivity of the fibre material under the conditions of thalla- tion therefore only describes one special kind of fibre reactions. Of particular interest from the technical point of view is the fine structure of water swollen fibres. Therefore a method for impregnation of fibres in water-swollen state was developed (5, 6). The water solution of a metal compound was made to penetrate the cell wall, it was reduced in situ and the metal was enclosed in the fibre structure by means of an effective drying. The metal content observed in the micrographs demonstrates an "origi- nal contrast", i.e. the contrast material is introduced without chemical reactions and before the prepara- tion and embedding of the material for sectioning. This kind of contrast treatment is suitable for com- pact materials like cellulose fibres and wood cells. As a consequence of very high metal content in several parts of the fibre wall, the fine structure can be studied only in micrographs of the cellulose rich areas with a low metal content. Such a situation is observed in fibre walls of holocellulose of spruce after some processing procedures, for example, after a hot alkali treatment. Electron micrographs of hot alkalized metal- impregnated fibre walls show morphological changes that appeared in the holocellulose during the treat- ment. Periods of broken, metal-filled canals are observed both in longitudinal and in cross sections. The number and shape of the metal-filled cavities in the fibre section before and after the chemical treat- ment show the distribution of the attacked parts and probably the distribution of some reactive parts in the original holocellulose fibre. The original holo- cellulose fibre does not usually show similar canals, or not to such an extent as the material after pro- cessing. In electron micrographs at good resolution a fibrillation is to be seen in the areas between the metal-rich canals (5). The fibrils observed presumably 294 SAARA ASUNMAA Fig. 1. Ultrathin section through a contact region between two perpendicularly arranged fibres in a sheet of Husum sulphate pulp ol' Piniis silvestris. Arrows point to the con- tact regions. Magnification • 100,000. Inset, magnitication X 140,000. bonded fibres, not only an arrangement of contact points. The contact region shows a tine structure; struc- tural components with a width of 50-100 A ("fibril strings") run between two adjacent fibre bodies. Particularly in micrographs of a contact between two perpendicularly arranged fibres the origin of the "fibril strings'" observed can be estimated. Figure 1 shows detail magnifications of a contact region between two perpendicularly arranged fibres in a paper sheet of Husum sulphate pulp of Piniis silvestris with a degree of beating 62 SR, 16,000 turns PFI. The electron optical magnification 25,000 diameters was used, an RCA EMU 2c electron micro- scope was employed. "Fibril strings" can be followed in the plane of the section, in the inset lengths of 2.5 cm, i.e. about 0,1 // can be measured. Several "fibril strings" are marked with dark lines. The strings at the left belong to the fibre I with a direction of fibre axis — *, the strings at the right to the fibre II with a direction of fibre axis f . The strings are observed to run parallel for distances of several hundred Angstrom. Thus the concept of fibre-to-fibre bonding has been brought one step further, to bonding between structural components of the fibre, which are only 50-100 A wide. Distances between such structural components are small enough to bring the concept of fibre-to-fibre bonding into the realm of chemical bonding. correspond to the "cellulose strings" or "fibril strings" observed in suspensions of hydrolyzed fi- bres and In the thallated undamaged fibre walls. The different layers of the fibre wall often show different impregnation qualities. In hot alkalized fibres the main secondary wall Is heavily impreg- nated, but Its outermost part shows a very low photo- graphic density (6). In beaten, metal-impregnated fibres of sulphate pulp of Pinus silvestris the outer secondary wall shows a higher metal content than the main fibre. The outer parts of the fibre wall are damaged during the mechanical treatment and are easily impregnated in water solutions (2). The metal Impregnation consequently gives a morphological analysis of the layers of the fibre wall in electron micrographs of the ultrathin fibre sections. An un- evenness of the fibre surface in beaten fibres, as clearly demonstrated in the electron micrographs taken by means of the reflection Instrument ( 1 ), is to be seen as a transit area in the sections. The contact region between two fibres in a fibre couple of a paper sheet can be studied in ultrathin sections of metal-impregnated samples. The real contact area as observed in the individual ultrathin sections Is often small and shows a length of the order of 0,1 /< up to 1 /<. According to serial sections there is often a contact line between the The author wishes to thank Dr. O. Andersson and Prof. B. Steenberg for the criticism of the manuscript and for many valuable discussions. She is very indebted to Dr. F. S. Sjostrand, Associate Professor of Anatomy, for facilities for using the electron microscope at the Laboratory for Biological Ultrastructure Research, Department of Anatomy, Karolinska Institutet, Stock- holm. References 1. Amboss, K., Emerton, H. W., and Watts, J., Brit. Paper and Board Makers' Assoc, Proc. Tech. Sect. 35, 487 (1954). 2. Andersson, O., Asunmaa, S., and Steenberg, B.. Svensk Papperstidn. (1957, to be published). 3. Asunmaa, S., Svensk Papperstidn. 57, 367 (1954). 4. — ibid. 58, 33 (1955). 5. — ibid. 59, 527 (1956). 6. — Proc. Conference on Textile and other Industrial Fibres, Leeds. Electron Microscopy Group, Institute of Physics (1957, in press). 7. Munkherjee, S. M., Sikorski, J., and Woods, H. J., /. Te.xtile Inst. 43, T 196 (1952). 8. Ranby, B., Fine Structure and Reactions of Native Cellulose. Dissertation Uppsala, Stockholm, 1952. 9. Sjostrand, F. S., E.xperientia 9, 68, 114 (1953). 10. — /// Oster and Pollister. Physical Techniques in Biological Research, Vol. III. Acad. Press Inc., New York, 1956. 11. VoGEL, A., Makronwiekiilare Cheniie IL 111 (1953). Die elektronenmikroskopische Darstellung groBer Perindcn in Celkilosefasern durch Einlagerung schwerer Alome K. Hess Lahoiaioi iiiiii J'iir Melil- iind Eiweififoiachiiii};, Ha/uiovcr Fasern aus Kunststoffcn (Polyamide, Polyester, Polyurethane, Polyathylene und Dcrivate) zeigen nach Hess und Kiessig (6) im Rontgenfaserdia- gramm Kleinwinkelreflexe auf dem Meridian, die Perioden in der Faserlangsrichtung zwischen 80 und etwa 200 A entsprechen (Abb. I). Die Erscheinung fand ihre begrundete Deutung in einer periodischen Wechselfolge gittergeordneter und gitterungeordne- ter Abschnitte in Faserlangsriciitung, wie sie sche- matisch in Abb. 2 dargestellt ist. Bei kiinstlichen Fiiden aus Polyvinylalii.. Fig. 1. Section througii a cell wail of date endosperm in its natural state. Magnification 5000. Fig. 2. Section through a swollen cell wall of date endosperm. Note the microfibrils between the swollen layers. Magnification 20,000. Fig. 3. Date endosperm after Turmix and ultrasonic treatment before extraction of mannan A. Magnification 25,000. Fig. 4. Date endosperm after Turmix and ultrasonic treatment after extraction of mannan A. Magnification 25,000. endosperm of palm seeds there is available a material which contains mannan, one of the most important hemicelluloses. in a more or less pure form. An attempt has therefore been made to gain some knowledge of the submicroscopic structure of man- nans by studying the endosperm of date seeds {Phoenix dactylifera L.) and ivory nuts (Phytelephas macrocarpa Ruiz et Pav.). Ludtke (4) has shown that there are two different mannans in the cell walls of ivory nut, i.e. mannans A and B which differ from each other by their solubility in dilute alkali. The chemical structure of the two mannans has been studied by Klages (2, 3) and recently by Aspinall, Hirst and coworkers ( I ). According to them these mannans are both com- posed of two types of molecules, one having a mannopyranose residue as non-reducing end group, the other a galactopyranose residue. The two man- nans should, according to Aspinall, differ from each other only in their molecular size. By determination of the end groups Aspinall has calculated a degree of polymerization of 10-13 for mannan A and of 39^0 for mannan B. It might however seem doubtful that the difference between the two mannans should lie in their molecular size only. In that case it would be very difficult to understand why there should not exist between the mannans A and B a series of molecules with a degree of polymerization between 10 and 40. Moreover, this small difference in chain length alone can hardly explain the very distinct difference in the alkali solubility of the two man- nans. In polarized light the cell walls of date and ivory nut endosperm are highly birefringent. It does not seem probable that this birefringence is caused only by the 6 "„ cellulose (total carboh\drate content = 100 "o) that is found in the walls. Therefore the man- nan forming about 90 "o of the cell walls must be at least partly crystalline. With the aid of the polariz- ing microscope the direction of the crystallites in the thick central layer of the walls can be easily deter- mined. In the date cells they run more or less per- pendicularly to the cell axis and surround the pits circularlv. 300 M. W. ANDREWS AND J. SIKORSKI In an ultrathin section through a cell wall of date endosperm the whole wall appears to be built up of small grains (fig. I ). However, if by chance a section through a swollen cell wall is obtained, microfibrils are visible between the swollen layers of the wall (fig. 2). It does not seem likely that they are of a cellulosic nature, since, as Reiss (5) has found, the whole thick central layer of the wall disappears when the seeds germinate. To obtain an answer to the question whether there exists a morhological difference between the man- nans A and B, the finely ground endosperm of date seeds was extracted with acetone and ether, macer- ated with acetic acid/hydrogen peroxide (1:1) and treated with a defibrator of the Turmix type and/or with ultrasonic waves. This material was found under the electron microscope to consist of two morpho- logically different components. There are small grains visible which are often aggregated to greater parti- cles and there is also about the same quantity of microfibrils (fig. 3). However, the microfibrils are mostly covered by the grains and therefore appear to be present in a smaller quantity. Chemical analysis of the sample yielded, after hydrolysis, 15 "„ glucose with traces of galactose and 85 °o mannose. The mannan A was extracted from this sample with 7 % potassium hydroxide at room temperature. This specimen free of mannan A appeared in the electron microscope to consist of microfibrils only. The small grains and their aggregates had disappea- red almost completely (fig. 4). It follows that the grains visible in fig. 3 are identical with the low molecular weight mannan A. If the latter is precipi- tated again from the extract it forms nearly cubic particles with a diameter of about 1 //. An ultrathin section through such a particle shows that it is built up of small grains very similar to those of the mannan A in its native state. The specimen that is shown in fig. 4 gave on paperchromatographic analysis 39 % glucose and 61 % mannose. Results analogous to those from date cells have been obtained with ivory nuts. After extraction with 7 % potassium hydroxide an ivory nut sample yielded on chemical analysis 20 °o cellu- lose and 80 "o mannan and also appeared in the electron microscope to consist only of microfibrils. This provides evidence that the high molecular weight mannan B is built up of microfibrils analogous to those of cellulose. Their width also seems to be of the same size. At present it is difficult to say whether there are pure mannan or only mixed mannan/cellulose microfibrils. However, the latter might be less probable because the mannan B can, after prehydrolysis of the material with 1 % hydro- chloric acid, be completely extracted with 24°o potas- sium hydroxide, whilst the cellulose is resistant to this treatment. This would hardly be possible if there were mixed crystallites of 80 % mannan and 20 % cellulose. It might be noted that without prehydrolysis mannan B cannot be extracted with 24 "o potassium hydroxide at room temperature. A sample that was first extracted with 7 ^o, then with 14 "o, 18 % and finally during three days with 24 % potassium hydroxide still yielded 77 '^o mannan. The conclusion of all these observations can only be that the mannan B builds up the framework of the cell walls of palm seeds in the form of microfibrils quite analogous to the cellulose in other plant cell walls. The mannan A on the contrary lies incrusted within the framework of the mannan B. This might possibly be an indication that the resistant hemicelluloses of woody fibers also form microfibrils. References 1. AspiNALL, G. O., Hirst, E. L., Percival, E. G. V., and Williamson, I. R., /. Chem. Soc. 3184 (1953). 2. Klages, P.. Liebigs Ann. Chem. 509, 159 (1934). 3. — Ibid. 512, 185 (1934). 4. LuDTKE, M., Liebigs Ann. Chem. 456, 201 (1927). 5. Reiss, R.,j9e/-. 22, 609(1889). A Contribution to the Structure of Keratin M. W. Andrews and J. Sikorski Textile Physics Laboratory, Department of Textile Industries, University of Leeds In spite of great advances in the study of keratin structure, during the last quarter of a century or so, it is, nevertheless, evident that the information available is still very incomplete. Thus, at the molecular level the difficulties to establish a convincing relation between the results of chemical analyses and x-ray diffraction data, are as many as they are formidable (I, 2, 9. II) and there can be little doubt that the ambiguity (except in the case of cortical cells: Woods (15)) about the individual effect of various histological components of keratin fibres, on the x-ray diffraction pattern, could be considered as, at least, a contributory factor in this respect. Recent discoveries of the bi- lateral structure in crimpy wool fibres (8, 10) call, among other questions, for the consideration of The Structure of Keratin 301 P I ^'^ >« K, -^-,*«*^ ^^^^HpPpr^ ^^^••'f*' ^ ...^t0^ '^itfl^^^^l ^^^^H ^Er^-^-""^ ^^i#^\^^iM^^^I ^^^^H ^H ^■i -^- * Fig. 1. Cortical ceil isolated from Australian merino wool and exposed to ultrasonic vibrations, in uater. for 12 hours. Magnification 10,000. Fig. 2. Cortical cell isolated from Lincoln wool and exposed to ultrasonic vibrations, in water, lor 12 hours. Magnification 29,000. Fig. 3. Cortical cell isolated from Lincoln wool and exposed to ultrasonic vibrations, in lithium bromide solution, for 4 hours. Magnification 21,000. Fig. 4. Fragment of cortical cell isolated from Lincoln wool and exposed to ultrasonic \ibrations, in lithium bromide solution, for 4 hours. Magnification 32,000. differences in the type of packing of the microfibrils in cortical cells of the two regions: ortho- and para- cortex. The type of lateral packing of the micro- fibrils must influence many properties of keratin fibres. We have discussed this problem in some detail in connection with our proposed mechanism of supercontraction in keratin (7) and have further suggested (14) that keratinization involves the stabi- lization of microfibrillar texture leading to the uniplanar cross-linking together of the microfibrils to form sheets. However, Fraser (4) and, independently, Rogers (private communication) suggested that the width ( 1 to 3 microns) of the sheets of microfibrils obtained by us (6, 7) may be rather due to the mode of bio- synthesis of macrofibrils; namely some "early" mi- crofibrils aggregating laterally, and their continuing to be synthesized in scroll-like fashion to form macro-fibrils. They, in fact, suggested that the chemical treatment of Jeffrey et al. (6, 7) was capable of reversing this underlying organization in the fully hardened keratin. This divergence o'i views ma\ be accounted for by the differences in chemical reactivity between the cortical cells from crimpy and straight animal fibres; a view confirmed recently by Satlow and Sikorski (13) who found that more resistant cortical cells show a more complex structure. This lends support to the observation of Rogers (12) that the intact cortical cells were apparent in the para-cortex of crimpy fully hardened and oxidi/cd keratin fibres, whereas macrofibrils were evident in the ortho-cortex. it must be admitted that the thickness of micro- fibrillar sheets observed in our experiments (6) can account only for a relatively small fraction of the diameter of cortical cells. We have, therefore, de- cided to re-examine this question in some detail and, in our experiments, used cortical cells which have been separated with the minimum of damage. How- 302 M. W. ANDREWS AND J. SIKORSKI ever, the cortical cells isolated from untreated keratin by mild digestion in trypsin (3) appeared to be little affected by the prolonged action of ultrasonic vibra- tions (25 Kc/s and 15 Watt/cm-, actual output) in water or various swelling agents: the separated fragments showed very little internal structure (6). It was thus advisable to increase the time of treat- ment in trypsin to obtain better retting effects. The experiments were carried out with non-medullated root-ends of Australian merino and Lincoln wool fibres. These were cleaned by extraction in petroleum ether at room temperature for about a week, followed by Soxhiet extraction first in benzene and then in ethyl alcohol for 24 hours. The fibres were washed in repeated changes of distilled water to remove water-soluble impurities and then dried at room temperature. Fibres were then treated in a mixture, containing equal proportions of 0.5"o w/v aqueous solution of trypsin and a buffer solution of pH 8.6, at 40 C, using a wool to solu- tion ratio of 1 :100. After three months both samples of wool retained their fibrous form, but on pressing gently between a microscope slide and a cover slip, they yielded cortical and cuticular cells; the latter were separated using Woods' technique (15). The purified cortical cells were kept (in a stoppered flask) in distilled water to which a drop of toluol had been added to prevent bacterial growth. Small samples of cortical cells were suspended in dis- tilled water, or lithium bromide solutions', in a cell of an ultrasonic apparatus (5) and were irradiated for periods up to 12 hours. The cortical cells isolated from Australian merino wool and irradiated by ultrasonic vibrations in water suspensions, appeared to break into large aggregates of the parallel macrofibrils, in agreement with the view that a macrofibrillar type of structure is characteristic in this type of keratin (see fig. 1). When, however, similar experiments were made with the cortical cells separated from Lincoln wool, their breakdown in water suspensions started with the separation of many superimposed layers (up to eight) of the sheets of microfibrils (see fig. 2). Some seemingly structureless material was found to be situated between these sheets (7); such films could often be seen over the cracks in the sheets of micro- fibrils (see fig. 2). In agreement, however, with our previous experi- ence with the chemically untreated keratin (6), we found it impossible to separate the very thin sheets of microfibrils from their thicker aggregates. The discrete steps of breakdown could, however, be obtained when the cortical cells, isolated from Lincoln wool, were subsequently exposed to the action of ultrasonic vibrations in lithium bromide solutions (fig. 3). During these experiments we have obtained evidence of the separation of many thin sheets of the microfibrils "peeling off" from the out- side surfaces of the cortical cells, even after relatively short time of treatment (four hours). We have also observed that some less resistant cells were disinte- grating into the thin sheets of microfibrils and the residues of some globular material could be seen scattered over their surfaces (see fig. 4). C C M ■MIF" Type 2 C.C M ■■MIF" Type 2 Fig. 5. Diagram illustrating cortical ceil. MIF, microfibril; MACF, macrofibril; C.C.M., cortical cell membrane. The results of our experiments indicate that the differences in stability of the cortical cells to the action of ultrasonic vibrations are not confined to crimpy wools only. The most resistant cortical cells isolated from the other (uncrimped) wool remain unaffected by the action of ultrasonic vibrations for considerable pe- riods of time. The less resistant cells show the sepa- ration of thin sheets of microfibrils, followed by the appearance of aggregates of the microfibrils. The presence of small debris (other than that of the intercellular origin) in the early stages of treatment of cortical cells suggests that there are also some cells of very low stability. Thus, there is no doubt that a considerable diffe- rentiation does exist in the organization of cortical cells, and we suggest that the following model (see fig. 5) could account for the available evidence. We suggest that the cortical cells are surrounded by a relatively high number of thin sheets of microfibrils and that there is also some less orderly arrangement of the aggregates of microfibrils inside the cortical cells; consequently we agree that the idea of the existence of the macrofibrillar structure in the interior of cortical cells should be regarded as consistent with the experimental evidence. The idea of the extensive sheets of microfibrils has been very recently supported by some x-ray evidence kindly placed at our disposal by Mr. Woods and 1 The lithium bromide solutions used in experiments de- scribed in this paper, were made by dissolving 100 g of salt in 100 ml of distilled water. Pigmented Kevatinous Materials 303 Mr. Skertchly. They found that all equatorial x-ray reflections, obtained from Lincoln wool, appear, within narrow limits, to be either orders of, or related to, the reflections near 250 A. They traced many orders up to the 57th and regard this as an indica- tion of a considerable degree of unidirectional order, consistent with a sheet-like structure. The same workers have also suggested the possibility of the existence of a hexagonal close-packing of the microfibrils. This point, however, can be resolved only by the direct examination of cross-sections of fully hardened keratin fibres. We wish to thank Messrs. H. J. Woods and A. B. Skertchly for placing at our disposal their unpublished results. One of us (M. W. A.) is indebted to the C.S.I. R.O. (Australia) for an overseas studentship which enabled this work to be carried out. References 1. ASTBURY, W. (1953). T., Proc. Roy. Soc. (London) B 141, 1 2. — Inlcni. Wool Text. Res. Conf., Australia, 1955 (in press). 3. Burgess, R., J. Textile Inst. 25. T 391 (1934). 4. Fraser, R. D. B., Intern. Wool Text. Res. Conf., Austra- lia, 1955; during discussion (in press). 5. Jeffrey, G. M. and Sikorski, J., Ann. Conf. of the Elect. Microsc. Group of the Inst, of Phys., 1956. Brit. J. Appl. PItys. (summary in press). 6. Jeffrey, G. M., Sikorski, J., and Woods, 11. ,1 , liuern. Woo! Test. Res. Conf., Australia, 1955 (in press). 7. — Te.\tile Research J. 25, 714 (1955). 8. KoNDO, T. and Horio, M., Textile Research J. 23. 373 (1953). 9. McArthur, I., Proc. Roy. Soc. (London) B 141, 33 (1953). 10. MiRCF.R, li. H., Textile Research J. 23, 380 (1953). 1 1. Pauling, L. and Corey, R. H., Nali4re 171, 59 (1953). 12. RoGi RS, G. E., Conf. on Hiectr. Microsc. of Text, and other Industrial Fibres, Inst, of Phys., Leeds. 1956; summary by Challice, C. E. and Sikorski, J., Brit. J. Appl. Phy.f. (in press). 13. Sailow. G. and Sikorski, J., Melliand Textilher. (in press). 14. Sikorski, J. and Woods, H. J., Conf. on Electr. Microsc. of Text, and other Industrial Fibres, Inst, of Phys., Leeds, 1956; summary by Challice, C. E. and Sikorski, J., Brit. J. Appl. Phys. (in press). 15. Woods, H. J., Proc. Roy. Soc. {London) A 166, 76 (1938). The Electron Microscopy of Pigmented Keratinoiis Materials J. Hope, J. Sikorski, and C. S. Whewell Textile Chemistry^ and Textile Physics'- Laboratories, Department of Textile Industries, University of Leeds The study of colouring matters found in keratinous materials has an important bearing on many prob- lems encountered in biology, genetics, medicine and industry. During recent years it has been possible to make accurate determination, for the first time, of the size and shape of the pigment-containing granules (4, 8, 11). In more recent studies valuable information has been obtained on the size and shape (1,6 and Bohren and Buss, private communication) and the disposition (1,2, 6, 9) of pigment-containing granules in the keratin matrix. The examination of ultra-thin sections of pig- mented follicles (1, 2, 9) and cf isolated melanin granules (2) has revealed that the latter are aggre- gates of dense particles. In some granules the mela- nin could be considered as deposited in parallel lamellae (1, 2, 9 and Charles, private communica- tion) which may form a succession of cylindrical or ellipsoidal shells. One of the main problems in investigating keratin pigment is to devise a method for isolating pigment- 1 J. Hope and C. S. Whewell. - J. Sikorski. 3 Three hours were required to isolate the pigment-contain- ing granules, by centrifuging the PHT suspension (10 ml) at 10,000 g (instead of 10 hours at 2000 g, in 50 ml tubes; Laxer et al. (6)). containing granules free from contamination by residual keratin without ad\erse moditication of their morphologi- cal and crystal structure (6). The following methods have been used for isolating pigment-containing granules from feathers and mammalian hairs: (i) enzymatic digestion (2); (ii) treatment in urea solutions, containing papain and sodium metabisulphile (2); (iii) boiling in Nji potassium hydroxide (I, 2); (iv) boiling or relluxing in 3 A' (2), 6 A (6, I and Bohren and Buss, private commu- nication) or concentrated h\drochloric acid (II); (v) refluxing in an iron-free solution containing 94 g of phenol, 9.0 g of distilled water and 8.2 ml of thioglycollic acid-PHT reagent (6). It is evident that granules isolated by most of the above methods are modified to some extent, if the method is modified so that the isolated granules appear to be intact, there is generally a great degree of contamination by residual keratin. This empha- sized thedesirability of electron microscopic examina- tion as control technique for indicating the efficacy of any process for separating granules from kerati- nous materials (6). The PHT (phenol hydrate and thioglycollic acid) method, however, was found to satisfy both criteria discussed above, but it does involve certain experimental difticullies of separating out the granules from viscous solutions.' An examination of various methods of isolating pigment-containing granules in keratinous materials. All the methods examined of isolating pigment- 304 J. HOPE, J. SIKORSKI AND C. S. WHEWELL Table 1. The examination of potential methods for isolating pigment-containing granules in keratinous materials. Reagent Details of treatment PH Material Sodium hydroxide 2.5 A' 3.0^ 5.0^ Sodium carbonate (15 %) Sodium sulphide (0.5 M) Sodium bisulphite (0.1 A') in 50/50 //-propyl alcohol/water Monothioglycol (0.2 M) in 50/50 77-propyI alcohol/water Thioglycollic acid (0.2 M) in 50/50 //-propyl alcohol/water Sodium thio-glycollate (0.5 M)with sodium hydroxide Monothioglycol (0.5 M) with sodium hydroxide Sodium bisulphite (0.3 M) and urea (10.0 M) Monothioglycol (0.5 M) and ammonium thiocyanate (9 M) Hydrochloric acid (6 A') PHT reagent (see text) Boiling for 5 min. At 98 C for 3 hr. Boiling for 24 hr. Boiling for 45 min. Refluxing for 24 hr. Refluxing for 48 hr. Refluxing for 24 hr. Refluxing for 24 hr. Refluxing 3 to 24 hr. Refluxing 3 J hr.'- Refluxing for 4i hr. Refluxing for 24 hr. Refluxing for 24 hr. Refluxing for 24 hr. Boiling for 48 hr. Refluxing for 24 hr. 12 Ch., BWM. C/R »» Ch., BWM. Ch.,BWM. 6.5 G BWM. 5.0 G, BWM. 5.0 G, BWM. 12 G BWM. BWM. 12 G 7 Ch., BWM 7 G BWM. 1 C/R— Crow or Rook; Ch.— Chicken; G— Grouse; BWM.— Black Welsh Mountain Wool. 2 Washed 3 times in boiling 2 N hydrochloric acid for few minutes. containing granules involved treatment with kera- tinolytic reagents. In the present investigation the suitability of several reagents to isolate pigment- containing granules, from black feathers and dark brown Welsh Mountain Wool, was examined. These not only provided data on the effectiveness of the reagents as a means of isolating granules but also yielded information on the relative resistance of the materials under different courses of treatment. The reagents used and other relevant data are sum- marized in Table 1. The experimental procedure was as follows: the feath- ers were washed in warm tap water and soap solution, and after being dried were purified by successive refluxing, for 24 hours, in diethyl ether and ethyl alcohol; finally they were washed in ten changes of distilled water. The quill and rachis were cut ofT and discarded, so that only web was used in the subsequent work. One tenth gram of purified feathers, or hairs, were treated in 10 ml of reagent. After extraction (see Table 1) suspensions were filtered through sintered glass filter and centrifuged. Washing was carried out by centrifuging in: (i) three changes of extracting liquor, (ii) three changes of distilled water, (iii) three changes of acetone and, finally, (iv) two changes of diethyl ether. Residues were then stored under vacuum over phosphorus pentoxide. For electron microscopy, residues were re-suspended by gentle grinding in a drop of distilled water on a micro- scope slide, using a glass stirring rod. A drop of the suspension was deposited on a supporting nitrocellulose, or carbon, film on the electron microscope grids and dried over phosphorus pentoxide. From the above experiments the following main conclusions can be drawn. The residues obtained from all treatments, except that involving the PHT reagent, were either contaminated by keratinous material (see fig. 1 ) or showed varying degrees of attack on the pigment-containing granules. Generally the amounts of keratinous contaminants varied only slightly from one reagent to another with perhaps one exception of treatment (for longer periods) in sodium thioglycollate, which was capable of yielding relatively clean pigment-containing granules. Fur- ther improvement in this respect was obtained, by additional washing for a few minutes in 2 A^ hydro- chloric acid. However, in view of the very high pH values involved in the original reagent, more detailed investigations are necessary before an answer can be given about the applicability of this method, particularly since greater electron transparency was Pigmented Keratinous Materials 305 Fig. 1. Residue from black feather of Red Grouse, after treatment in 0.5 M sodium thioglycollate, at pH 12. ? hours reflux. Note keratin contamination in the background. Magnification 10,000. Fig. 2. Pigment-containing granules from black tail hair of cow. Magnification 10,000. Fig. 3. Pigment-containing granules from light-brown feather of Mallard. Magnification 10.000. Fig. 4. Pigment-containing granules from black feather of pigeon.Twopopulationsof granules. Magnification 10,000. observed of the pigment-containing granules isolated from black chicken feathers by boiling in 2.5 A' sodium hydroxide; the removal of some material was evident along the major axes of the granules. The pigment-containing granules, free from keratin impurities, could be isolated if, in addition to the milder reagents listed in Table 1, use was made of enzymic digestion (2) to remove such histological components, of hairs or feathers, which may resist the action of keratinolytic media (5, 7, 10). However, this is an additional complication which is not neces- sary in the case of PHT method; consequently, this latter method was adopted in all experiments de- scribed below. Electron microscopy of tlie pigment-containing granules isolated by the PHT method. The pigment- containing granules were isolated from five mamma- lian materials and twenty samples of feathers, and relevant data are collected in Table 2. The shapes of granules from mammalian hair arc similar to those described before (6) and their dimen- sions generally fall within the same range. However, 20 - 568204 Electron Microscopy the pigment-containing granules isolated from the skin of grey Fin Whale and the black tail hair of Cow (see fig. 2) show higher width length ratios, 0.7 and 0.67, respectively. Two populations of gran- ules were isolated from black hairs of Cat. The data on the granules isolated from Hedgehog quill are of some significance in the x-ray diffraction studies of keratin (6). The granules separated from feather show generally lower width/length ratios (0.27 to 0.10, for Hclmeted Guinea Fowl and Mallard (see fig. 3), respectively) except hM- the black feather of Pigeon (see fig. 4) which yielded two populations of granules. Generally, the lengths of the major axes of pigment-containing granules vary from 1.76 // to 0.76 // (for Dipper and Crowned Crane, respec- tively), and those of the minor axes from 0.36 // to 0.12 // (for Dipper and Mallard, respectively). No simple relation was found between the shape and the dimensions of pigment-containing granules and the apparent colour of hairs or feathers, from which they were isolated; this is in agreement with the previous work of Laxer et al. (6). 306 J. HOPE, J. SIKORSKI AND C. S. WHEWELL Table 2. Dimensions of PHT-isolated pigment-containing granules. So lurce Length Width Width/length ratio Fibre/ Feather Co'our /' No. measured C.V. % /' No. measured C.V. % W/L No. measured C.V. 0/ Cat Black 0.85 50 33.3 0.23 50 9.6 0.27 50 34.3 0.84 28 29.9 0.17 28 10.6 0.20 28 31.9 Hedgehog quill Dark brown 0.74 243 20.3 0.26 312 16.9 0.35 204 20.8 Dog Black 0.65 75 17.6 0.25 75 17.3 0.38 75 19.9 Finwhale (skin) Grey 0.64 79 19.6 0.45 79 16,9 0.70 79 23.5 Cow (tail) Black 0.45 73 15.0 0.30 73 17.1 0.67 73 21.6 Dipper Dark brown L76 17 22.5 0.36 17 24.6 0.20 17 26.1 Red grouse Black 1.52 16 17.9 0.28 16 13.5 0.18 16 21.0 Budgerigar Black L38 30 21.2 0.20 30 19.1 0.14 30 22.9 Black grouse Dark brown 1.37 24 23.8 0.20 24 10.6 0.15 24 29.5 Red grouse II Mottled black/brown 1.33 33 27.0 0.27 33 16.9 0.20 33 39.1 Amazon parrot Black 1.33 60 15.9 0.31 60 14.0 0.23 60 20.9 Lapwing II Brown 1.31 61 15.6 0.16 61 23.6 0.12 61 31.2 Golden pheasant Mottled black/brown 1.28 23 14.4 0.30 23 16.7 0.23 23 16.3 Delamere Black 1.28 120 16.0 0.24 120 13.1 0.19 120 17.7 whydah Arctic skua Brown 1.22 47 19.3 0.22 47 16.6 0.18 47 32.8 Oyster catcher Dark brown 1.20 81 16.8 0.27 81 12.8 0.22 81 19.9 Glossy ibis Dark brown 1.14 28 14.7 0.30 28 19.0 0.26 28 19.7 Mallard Light brown 1.13 55 18.3 0.12 55 12.3 0.10 55 22.6 Chicken Black 1.08 51 24.6 0.27 51 14.7 0.25 51 30.3 Sparrow-hawk Brown L04 50 17.0 0 16 50 17.6 0.15 50 26.2 Jay Black 0.93 69 23.6 0.18 69 23.5 0.19 69 20.4 Helmeted Black 0.93 67 21.2 0.25 67 17.0 0.27 67 32.0 Guinea fowl Pigeon Black 0.87 64 20.7 0.27 64 32.9 0.31 64 18.1 0.57 67 16.6 0.39 67 19.0 0.68 62 28.4 Pea fowl Black 0.86 64 17.3 0.17 64 31.5 0.20 64 32.2 Crowned crane Black 0.76 55 21.2 0.16 55 19.5 0.21 55 28.1 The authors wish to thank Professor B. B. Bohren and Mr. A. O. T. Charles for making it possible to examine the results of their unpublished work. References 1. Barnicot, N. a., Birbeck, M. S. C, and Cuckow, F. W., Ann. Human Genetics 19, 231 (1955). 2. Birbeck, M. S. C, Mercer, E. H., and Barnicot, N. A., E.xptl. Cell Resecinh 10, 505 (1956). 3. Buss, E. G., Thesis. Purdue University, U.S.A. 1949. 4. Dalton, a. J. and Felix, M. D., Pigment Cell Growth, p. 267. Ed. by M. Gordon, Academic Press Inc., New York 1953. 5. Gralen, N., Bioclum. et Biophys. Acta 6, 497 (1951). 6. Laxer, G., Sikorski, J., Whewell, C. S., and Woods, H. J., Bioclum. et biophys. Acta 15, 174 (1954). 7. LiNDBERG, J., Philip, B., and Gralen, N., Nature 162, 458 (1948). 8. Mason, H. S., Kahler, H., and MacCardle, R. C, Proc. Soc. E.xptl. Biol. Med. 66, 421 (1947). 9. Rogers, G. E. and Horne, R. W., Ann. Cont". of the Electr. Microsc. Group of the Inst, of Phys., sum- mary by Challice, C. E., Brit. J. Appl. Phys. 7, 89 (1956). 10. Schuringa, G. J. and Algera, L., Biochini. et Biophys. Acta 6, 497 (1950). 11. Shackleford, R. M., Genetics 33, 311 (1948). Uber den Feinbau der Spinnenfiiden R. Lehmensick und E. Kullmann Zoologisches Inslitiil tier Univcrsitat Bonn, Farasitolof^iscJic Ahtciliing Das Netz der Spinncn bcsicht aus feinsten Fiiden unterschiedlichcr Quaiitat. Die fadigen Elemcnte dcs Grundbaues (Radial- und Rahmenladcn) sind glatt. Uber dieses Gerust ist der Fangfaden meist in Form einer Spirale aufgezogen. Dieser Faden ist so beschaf- Icn, daB die Beutetiere, sobald sie mil ihm in Beriih- rung kommen. hiingen bleiben und dadurch von dem riiuberischen Herstelier iiberwaltigt und aus- gesaugt werden konnen. Wir unterscheiden 2 Gruppen der Aranccic: die Ecribellata und die Crihcllata. Wahrcnd die ersteren zur Erzeugung des Spinnfadens am Abdomen ledig- lich Spinnwarzen besitzen. die in 3 Paaren zu einem Spinnfeld vereinigt sind, hab^n die Cribellaten zu- satzlich ein Feld mit vielen feinen Spinn-Diisen, das Crihellum (s. Abb. \ a) und am Metatarsus einen aus Borsten gebildeten Kamm, das Calaniistnini (Abb. 1/?). — Die Fangfaden dieser beiden Spinnen- arten unterscheiden sich im Aufbau erheblich. — Der Fangfaden der Ecribellaten besteht aus einem Achsenfaden als Grundelement, auf dem eine Kleb- substanz in Form feinster Tropfchen aufgetragen ist. Cribellum vordere Spinnwarzen mittlere Spinnwarzen hintere Spinnwarzen After Abb. 1. Spinnwerkzeuge und Fadenschcma der Cribellaten. rt) Spinnfeld mit Crihe/liini; h) Tarsus und Metatarsus des 4. Beines mit Calamistruni {a und h von Diclyna viridissiina aus Kiikenthals Handbueli der Zoologic); t) Fangfaden von Menneus (nach Akermann). — Der Fangfaden der CribeUaten hingegen hal eine viel kompliziertere Struktur. Die lichtmikro- skopische Analyse seines Feinbaus stolit auf so groBe Schwierigkcitcn, daB es bis jctzt noch nicht gelungen war, eine eindeutige, klare Vorstellung davon /ii eiitwiekeln. Das heute am meisten aner- kannle Hau-Schema stammt von Akermann (s. Abb. \ c): auf 2 Achsenfaden (A) ist cine Klebmasse in Bandform aufgetragen (K), die beidseitig von je einem feinen Kriiuselfaden (KF) begrcnzt vvird. Den seitUchen AbschluB bildet jcderseits ein regelmiiBig geweUter Randfaden (R). Die Natur des Klebbandes mit seinen paarigen Kriiuselfaden war aber bis heute sehr umstritten, und es schien mir daher lohnend, mit Hilfe des Elektronenmikroskopes cine Kliirung zu suchen. Zusammen mit meinem Schiiler und Mitarbciter Herrn E. Kullmann habe ich begonnen, dieses Problem ciner Losung zu/ufiihren und mochte Ihnen hier unsere gemeinsamen Ergebnisse mitteilen. I »l f\ n ' I \ *■• ♦ . >. ■\ J^A -2 Abb. 2. Fangwollc von Hyptiotes paradoxus. VergroBerung: El.-opt. 1 1 300 , total 56000 x . 308 R. LEHMENSICK UND E. KULLMANN Abb. 3. FangwoUe von Filistata iiisidiatrix. VergroBerung: El. -opt. 11300 , total 31000 x FiJr die Untersuchung wurden die Netze auf einer siebartig durchlocherten Plexiglasplatte aufgefangen. Die einzelnen Faden, soweit sie die Locher des Siebes frei iiberspannen, konnen dann unter dem Binokular leicht auf unbefilmte Netzblenden aufge- tragen und ohne weitere Priiparation eingeschleust werden. Durch diese hochst einfache Technik bleibt die naturliche Struktur des Objektes (von einigen gelegentlichen Zerrungen abgesehen) unverandert. Die Fiiden der ecribellaten Spinnen ergaben zu- nachst keine sehr ermutigenden Bilder, weil sie sich als undurchstrahlbar erwiesen. Wenn man aber ein Netz Liber Wochen unberiihrt in moglichst staub- freier Umgebung hiingen laBt, beginnen sich die Faden aufzuspalten. Hierbei zeigt sich dann, daB sie aus einer groBenZahl von Elementarfaden bestehen, die fest miteinander verklebt sind. Sehr viel interessanter gestaltete sich die Analyse der Fangfaden cribellater Spinnen. Das elektronen- optische Bild der Klebmasse (K) des Schemas (Abb. if) bot eine besondere Uberraschung. Die ,, Kleb- masse" erwies sich als eine aus feinsten Faden gebil- dete ,,Fangwolle" (Abb. 2). In diesem Filz verstricken sich die Opfer mit ihren Borsten und bleiben darin wie an einem Leim hangen. — Damit bestiitigt sich eine alte Vermutung Bertkaus, der bereits Ende des vorigen Jahrhunderts auf Grund logischer (Jber- legungen zu der (Jberzeugung gekommen war, daB die Klebmasse der Cribellaten eine feinfadige Struk- tur haben miisse. Er konnte sich mit dieser Vermu- tung aber nicht durchsetzen. — Betrachtet man die Elementarfaden dieser Wolle genauer, so erkennt man deutlich, daB sie eine perlschnurartige Eigen- struktur besitzen. — Die weiteren Untersuchungen ergaben, daB, wie zu erwarten war, alle von uns daraufhin angesehenen Spinnen dieser Artengruppe als ,, Klebmasse" eine ,,Fangwolle" produzieren, die im Prinzip stets den gleichen Grundaufbau hat. Es kann wohl keinem Zweifel unterliegen, daB sie im wesentlichen das Produkt des Crihelliims ist, welche mit Hilfe des Calamistiums in der eigen- artigen Form auf die Achsenfaden aufgetragen wird. Bei Ulohonis spec, und Dictyna latens sind die Elementarfaden fast genau so gebaut wie bei Hyptio- it's. Auch ihre MaBe liegen in gleicher GroBenan- ordnung (0,01-0,02 //). — Es kommt zuweilen vor, daB die Elementarfaden teilweise zu Bandern und kleinen Platten verklebt sind, ein Phanomen, das interessante Riickschliisse auf die Bildungsweise der Wolle zulaBt. Die aus einer DUse austretende Faden- substanz bleibt ofTenbar nur dann als distinkter Faden erhalten, wenn sie sofort erhartet. Verzogert sich Uber den Feinbau der Spinncnfdden 309 Abb. 4. Bandformiger Fangfaden von Lo.\osceles nifescens. VergrolJerung: El. -opt. 5700 >• , total 25 500 dieser Hartungs-ProzeB, dann verkleben diese Fiiden mehr oder minder fest miteinandcr, so daB eventuell richtige Schleier entstehen. Vielleicht ist aber auch eine zusatzliche diinnflussigere Substanz im Spiel. Es ist sehr wahrscheinlich, daB die Spinnen einen gewissen EinfluB auf diese Prozesse nehmen und dadurch ,,willkijrlich" den Faden einen unterschied- lichen Charakter verleihen konnen. — Besonders interessant sind die Elementarfiiden von Filistata insi- diatrix (Abb. 3). Sie sind nicht rund und perlschnur- artig, sondern haben die Form schmaler Bander. Irgendwelche Strukturen, die den Knotchen der Ele- mentarfadenvorgenannter Spinnen entsprecinen, kann man nicht an ihnen entdecken. Es ist anzunehmen, daB die Struktur der Elementarfaden u. a. abhangig ist von den Dimensionen der Austrittsoffnung und der Viskositat der die Faden bildenden Driisensub- stanz. — Die Elementarfaden scheinen bei ailer Zartheit auch noch eine gewisse innere Eigenstruktur zu besitzen: einige haben niimlich eine wellige Grundform, andere sind ausgesprochen gradUnig gebaut. Das Farbenspiel bei Betrachtung mit ge- kreuztenNicolsim Polarisationsmikroskopverrat,daB sie alle geladen sind mit inneren Spannungen. Diese Verhaltnisse ermogUchen es wahrscheinlich, daB die zarten ,,Wollhaufen" iibcr viele Tage selbst im wechselvoUsten Mikroklima ihre lockere Struktur beibehalten und funktionstiichtig bleiben. Die in den Mittclmcerlandcrn heimische ccribel- late Spinne Loxosceles rufescem lebt am Boden und fertigt ein eigentiimliches Fanggcwebe an, dessen Elemente mit ihrcm bliiulichcn Schimmer schon mit bloBem Auge mehr an die Faden cribellater Spinnen erinnern. Es ist Herrn Kullmann gelungcn, diese Spinne in Sardinien zu finden und Icbend mit nach Bonn zu bringen, so daB wir auch ihre Fiiden unter- suchen konnten. Wie Abb. 4 zeigt, besteht dieses Fanggewebe aus breiten Biindern, die sehr den Elementarfaden von Filistata ahneln. Diese Bander sind freilich um vieles breiter, aber ihre Dicke betriigt ebenfalls 0.02 //. Sie weisen eine sehr interes- sante Liingsstruktur auf, die vielleicht mit der Entstehungsweise aus Einzelfiiden zusammenhiingt. — Damit riickt Loxosceles auch von dieser Seite her in nahe Beziehungen zu den Cribcllaten. Diese unsere ersten, naturgemiiB noch sehr lucken- haften Ergebnisse zeigen, daB eine solche elektro- nenoptische Analyse imstande ist, Beitrage zu allge- meinen biologischen und systematischen Problemcn zu liefern. Die Ergebnisse konnen andererseits aber eines Tages auch wertvolle Erkcnntnisse fur die Praxis der Herstellung feinster Kunstfiiden bringen. Die cicktroncnoptischcn Aiifiialimcn stammen aus dcm Labor fur angewandte Ubcrmikroskopie in Bonn. W ir spiechcn scincm Lcitcr, Hcrrn Dr. Wohlfarth-Boltcrmann und seinen Mitarbeitern fiir die Anfertigung der Aufnahmen unscrcn her/lichsten Dank aus. LrrpRATUR 1. Akirmann, Ann. Natal Museum 5 (1926). 2. BrRTKAU, Wiegnwnn Arch. Naturgeschichle 48 (1882). 3. CoMSTOCK. The Spider Book. New York. 1912. 4. Kastnkr, Kukenlhals Handbucli der Zoologie 3 (1941). 5. Nielsen, The Biology of Spiders. Copenhagen, 1932. 6. Peters, Z. Satuiforscli. 10b (1955). XIV METALLOGRAPHY AND OTHER INDUSTRIAL APPLICATIONS Direct Observation of Dislocations and Their Movement in Metal Foils P. B. HiRSCH, R. W. HoRNE, and M. J. Whelan Cavendish Laboratory, Cambridge In order to explain the low values of the shear stress required to start plastic flow in metal crystals, it is necessary to postulate the existence of a lattice imperfection, known as a dislocation. There is a considerable amount of indirect evidence from etch- ing and precipitation experiments (9, 13) that such imperfections exist in metal crystal, whilst for inor- ganic crystals such as AgBr and NaCl (1,5) the dislocation networks may be decorated by various techniques and made visible under the optical micro- scope. However no similar direct observations have been made on metals, simply because it is not possible to examine interior structures with optical tech- niques. It occurred to the authors that much useful information on the arrangement and movement of dislocations in metals might possibly be obtained by examining thin foils directly in the electron microscope. Electron optical transmission experi- ments with gold foils had shown that the contrast was essentially due to Bragg scattering which is structure sensitive. It was therefore thought that dis- locations might be made visible by virtue of their strain fields. This paper is a short account of some observations on dislocation distributions and move- ment in aluminium foils. A fuller account is published elsewhere (8). Aluminium was used in this work on account of its transparency to electrons, and also because much is already known about the substructures formed by de- formation in this metal. Beaten foils \ /i thick, either cold worked or annealed at 350 C in vacuo, were etched in dilute hydrofluoric acid. The foils were examined directly in the Siemens and Halske "Elmiskop 1" operating at 80 KV at an instrumental magnification of 40,000 -< . Evidence for the visibility of dish^cation lines. — Figs. 1-5 show some typical micrographs obtained. The following facts leave little doubt that individual dislocation lines are being observed. (a) Fig. 1 shows that the specimens contain a substructure of subgrain diameter about 1 /<. The misorientations across the boundaries have been determined by diffraction experiments and are found to be about I' or 2°. Fig. 2 shows a typical boundary under higher magnification, showing that it is pos- sible to resolve individual dislocations in the boun- daries. An interpretation in this case is that this is a simple (110) tilt boundary consisting of edge disloca- tions traversing the foil. MicrodiflFraction patterns show that the foil has a strong preferred orientation with a [100] cube axis normal to its surface. This orientation is favourable for the formation of (110) tilt boundaries normal to the foil. Examples are observed of subgrain boundaries traversing the foil at an angle, in which the three dimensional character of the dislocation networks is evident. (b) The dislocation density estimated from several micrographs similar to fig. 2 is about 10'" per sq.cm. This is in good agreement with previous estimates from x-ray evidence (4, 7), and with the average misorientation angles 1 or 2" observed from diffrac- tion patterns. Fig. 1. Siibgrains in cold beaten Al. The average subgrain size is about 1 //, the average angular misorientation about 1^°. The dislocation density is lO^" per sq.cm. Extinction contours due to large range strains are shown at A and B. Magnification 12,000. Fig. 2. A sub-boundary consisting of uniformly spaced dis- locations. The average spacing of the dislocations is about 175 A. Magnification 120,000. Dislocations and Their Movcnwnl in Mclal Foils 313 Fig. 3(«). An example of a square cross-grid of screw disloca- tions forming a iwist boundary on (100). Dislocations come to an end on the surface at A. Magnification 100,000. Fig. 3 (ft). A hexagonal network of dislocations. Magnification 200,000. Fig. 4. Fast dislocation slip trace showing cross-slip. Ihc dislocation penetrated the boundary at C. CD is another slip trace with the dislocation held up at D. Magnification 60,000. (c) Some of the arrangements of lines can be explained readily on dislocation theory (see below), but not on any other basis known to the authors. (d) On tilting the illumination or the specimen through small angles, the lines remain fixed in posi- tion, although the contrast changes. This shows that the lines are a definite property of the specimen. (c) When working with large beam current den- sities the lines are observed to move; the movement occurs along straight lines parallel to the traces of (111) slip planes, fig. 4. Contrast mechanism. — The diflFerences in contrast between one subgrain and another are due to differ- ences in intensities of Bragg refiections caused by the small misorientation angles. This is easily dem- onstrated by tilting the object in a stereo holder through an angle of 1° or 2'', when it is possible to illuminate grains which were originally dark and vice versa. The thickness of the foil is sufficient (about 500 A) for the scattering of electrons to be dy- namical, and depending on the orientation much inten- sity may be abstracted from the direct beam and the Obiect Ezzz Ezzzzz; czza Len s (a) Objective Aperture ---- Low Intensity — • Hiqh Intensify (b) Fig. 5((/) and (h). Illustrating Bragg contrast between sub- grains. («) Grain appears light, (h) Grain appears dark. low angle scattering by intense Bragg reflections. Owing to the physical objective aperture (30 // in these experiments) these Bragg rcficctions do not contribute to the image and such a subgrain appears dark, see fig. 5 (a) and (/;). It will also be noticed that the boundaries and the dislocation lines are darker than the surrounding regions. The dislocation line contrast was at first thought to be due to impurities, but similar observa- tions on high purity foils eliminate this ctTect. It is now thought that the contrast is again due to in- creased Bragg refiection from the strained region around a dislocation line. It is known that the intro- duction of a simple dislocation into an otherwise perfect lattice can produce difi'useness of the reci- procal lattice points (14). The dark appearance of the dislocation lines in bright field is therefore due to an increased probability of refiection into several Bragg reficctions from the strained region around the dislocation. This is shown schematicalK in tig. 6. (ooo) • (ooo) (a) Strain-Free Stronq Reflection Weak Reflection (b) Strained Fig. 6((/) and (/>). Schematic diagram of the reciprocal lattice and reflecting sphere, showing how the strain ticid of a dis- location may produce increased Bragg scattering. («) A strain-free region may only give rise to one or two strong Bragg reflections, (h) A strained region may give rise to several strong reflections. 314 p. B. HIRSCH, R. W. HORNE AND M. J. WHELAN Table 1. Siininiary of results of boundary misorientation measurements in three cases. Plate numbers Normal to foil Type of boundary assumed Rotation axis Component angle of rotation Calculated dislocation spacing Observed approximate spacing of dislocations Resolution 94, 95 [001] 293, 294, 296 [001] 299, 300, 302 [001] (110) Simple tilt (211) Mixed tilt Hexagonal network near (111). Twist boundary [112] [Oil] [111] 2.1° 1.36 0.35 = 80 A 94 A 270 A 130 A Poor 100 A Good in places 300 A Fair Dark field experiments, in which the 30 // objective aperture is moved to accept several Bragg reflections in turn, have shown that the contrast is reversed, and that the dislocations and boundaries are illu- minated usually in several Bragg reflections, whereas strain free areas show up in only one or two reflec- tions. The contours visible at A and B in fig. 1 are extinction contours first studied by Heidenreich (6). It is thought that these may be due to bending in the foil produced by large range strains in otherwise perfect grains. Arrangement of dislocations. — The direct observa- tions of dislocations in the subgrain boundaries en- able dislocation theory to be checked quantitatively. Foracomplete discussion of networks and boundaries in face-centred cubic crystals reference should be made to Frank (3), Ball and Hirsch (2), and Amelinckx ( I ). Dislocation theory predicts that simple boundaries may be considered as surface arrays of dislocations. A probable example of a simple (1 10) tilt boundary has already been given in fig. 2. This type of boun- dary contains dislocations of one of the twelve cubic slip systems only. More complicated boundaries may contain dislocations from two or more slip systems. Fig. 3 (a) shows an example of a simple twist boundary on (100), consisting of a crossed grid of screw dislocations. The crystallographic orientation of this field of view was determined by diffraction. The normal to the foil was close to [100] and the lines run parallel to [110] directions. The network therefore makes a shallow angle with the surface. It can be seen at A that dislocation lines end on the surface. Another network, predicted theoretically (3) and observed extensively in AgBr and NaCI crystals (1, 5), is a hexagonal network of screw dislocations lying in a (111) plane, forming a twist boundary. Fig. 3 (b) shows a small piece of hexagonal network in aluminium. Large hexagonal networks are not observed, presumably because the foil is thin and areas of [111] orientation are rare because of the preferred orientation. Experiments have been performed to check the theory quantitatively by microdiffraction by selecting a small area across a subgrain boundary. From the splitting of the diffraction spots it is possible to obtain the component misorientation angle about the axis of the instrument. A high resolution micro- graph often enables the type of boundary to be de- termined. This is possible when the resolution is particularly good and when interference effects pro- duced by the overlap of crystals at the boundary are absent. These interference effects produce fringes at the boundary, which tend to mask partially or com- pletely the dislocation network. Table 1 contains a summary of the results obtained in three cases. The dislocation spacing, calculated from the observed misorientation angle and the assumed network, is compared with the approximate observed spacing. In these three cases the type of boundary present is only tentatively proposed. In the first two cases the boundaries were inferred from the trace of the boun- dary plane. In the third poorly resolved hexagons could be seen. It is seen that there is agreement to a factor of 1.5. This is evidence in favour of disloca- tion theory, and shows that the boundaries do con- tain dislocations of unit Burgers vector. Movement of dislocations. — When working under fine focus illumination conditions with the double condenser lens, dislocations are observed to move. The reason for the movement is not at present clear, but it is thought that it may be due to a combined effect of heating and straining due to thermal gra- dients in the specimen. The temperature rise in the object is not known.Thefirst indication of movement is that certain lines bow out. This is presumably direct confirmation of the mechanism suggested for the decrease in elastic modulus due to dislocations (11). Two types of movement can be observed. The first is slow and jerky, probably requiring consider- able cross-slip and climb, which possibly corresponds to creep. The second type of movement is very fast and usually it is only possible to observe the appearance Dislocations and Their Movement in Metal Foils 315 of the band of contrast left after the dislocation has passed. Cine film measurements have shown that these bands appear in less than .5^ second. Fig. 4 is a typical example of fast movement. Here a dislocation has approached a boundary along AB, cross-slipped several times, and eventually pene- trated the boundary at C. Another fast dislocation has left the boundary at C and can be seen held up at the end of the slip-trace CD. The interpretation of the bands is that the dislocation leaves evidence of its passage on the slip plane, and that the width of the band is due to the finite thickness of the foil. In most areas, such as in fig. 4, the normal to the foil is [100] so that the thickness of the foil is roughly equal to the width of the slip-trace. Measurements of the slip traces gives a thickness of roughly 500 A in an average area of the foil. The directions of the slip traces have been determined by difl'raction, and are known to be parallel to the traces of the (111) slip planes in aluminium. At points such as E the dislocation is observed to transfer from one slip plane to another. This is direct evidence of the cor- rectness of the Mott-Frank mechanism of cross slip of a screw dislocation (10). Slip trace contrast. — The band of contrast left by a moving dislocation slowly fades, and in about 10 seconds has completely disappeared. The dis- appearance rate is higher at the larger illumination intensities obtained under fine-focus illumination conditions. The contrast may be either lighter or darker than that of the surrounding region, and is more intense near the edges of the slip trace, i.e. at the surface of the foil (fig. 4). This contrast is tenta- tively interpreted in terms of the generation of point defects by moving dislocations (12). A dislocation line containing jogs may generate vacant lattice sites and interstitial atoms by non-conservative mo- tion of these jogs. It is thought that the short range strain fields associated with a distribution of vacan- cies or interstitials on the slip plane may give rise to Bragg contrast in a manner similar to the strain field of a dislocation. Some preliminary observations indicate that the contrast is also reversed in dark field illumination. The disappearance of the slip trace contrast would then be due to the diffusion of the defects away from the slip plane. A calculation shows that the right order of magnitude for the persistence time is obtained for vacancies if the tem- perature is of the order of 100 to 200 C. The more intense contrast at the edges of the band may be due to the larger number of defects generated near the oxide film, or due to the stresses set up by the interac- tion of the oxide film with a dislocation trying to penetrate the surVdce. (Note in proof. — Recent obser- vations have shown that the slip trace contrast is due to this latter mechanism, and not to point defects.) Conclusion. — There is little doubt thai il is possible to examine the dislocation structure of metals directly by transmission electron microscopy. This study of aluminium foils has shown that the ideas of disloca- tion theory applied to networks and surface arrays are essentially correct. Moreover it has shown in the case of aluminium, typical of a well polygoniscd metal, that most of the dislocations are in the boundaries even in the cold worked state, and very few dislocations remain inside the subgrains. It should be possible to extend these observations to other metals. Some preliminary observations on Pd, Au and Ag foils show that similar dislocations are visible, and il is also possible to see their movement. Many of the effects observed during movement of dislocations are complicated and not completely understood. It has not yet been possible to locate dislocation sources, although expanding loops are often observed. It is however clear that dislocations come out of the boundaries. Some slip traces are observed to terminate at the boundaries while others penetrate them. The detailed movement of the dis- locations in aluminium has been recorded on 16 mm cine film. References 1. Amklinckx S., Phil. Mu};. I, 269 (1956). 2. Ball, C. J. and Hirsch, P. B.. Phil. Mag. 46, 1343 (1955). 3. Frank, F. C, Report of Bristol Conference. 1954. 159. Proc. Phys. Soc. (1955). 4. Gay. P., Hirsch. P. B., and 315 (1953). 5. Hedgls, J. M. and MrrrHrLL (1953). 6. Heidenreich, R. D., J. Appl. Phys. 20. 993 (1949). 7. HiRsrn, P. B.. Acta Ciysi. 5, 172 (1952). 8. HiRSCFi, HoRNE and Whelan, Phil. A-/«^. 1, 677 (1956). 9. Lacombe. p. and Bi aujard, L.. Rev. met. 45. 317 (1948). 10. MOTT, N. F., Proc. Phys. Soc. B 64, 729 (1951). 11. — Phil. Mag. 43, 1151 (1952). 12. Seeez, F., Actvanccs in Physics I, 43 (1952). 13. WiLsnoRF, H. and Ki hlmann-\Vii sdore. of Bristol Conference, 1954, 175. Proi (1955). 14. Wilson, A. J. C, Research 2, 541 (1949). Ki I E'l , A., Acta Met. 1, J. W.. Phil. Mag. 44, 223 D.. Report Phvs. Soc. Dislocations in Stainless Steel W. BOLLMANN Battelle Memorial Institute, Geneva The first experiments on the study of metals by electron microscopy in transmission were under- taken by Heidenreich (5) on aluminium- and alu- minium-4 ",, copper alloy; he mainly investigated cold working and the formation of precipitates using an electropolishing technique for the preparation of the specimen. In the same paper, this author gives a theory on some diffraction effects occurring in the electron microscopy of thin crystal foils. Castaing (3) used, in addition to the electro- polishing, an ion bombardment for the preparation of transparent specimens of the same metals as studied by Heidenreich. The present work was undertaken with the aim of studying creep in certain austenitic temperature- resistant steels. To develop the preparation technique, (18 %) chrome- (8 %) nickel-steel was chosen. The raw specimen, a circular disc (2 cm diameter, 0.2 mm thick) insulated around the edge with varnish was attacked electrolytically from both sides. Thecathodes consisted of pointed electrodes which were insulated except for the points. At the start the electrodes were placed at a distance of 1-2 mm from the disc and the attack was continued up to the moment when a central hole appeared. Then the cathodes were placed at about 1 cm from the specimen. Under these conditions, the specimen was preferentially attacked near the insulated edge and there a second hole opened. The attack was continued until the space between the two holes was about 1 mm. Then the current was applied in pulses up to the moment when the two holes joined. In that region the specimen showed fairly large areas which could be observed by transmission. The electrolytic solution consisted of 40 °o sulfuric and 60 °o orthophosphoric acid. The current applied was about 3 amp. Much attention had to be paid to the cleaning of the specimen; it was rinsed as soon as possible, first in hot running water, then in a hot sequestrol solution in order to dissolve all the salts and finally in hot distilled water. ExperiiiK'itts with ion hoiiihardmetit. — Some experi- ments were undertaken to apply the method of Castaing (3) on stainless-steel specimens. In the first trials the ion bombardment was applied through a small ion gun (ca. 1 //A 1500 Volt). The ions of air and of pure nitrogen produced a strong corrosion of the specimen; so the experiments were continued with argon. Here, the specimien was directly placed as cathode into a gas-discharge. The current in this gas discharge proved to be practically independent of the gas pressure in a wide range (1-0.05 mm Hg). As the current density is the product of the ion density multiplied by the mean velocity of the ions, this ve- Fig. 1. Rows of dislocation lines crossing a steel foil (piled up against obstacles). Fig. 2. Network of dislocation lines in lightly rolled steel. Dislocations in Stainless Steel 317 ■% mm Fig. 3. Stacking faults. Fig. 4. Steel reduced 25 "„ by rolling. locity increases with decreasing gas pressure (because the ions suffer a smaller number of collisions with gas atoms). With increasing mean velocity the mean energy of the ions impinging the surface also increa- ses. High energetic ions strike out of the specimen a large number of atoms per ion, producing in this way small holes. Therefore, the lower the gas pres- sure, the rougher will the treated surface become. Thus the strength of the attack can be regulated by varying the gas pressure. For our purposes, the ion bombardment did not present an advantage. The smoothest surface was obtained directly by electropolishing. The importance of ion bombardment for the work of Castaing on Al-4 "o Cu alloy seems to lie, at least partially, in the fact that the Al is preferentially electropolished (as pictures of Castaing show), but that the ion bom- bardment released the copper preferentially from the surface. It is known that it is relatively simple to sputter (release by ion-bombardment) heavy met- als such as copper, but that it is difficult to sputter aluminium because of the oxide layer (7). Micrographic appearance of dislocations in crystal foils. — Much theoretical work has been done on dislocations in crystals to explain the behaviour of metals during plastic deformation (sec 4, 6). In a plastically deformed metal, the strongly distorted regions are concentrated on definite lines, the so- called "dislocation lines". The material around these lines is more or less elastically deformed. The theory shows that the dislocation lines have to be closed loops or have to end at the surface of the grains; they can be branched into extensive networks (figs. 1. 2). Until recently, experimental evidence on disloca- tions in metals was given by etching the surface, but here only the end-points of dislocation lines appear as etchpits (2). In salts it was possible to make dislocations visible by metal precipitations (I). The reason why dislocations can be seen directly by electron microscopy inside the material is due to the fact that diffraction effects contribute markedly to the image contrast in the electron microscopy of transparent crystal foils. As electron diffraction (Bragg condition) depends among other things on the lattice constant, a local distortion of the lattice becomes visible in suitable orientations. Sometimes, dislocation lines appear as dotted lines. Using the dynamic theory of diffraction, Hei- denreich showed that when a Bragg angle is nearly attained by the incident electrons, the brightness of the picture of a crystal foil varies periodically with growing thickness. Thus, a foil with varying thickness shows interference fringes. It can be shown from different pictures that the dotted dislocation lines might be related to these fringes. In fig. 3 interference fringes are visible both at the grain boundary and within the grains; at the same time the dislocation lines in the upper part are dotted. If the angle be- tween the object plane and the electron beam is changed, the fringes disappear and the corresponding dislocation lines become smooth. 318 R. ARNAL AND M. SOREL A stacking fault in a cubic face centered crystal is a 1 1 1 -plane which is in a false position with respect to its neighbouring planes. In the case of a stacking fault, the periodicity of the crystal lattice is disturbed across an extended surface. Fig. 3 probably shows such faults. When this type of plane ends inside the crystal, its edge is a so-called "partial dislocation". In the present work, the steel was rolled in order to produce dislocations uniformly throughout the whole specimen. The pictures show qualitatively the general elTects of cold working. A detailed evalua- tion in terms of the theory of dislocations has yet to be made. Fig. 1 shows how in a slightly rolled specimen ( 1-2 "„ reduction in thickness) dislocation lines begin to spread out from the grain boundary along slip- planes into the interior of the grains. With heavier rolling the density of dislocation increases very fast (fig. 2). In a 25 »„ reduced specimen the slip-planes become curved (fig. 4). For a study on dislocations, stainless steel is con- venient especially for the observation of static con- figurations, because the annealing temperature, where the dislocations move, is too high to be attained by heating the object by the electron beam. The situation is different with aluminium, however, as it is possible with this metal to study dislocation movements as shown by Whelan, Home, Hirsch (8). So, for this type of research these two materials are in some ways complementary. The present work has been sponsored by the Union Miniere du Haul Katanga and the Battelle Memorial Institute. I wish to thank Dr. W. Siegfried for having suggested the research. Our work has been undertaken and the main results have been obtained independently from the group Whelan, Home, Hirsch. I thank Dr. P. B. Hirsch for useful discussion on the interpretation of some special points, and Dr. J. W. Menter for his criticism of the manuscript. References 1. Amelinckx, S., Phil. Mag. 1 (8th Series) 269 (1956). 2. BiLLiG, E., Internationales Kolloqiiium iiber Halbleiter und Phosphore. Garmisch-Partenkirclien, August 1956. 3. Castaing, R., Rev. met. 52, 669 (1955). 4. CoTTRELL, A. H., Dislocations and Plastic Flow in Crystals. Oxford, 1953. 5. HEiDFNRriCH, R. D., /. Appl. Phys. 20, 993 (1949). 6. Read, W. T. Jr., Dislocations in Crystals. New York, 1953. 7. Wehner, G. K., Phys. Rev. 102. 690 (1956). 8. Whelan, M. J., Horne, R. W., and Hirsch, P. B., Annual Conference of the Electron Microscopy Group of the Institute of Physics. Reading (England), July 1956. Migrations of Grain Boundaries Studied with the Emission Electron Microscope R. Arnal and M. Sorel Lahoratoire d'E/eclroniqiie et de Radioelectricite de la Faciilte des Sciences de Paris Since the growth of metallic crystals can be ob- served continuously with the help of cinematographic recording, it is possible to get quantitative informa- tion on the cinetics of grain boundaries. It was noticed, some time ago, that there was a geometrical analogy between crystal boundaries and two-dimensional soap froth; in this latter case, the mechanism is well known; the equilibrium of inter- faces is due to the presence of surface tensions. To show the analogy in a better way, we have tried to observe bidimensional crystals in order to have the same conditions as for soap froth. On the other hand, when the mean diameter of crystals is smaller than the sample thickness, the growth or the disappearance of an internal crystal carries a change in topography and one can only observe secondary effects, the internal cause of which is unknown. The emission electron microscope has been used with a small magnification (30 times for a screen distance of 70 cm) and the sample heated by elec- tronic bombardment is about 0.2-0.3 mm thick. On a screen (8 8 cm), it is possible to see several crystals the diameter of which is greater than the sample thickness. As soon as the diameter of the image of a crystal becomes greater than j^„ of the side of the screen, one can say that it is the bidimensional case. The fluorescent screen is filmed at the rigorously constant rate of two frames per second. During the projection, the crystal growth is therefore shown ten times the speed. The luminous intensity is measured with a photomultiplier, large range variations of luminosity are balanced by a hand-regulated dia- phragm. The sample is heated by electronic bombardment and the temperature is measured by a chromel- alumel thermocouple, directly soldered on to the sample. After testing many metals, titanium, produced by Pechiney, has proved to be the material for which the crystal growth after the phase transformation (at about 1100 C) is the quickest and the more in- teresting. Pcrlit- unci Bainitgefiigc in i/ici Kohlcnstoff'siahlcn 319 To sum up, in this work the emission electron microscope is not used as a magnification tool, hut as a tool permitting the study of the migrations of crystal boundaries in a continuous way at a high temperature. It has been possible to observe three kinds of recrystallisation: primary recrystallisation, secondary recrystallisation and strain-induced growth. In about five seconds, the temperature is raised to 1100. then maintained constant during the growth. The primary recrystallisation of the /^ phase takes place very rapidly in these conditions (about 30 seconds). It is known that the energy necessary for this kind of recrystallisation comes first from the strain existing in the metal; then, when the grains tilt each other, the equilibrium of the boundaries is reached as a result of surface tension and the dis- appearance of some instable grains occurs as in a soap froth. Then the preferential growth of a crystalline plane can be observed. This technique of filming evidently shows that the superficial energy is not the driving force of this growth; in fact, one observes "buddings"" at the boundary of the growing grain. However, this migration is slow enough to allow the intercrystalline boundaries to reach equilibrium under the influence of surface tension; therefore, grain boundaries mi- grate toward their centers of curvature ar.il the shapes of metal grains are identical to shapes of cells in foams. When this growth has stopped, the sample is cooled below the transformation temperature, then is heated again to I 100 C. The shape o\ the [i crystals is the same as before, but the crystals are strained. So a new kind of migration appears; the boundary moves in an irregular front consisting of curved sections. The same "buddings"" as those of the secondary recrystallisation are observed on the film. It does not seem to us that there is a fundamental dilVerence between secondary recrystallisation and that kind of growth, both being induced by strain: hut in ihc former case, the secondary recrystallisa- tion grows among a structure of small crystals; the growth speed, induced by the residual strain, is smaller than that restoring the equilibrium of grain boundaries, and for this reason the grain boundaries move toward their centers of curvature. In the latter case, the migration, with the same characteristic buddings, does not take place in a structure of small crystals, which accounts for the boundaries migrat- ing away from their centers of curvature. Perlit- Lind Bainitgeflige in drei KohlenstofTstahlen mil 0,18%, 0,50^,und 0,86% C S. MODIN Metallografiska Iiisiiliilcl, Slockholni 1m Metallografischen Institut, Stockholm, werden Mikrostrukturen in Kohlenstoffstiihlen nach ver- schiedenen Warmebehandlungen untersucht. Der vorliegende vorlaufige Bericht behandelt die Struk- turen in zwei untereutektoiden Stiihlen nach iso- thermer Austenitumwandlung im Vergleich mit den entsprechenden Strukturen in einem eutektoiden Stahl. In einem vorhergehenden Bericht sind die in einem eutektoiden Stahl auftretenden Strukturen schon beschrieben worden (3). Die bei der Wiirme- behandlung erhaltenen Strukturen sind auf zwei ver- schiedene Arten untersucht worden. Einerseits wur- den polierte und geiitzte SchlitTe im Metallmikroskop und nach Herstellung von Lackabdriicken auch im Elektronenmikroskop untersucht, anderseits wurden chemisch isolierte Karbidkorner direkt im Flcktro- nenmikroskop abgebildet. Prcipariening. — Die Proben hatten Schcibcnform mit einer Dicke von nur 0,8 mm, um eine Umvvandlung bei der gewiinschten Tempcratur zu gevvahrlcistcn. Die Austenittemperatur war llOOC und die Haltezeit 10 Min. Die isotherme Umwandlung erfolgte in einem Metallbad. Sic wurde durch Abschreckung in 5 "„ Natronlauge abgcbrochcn. Nach Schlcifen und me- chanischem Polieren warden die Pmbcn gciit/t: fiir metallmikroskopische Beobachtungen in 4 "„ Pikrin- siiurc in Ethylalkohoi, fiir clcktroncnmikroskopische Unlcrsuchungen in -) <> Losung von Saipctcrsaurc in Amylalkohol. Die Lackhiiiitchen warden aus einer Lo- sung von Formvar oder Mowital, gelost in neu destillier- tem Chloroform, hergestellt. Die Konzentration der Lo- sung war 0,7 g Lack auf 100 ml Losungsmittcl. Vor dem Abziehen wurden diese Hautchcn mit cincr KolloJium- schicht verstiirkt. Das Kollodium war in Amylacetat gelost und die Konzentration war 4 "„. Die Doppel- hiiutchen wurden in iiblicher Weise unter Wasscr abge- zogcn. Es hat sich herausgestellt, daB man bei diesem Abziehen oft auch eine Mcnge Zementitkorner vom Sehlitf mitbckommt. Man kann die Abdriicke von diesen anhaftenden Korncrn diircli Baden in einer verdiinnten Siiure befreien, z. B. in Seliwefelsaure I :20 (2). Die Lackabdriicke wurden ohne vorhergehende Be- schattung in das Elektronenmikroskop eingesetzt. Um guten Kontrast zu erhallen, ist es vorteilhaft, mit nieht zu groBer Apertur im Objekliv zu arbeilen. Die Karbidkorner warden aus ganz umgewandeiten Proben folgendermaBen chemisch isoliert: sie wurden in Salpelersiiure 1:1 wiihrend ungefiihr 20 Sek. getaucht. An der Oberfliiche geht der Perril dabei in I.osung. Die Probe wurde schnell in Wasser, Alkohol und zuletzt in Amylacetat getaucht. Die an der OherfJiiche der Probe 320 S. MODIN Fig. 1. Isolierter Perlitzementit. C 0,18 "„. 650"C, 5 Min. Vergr. 18000 . Fig. 2. Perlit und Fcrrit. C 0,18 %. 650 C, 2 Min. Lackab- druck. Vcrgr. 3700 >■■ . Fig. 3. Beginnendc Bainitbildung. C 0.86 %. 500°C, 6 Sek. Lackabdruck. Vergr. 6000 . Fig. 4. Isolierter Bainitzementit. C 0,18 %. 450-C, 16 Sek. Vergr. 28000 x. anhaftenden Karbidkorner wurden in Amylacetat mittels Ultraschall abgeschleudert ( 1 ). Die in beschriebener Weise isolierten Karbidkorner wurden im Elektronenmikroskop untersucht. Perlit. — Wenn Austenit sich in Perlit umwandelt, bilden sich an verschiedenen Stellen im Austenit, vorzugsweise an den Korngrenzen, Perlitgruppen (englisch ,, nodules") bestehend aus mehreren Perlit- einheiten (englisch ,, colonies") mil einheitlicher La- mellrichtung.Wahrscheinlich besteht jede Perliteinheit aus einem Fcrritkorn. Die Lamellrichtung der Per- liteinheiten scheint meistens unabhangig von dem radialen Zuwachs der Perlitgruppe zu sein, d. h. sie bildet einen beliebigen Winkel mit ihm. Bei sinkender Umwandlungstemperatur geht die Umwandlung schneller. Eine groBere Anzahl von neuen Perlitgruppen entstehen pro Zeiteinheit, auch die Zuwachsgeschwindigkeit der Perlitgruppe steigt. Man nimmt allgemein an, daB der Zuwachs einer Perliteinheit einerseits dadurch erfolgt, daB schon vorhandene Lamellen an der Kante in der eigenen Ebene wachsen, anderseits dadurch, daB sich neue Lamellen von Ferrit und Zementit neben den schon vorhandenen bilden. Die isolierten Zementitkorner haben Scheiben- form. Diese Scheiben sind stark zerfetzt und haben Locher (Fig. I). Bei sinkender Umwandlungstem- peratur werden die Scheiben diinner, kleiner, werden mehr zerfetzt und bekommen mehr Locher. Der Perlit, der sich in untereutektoiden Stahlen bildet, unterscheidet sich von dem in einem eutek- toiden Kohlenstoffstahl. In solchen Stahlen fiingt die Umwandlung mit proeutektoidem Ferrit an. Dieser Ferrit verhindert den kugelformigen Zuwachs von Perlit und diirfte auch hauptsiichlich die iiuBere Form des Perlites bestimmen (Fig. 2). Bei der Perlitumwandlung in einem eutektoiden Kohlenstoffstahl geschieht wahrscheinlich keine Veranderung in der chemischen Zusammensetzung des Austenits. Der gebildetc Perlit iibernimmt dessen Kohlenstoffgehalt. Bei der Umwandlung in untereu- tektoiden Stahlen geht eine Ferritbildung der Perlit- bildung voraus. Bei dieser Ferritbildung wird der Kohlenstoff in dem noch vorhandenen Austenit ange- reichert. Dieser Austenit diirfte dabei aber keinen gleichmiissig verteilten Kohlenstoffgehalt bekom- men. Der Gehalt ist wahrscheinlich am hochsten an der Grenze zum Ferrit und am tiefsten in weit vom Ferrit entfernten Gebieten. Dieses Verhiiltnis ist auch deutlich merkbar im Perlit, der sich aus solchem Austenit bildet. Die Struktur des Perlites ist stark abhiingig von der Bildungstemperatur. Wenn Perlit sich in einem untereutektoiden Stahl unmittelbar unter Ai bildet unterscheidet er sich nicht nennenswert in seiner Lamellenstruktur von dem, der sich in einem eutek- toiden Stahl bei gleicher Temperatur bildet. Mit sinkender Umwandlungstemperatur vergroBert sich der Unterschied. Bei untereutektoiden Stahlen vermindert sich bei sinkender Umwandlungstemperatur der Zementitan- teil im Perlit und damit der Kohlenstoffgehalt. Dem- zufolge vermindert sich auch die Menge von proeu- tektoidem Ferrit, wiihrend die Perlitmenge zunimmt. Wenn der Perlit auf diese Weise mit Ferrit ver- diinnt wird, nimmt die Lamellenstruktur mehr und mehr ab. Statt dessen sieht man auf der Schlifflache einen kornigen Zementit in ferritischer Grundmasse. DieZementitlamellen im Perlit sind deutlich dicker in der Nahe des proeutektoiden Ferrites. Oft haben sie auch einen Zementitklumpen an dem Ende, mit dem sie gegen den Ferrit stoBen (Fig. 2). Manchmal liegen mehrere solche Klumpen von Zementit so nahe aneinander, daB der proeutektoide Ferrit und der Perlit durch eine Borte von solchem Zementit getrennt sind. Weiter weg vom Ferrit werden die Zementitscheiben diinner, da der Austenit hier ge- ringeren Kohlenstoffgehalt gehabt hat. Im Metallmikroskop wie auch auf den Lackab- driicken im Elektronenmikroskop sieht es aus, als ob solche diinnen Zementitscheiben in eine Reihe kleinerer Bruchstiicke iibergehen. Zwischen den Bruchstiicken hat man Ferritbriicken (Fig. 2). Die von einem untereutektoiden Stahl isolierten Perlit- und BainitgefUge in dici Kohlenstoffstcihlen 321 Zementitscheiben sind oft viel dicker an dem einen Ende als am anderen. AuBerdem sind die Scheiben zerfetzt und voll Locher. Die isolierten Karbidkor- ner bestiitigen das Zementitaussehen im SchlifTbild. Ein Perlit, der im Mikroskop keinen besondcrs la- mellaren Charakter hat, diirfte aus unvollkommenen Zementitlamellen, mit Fetzen und Lochern, in Ferrit bestehen. Die Frage wie die Perlitumwandlung bcginnt, d. h. ob der Ferrit oder der Zementit keimbildcnd vvirkt, ist besonders untersucht worden. In eutektoidem Kohlenstoffstahl konnte sie nicht entschieden werden. Die kleinsten Perlitgruppen bestanden immer aus mehreren Ferrit- und Zementitlamellen. Die Ferrit- und Zementitlamellen scheinen gleich schnell zu wachsen. In den beiden untereutektoiden Stahlen konnte dagegen folgende Beobachtung gemacht werden. Bei der hochsten Umwandlungstemperatur 700 C sieht man im Metallmikroskop bei hochster VergroBe- rung, daB der proeutektoide Ferrit in ca 30 % der Falle ohne sichtbare Korngrenze in Perlitferrit iiber- geht. Bei niedrigeren Umwandlungstemperaturen reicht das Auflosungsvermogen im Metallmikroskop nicht mehr aus, um Beobachtungen zu machen. Aus den elektronenmikroskopischen Lackabdriicken geht es doch deutlich hervor, daB der proeutektoide Ferrit ohne sichtbare Korngrenze in 20 bis 30 °o der Fiille in Perlitferrit iibergeht. In Perlitgruppen, bei denen man in der Schlifflache keinen solchen korngrenzenfreien Ubergang sehen kann, hiitte doch moglicherweise ein solcher fest- gestellt werden konnen, wenn man die Schlifflache anders gewahlt hatte. Korngrenzenfreie Ubergange diirften darum sehr gewohnlich sein. Es liegt verlockend nahe, aus den gemachten Beobachtungen die SchluBfolgerung zu ziehen, daB der Ferrit keimbildend auf die Perlitumwandlung in untereutektoiden Stahlen wirkt. In alien den Fallen, in denen korngrenzenfreie ijbergange zwischen einer Perlitgruppe und dem umgebenden proeutektoiden Ferrit beobachtet wor- den sind, geht nur ein Ferritkorn ohne sichtbare Grenze in Perlitferrit uber. Bainir.—Dev Bainit scheint sich in gewissen Kristall- ebenen im Austenit auszuscheiden und zu wachsen. Das ist nicht der Fall im Perlit, wo die Perlitgruppen radial wachsen, d. h. unabhiingig von der Orientie- rung des Austenits. Die Ferritlamellen im Bainit auf beiden Seiten einer Austenitkorngrenze bilden immer Winkel miteinander. Bei hoher Umwandlungstemperatur im Bainitbc- reich bildet sich in alien drei untersuchten Stahlen Bainit in Federform an den Austenitkorngrenzen. Bei allmahlich sinkender Umwandlungstemperatur bil- det sich Bainit auch im Inneren der Austenitkorner, dann in Stiibchenform und spiiter in Lanzettenform. Fine Bainilcinheil scheint nie iiber cine Austenit- korngrenze zu wachsen. Auf beiden Seiten einer Austenitkorngrenze habcn die Bainiteinheiten immer verschiedene Richtungen. Der Bainit, der sich in untereutektoiden Stahlen bildet, unterscheidet sich von dem in einem cutekto- iden Stahl. Dieser Unterschied vergrtiBert sich bei Verminderung des KohlenstofTgchaltes. Die Bainit- struktur wird bei sinkcndem KohlcnstolTgehalt gro- ber, wenn die Bildungsbedingungcn im iibrigen die- selben sind. Die Bainitumwandlung wire! in ;illcn drei unter- suchten Stahlen bei alien Temperaturcn mit einer Ferritbildung eingcleitet. Im eutektoiden Kohlen- stolTstahl scheiden sich zuerst mehrere parallcle Scheiben von Ferrit aus. Etwas spiitcr wandelt sich der an KohlenstofT angereicherte Austenit in Ferrit und Zementit um (Fig. 3). In untereutektoiden Stahlen mit geringem Koh- lenstofTgehalt ist die einleitende Ausschcidung von Ferrit sehr reichlich. Diese Ausscheidung geschieht in Widmannstattenanordnung. Zwischen diesen Fer- ritscheiben befinden sich Austenitgebiete, die an KohlenstofT angereichert sind. Diese Austenitge- biete haben oft auch Scheibenform. Proben, die bei diesem Umwandlungsgrad abgeschreckt werden, haben eine lamellare Struktur, die der des Perlits ahnlich ist. Die Scheiben bestehen aber aus Ferrit und durch Abschrecken gebildetem Martensit. Wenn die Umwandlung bei konstanter Temperatur weiter- gehen kann, bildet sich aus dem Austenit Ferrit und Zementit. Der Ferrit scheidet sich dann ohne Korn- grenze an dem schon vorhandenen Ferrit aus. Bei sinkender Umwandlungstemperatur werden die Ferritscheiben diinner. Die Bildung von Zementit folgt schneller auf die Ferritausscheidung. d. h. der Ferritvorsprung wird kleincr. Fine ahnliche Wirkung erhiilt man, wenn der KohlenstolTgehalt im Stahl zu- nimmt. Die Untersuchung zeigt, daB der Ferrit keimbil- dend auf die Bainitumwandlung wirkt und daB der Ferrit die fuhrende Phase ist. Der Ferrit, der sich aus dem Austenit ausscheidet, diirfte nur einen ge- ringen Kohlenstoflfgehalt haben und der Austenit reichert sich bei dieser Ferritausscheidung an Koh- lenstofT an. Die isolierten Karbidkorner sind drcidimcnsionalc Gebilde. Von einer Scheibe mit Fetzen und Lochern wachsen in den Raum Stiibchen und Vorspriinge aus (Fig. 4). LiTERATUR 1. MouiN, Hi 1 1 Kii), Jcrnkontorets Aunalcr 139, 516 (1955). 2. — ihiil. 139, 521 (1955). 3. MoDiN, H ELI RID unci MoDiN, Sten, Jernkontorets Annaler\yi,\%\ (1955). 21 - 568204 Electron Microscopy Selective Oxidation due to the Heating of the Evaporated Film of a-Brass N. Takahashi and K. Mihama Yainanashi University, Kofii, and Japan Electron Optics Laboratory Co., Ltd., Tokyo It is well known that, when an alloy is heated in the air, one of the constituent metals is oxidized selectively [6]: these conditions are determined by the temperature and the pressure [3]. On a-brass, it has been known that CuoO is formed at a low temperature and ZnO is developed at a high tempe- rature [6]. We have prepared the single crystal film of a-brass by evaporation /// vacuo and we have performed the experiment on the selective oxidation by heating in the electron microscope when observing the electron microscopic and the electron diffraction images continuously. (At the place of the specimen chamber of the JEM-5 electron microscope with a four-lens system, the specimen heating or cooling adaptors can be introduced. These adaptors make it possible to study continuously the thermal change of the specimen in the ranges from room temperature to about 1000 C or from about r200 C to the liquid nitrogen temperature respectively.) A single crystal of a-brass has been prepared by successive evaporation of copper and zinc /"// vacuo onto the cleavage surface of rock salt which was kept at about 400''C. To obtain a single crystal of a-brass, copper must be evaporated prior to zinc without breaking the vacuum. «-ii£3i! J-. Fig. 1. Electron diffraction pattern of a-brass. Fig. 2. Corresponding electron micrograph of fig. 1. (3) (2) Fig. 3. Electron diffraction pattern of a-brass with CujO. Fig. 4. Three relative orientations between a-brass and Cu.^O. Single crystal of oi-brass. — According to the equi- librium diagram of Cu-Zn alloy, the a-phase, which is the solid solution in substitution type, exists between 0 and 40 °o in weights of zinc. Electron diflfraction image of the above a-brass film at room temperature is shown in fig. 1 : the plane (100) is always parallel to the cleavage surface of rock salt. Appearances of irrational spots have been observed and discussed by Briick [1], Menzer [5], Laue [4], Gottsche [2], etc. and have been rec- ognized in the present case also. The corresponding electron micrograph is shown in fig. 2. The regions of black and white are due to the interferences of equal inclination of the illuminated electron beam for the specimen and the directions seem to corres- pond to [100] direction or a-brass. Formation of Cii^O. — CujO can frequently be seen even at room temperature. The composite film of copper and zinc may be oxidized on account of the water action at the time of dissolution of rock salt. Heating the specimen at a temperature below 300°C, CuaO appears as shown in fig. 3. Three rela- tive orientations between Cu.>0 and a-brass are observed, that is, (00r)cu,o//(001)a.brass and [010]cu,o//[010],_brass, (D (111 )cu.O /(OOl)^.brass and [1 T0]cu,o//[lT0]a-brass (2) and (111 )cu.O;'/(001 )a-brass and [lT0]cu,o//[l 10]a-brass- (3) These three relative orientations are shown sche- matically in fig. 4. The effect of secondary scattering can be observed and it results in the appearances of irrational electron diffraction spots. The correspond- ing electron micrograph is shown in fig. 5. The inter- ference fringes disappear only due to the heating. A. Fig. 5. Corresponding electron micrograph of Fig. 3. Fig. 6. Electron micrograph of CU2O formation in lower vacuum. Selective Oxidation of y:- Brass 323 Fig. 7. Electron ditVraction pattern ofa-brass withZnO. Fig. 8. Two relative orientations between a-brass and ZnO. Fig. 9. Corresponding electron micrograph of Fig. 7. Fig. 10. Electron micrograph for fmal stage of oxydalion for a-brass. Fig. 1 1. Secondary diffraction effect for Cu.,0. and the small black points are distributed uniformly on the surface of a-brass. Furthermore, when a-brass is heated in some bad vacuum, one may see the agglomerations with the definite external shape as shown in fig. 6, which can be considered as CuaO. Formation of ZnO. — When heating the a-brass over 450 C in the electron microscope, ZnO appears in the state of single crystals. Fig. 7 shows the electron diffraction image of ZnO which was obtained by heating a-brass at 450 C during 30 min. The larger spots correspond to a-brass and the other spots are due to ZnO. The arrangement shows that ZnO develops as a hexagonal prism which has the situation of perpendicularity to the cubic face of a-brass. The diagonal of (0001) of ZnO is parallel to the direction of [100] of a-brass. Fig. 8 shows schematically the two relative orienta- tions between ZnO and a-brass, that is, (000 1 )z„o// (00 1 ),.brass and [ 1! 20]znO, \ [ 1 TOJa-brass (4) and (0001)zno//(00I),.b,.assand [1 I 20]z„o//[l lOUrass- (5) The diffraction spots around the direct electron beam in hg. 7 show the above two orientations. Discussion. — ( 1 ) Relative orientations between t/ie oxides and y.-lvass. The appearance of the orienta- tion (I) can be understood easily by the fact that oxygen atoms enter in the lattice of a-brass to form CU2O. The orientations (2) and (3) seem to devel- op after the formation of the orientation (I): copper atoms traverse into the layer of oxide already pro- duced and at the surface of the oxide layer copper atoms combine with oxygen atoms. One of the authors has earlier shown that the oxidation of brass starts in the direction of the most dense atomic plane, that is (100) plane for a-brass and (III) plane of /^-brass [7]. As in the case of Cu.O, the direction [1120] of ZnO appears parallel to the direction of the most dense atomic plane (110) of a-brass, and the plane (0001) of ZnO appears for the same reason parallel to the plane of (III) for CuaO as the result of the replacement of copper atoms by zinc atoms at the superficial layer. The development of the orientations (4) and (5) can be explained in the same manner for the case of CuoO. It is also explained by the hypothesis of "rotational slip" which was proposed by Wilman [8]. The important role of the direction [110] of a-brass is also explained and its evidence can be observed in the electron microscope as shown in fig. 10, in which ZnO appears with some unknown substance at the final stage of the oxidation of a-brass at Fig. 12. Schematic representation of secondary diffraction effect due to ZnO. The small black and white circles corres- pond to the orientations (4) and (5) respectively. (Cf. fig. 7.) 324 A. W. AGAR AND R. S. M. REVELL 600''C. By means of microdiffraction, the two re- stricted directions can be determined to correspond to the direction of [110] of a-brass. (2) Secondary scattering effect. The secondary scattering effect for CuoO is shown in fig. 11. For ZnO, the irrational diffraction spots due to this effect are distributed more densely than those of CuaO. Fig. 12 shows schematically this effect for only the first quarter of the electron diffraction image of a-brass. (3) Selective oxidation. For low pressure of oxygen and the temperature under the melting point of zinc, the surface of a-brass becomes rich in copper atoms, because the vapour pressure of zinc is greater than that of copper and the mobility of the copper atom is greater than that of zinc. Accordingly, it results in the formation of CuoO at an initial stage. At the high temperature over the melting point of zinc, 419 C, the zinc atom will easily combine with the oxygen atom, because the zinc atom be- comes easily displaceable in the specimen. Further- more, Cu.>0 has some tendency of reduction by the above high diffusion of zinc or the dissociation itself in vacuo. References 1. BRiJCK, L., Ann. Physik 26, 233 (1936). 2. GoTTSCHE, H.,Z. Physil< 134, 517 (1953). 3. HONJO, G.. Proc. Pins. Soc. Jap. 8, 113 (1953). 4. VON Laue, M.. Ann. Pliysik 26, 55 (1936). 5. Menzer, G.,Z. Krist. A 99, 410 (1938). 6. MiYAKE, S., Sci. Papers Inst. Phys. Cheni. Research 29, 167(1936). 7. Takahashi, N., Rikagal / — ^ \f/ — M / — 20 30 kO 50 60 TO 80 90100 Durdimesser in mji Bild 6. Kornungskennlinien im Wahrscheinlichkeitsnetz bei Konverterstaub. Bild 8. Staub aus dem Graef-Rotor. gemessen, so daB als Unbekannte nur noch die Extinktionskonstante k verbleibt.) Die Miesche Theorie (5) beschreibt die Phano- mene der Lichtbeugung und Absorption an metalli- schen und nichtmetallischen Kugein und ist damit auf den industriellen Staubauswurf anwendbar, worauf zuerst Pepperhoff (6, 7) iningewiesen hat (vgl. auch (8)). Nach ihr kann die Extinktionskon- stante aus einigen konvergenten Reihenentwick- lungen errechnet werden. (Nulite Niiherung fiir den Grenzl'all Teilchendurchmesser: Lichtwellenlange < 1 ist das Raleighsche Gesetz.) Die Raleighsche Naherung war jedoch wegen der vorliegenden Teil- chengroBen nicht anwendbar; die Reihen muBten bis zum 8. died ausgewertet werden. In die Berech- nung gehen als StoflF konstanten der Brechungsexpo- nent und der Absorptionskoeffizient ein. Da die Vermutung bestand, daB der Staub beim Austritt aus dem Konverter noch metalUsch sein wiirde, wur- den fiir die genannten Konstanten die Landolt- Bornstein-Werte fijr metalHsches Fe fiir A = 546 m/< wie folgt eingesetzt: Brechungsexponent = 1,436 Absorptionskoeffizient = 1,553. Die hieraus berechnete Extinktionskonstante zeigt Bild 7 in Abhiingigkeit von der KorngroBe des absorbierenden Staubes. Ergebnisse. — In Bild 7 sind neben der theoretisch errechneten Kurve fiir die Extinktionskonstante auch die experimentell gewonnenen Werte eingetragen. Theorie und Experiment stimmen innerhalb der MeB- ♦o »« I I 35 JO 25 20 IS 10 5 Theoretische kurve erre&tnet nach Mie fur A » 5'*6 m/j UMdurchlali / = e'""'^ t Jo c • konientration [an '/cm ^J cf- Metislrecke [cm] 50 wo 150 Teilchendurchmesser d in mpt 200 Bild 7. Extinktion von Fe-Staub im Konverterabgas Ver- gleich zwischen Theorie und Experiment. 5^ 1 s e 10 99.91 1 - 1 I M I 1 "lomaskonvei 99.9 1 A 'rer fjfoef-Rptvr-- yni 1 1 1 1 1 I 1 98, J F \ y I ■^ m ^' Hi / •■.y jf \^ 90 / f / SO 70 / ^J t — y f y 60 '^-T*- Vj"^ - W m^ — — /\\\ V i y A \uyr U) y/\\\ \\\ J 9 j<^ r^ JO \\\\\\ f JW^ •a /\ \\\\\\\ A .^ u /A ''\y / Bradley (3) has been found to be particularly etVective. Howe\er, most of the two-stage replica techniques are somewhat lime absorbing and electron micrographs of single stage formvar replicas can be photographically re\erse printed in order to fa- cilitate interpretation of the correct perspecti\e of the original specimen. Reverse prints are made from a positi\e plate, contact printed from the negati\e. Replicas from the worn surfaces of specimens, taken from a wear machine in which upper and lower annular specimens are rubbed together under con- trolled conditions of speed, load and lubrication, showed clearly the abrasions of the original surface together with deformation, scoring, and build up of material caused b\ the rubbing action. 332 D. SCOTT AND H. M. SCOTT Fig. I. Initiation of Wear Damage on Soft Steel. An area of running track near the start of a test run. Tiie material of the cylinder is E.N. 8 steel of hardness aproximately 200 V.P.N. Plastic deformation of this fairly ductile material is evident, also small areas of surface damage produced by the displacement of material. Magnification 7500. Fig. 2. Initiation of Wear Damage on Hard Steel. Initial damage to the surface of a case hardened E. N.36 steel cylinder of hardness over 800 V.P.N. There is little plastic deformation of this hard material of low ductility and initial damage appears to be by displacement of material and cracking following overstressing of the surface. Magnification 6000. Visual damage by optical microscopic examina- tion after only one revolution of the wear machine appeared to be confined to circumferential scratches, but from the electron micrograph it appeared that, within each individual scratch, plastic flow had oc- curred over a narrow track and a step-wise structure had been formed suggesting a stick-slip method of formation. It is not to be expected that the complex mecha- nism of wear should be easily elucidated for the very process itself eliminates its own initial stages as cumulative action develops towards catastrophic failure. For the study of factors initiating such a failure it is desirable that a single traverse be made of a surface so that the development of surface failure may be followed. For this purpose a special crossed-cylinder machine (1) has been developed at the Mechanical Engineering Research Laboratory in which one cylinder is rotated and a mating cylin- der, at right angles, so traversed that the zone of contact moves along the surface of both cylinders and fresh contact surfaces are continually being presented. The helical track round the cylinder reveals how the damage to the surface builds up with increase of load. Using this machine and cylinders with a fine surface finish of average rough- ness 1 // inch C.L.A. some interesting electron micro- graphs have been obtained of the initial stages of the wear process when using a conventional lubri- cant. The electron micrographs 1-3 are reversed prints of shadowed single-stage formvar replicas. Another important wear problem is fretting corro- sion, the term given to the surface damage that results when there is slight relative oscillatory motion between solid surfaces in contact. The damage mani- fests itself by severe pitting of the surfaces and the generation of considerable amounts of oxidised deb- ris. In conjunction with the research on fretting (8) the electron microscope has been used to study the initial stages of surface damage (fig. 4). Formvar replicas have been taken from taper sections (7) prepared for metallographic investiga- tion of rubbing surfaces. E.M. examination of these shadowed replicas has helped considerably in eluci- dating the metallographic changes in the immediate sub-surface areas of contact. Interesting electron micrographs have also been obtained of the deforma- tion of crystals in the sub-surface area of contacting points of rubbing surfaces. The electron microscope has been used for the examination of debris from fretting corrosion and wear tests, since test conditions giving reasonably low rates of wear require the high magnification of the electron microscope for satisfactory examina- tion. When debris particles are available in the dry state they can be brushed directly on to a prepared formvar film supported on a copper microscope grid and examined in the microscope. Shadowing improves definition and enables a better interpretation to be made of the shape of the particles. When a lubricant is used in a test it is necessary to separate the debris froin the lubricant. This is done by first centrifuging the sample with a solvent for the lubricant. The solvent is replaced during .f?^*'*'w» ■Uk Fig. 3. Effect of E.P. Lubricant on the Initiation of Wear Damage. The build up of a surface film when using extreme pressure additives in the lubricant which seem to form a contaminated surface film in the area of contact. This film may be expected to be of low shear strength and may prevent metallic contact and seizure. The material of the cylinder is again case-hardened E.N. 36 steel. Magnification 15,000. Fig. 4. Initial Stages of Fretting Corrosion. The surface of a highly polished tool steel specimen after 15 seconds fretting (750 oscillations under a load of 10 kg and amplitude of slip 10-3 inches). Overstressing of the surface has mani- fested itself by plastic deformation and displacement of surface material; the mechanism being similar to that observed during the initial stages of wear of hard steel cylinders. Magnification 7000. Mikrotomschnitt- Technik and Mineraluntersuchungen 333 further centrifuging operations by ether until finally a dispersion of the debris in ether is obtained. One drop of this dispersion is then used for examination. For a statistical determination of particle size in wear debris the debris from short duration tests was centrifuged on to a formvar film supported on a microscope grid which in turn, supported by a plug, formed the end of the centrifuge tube. By this method all debris from the test was collected on the formvar covered grid. The debris taken from a steel surface although generally irregular in shape, often contains rod like particles the presence of which is believed to indicate that some of the debris may be produced by fragmen- tation of the pearl ite structure. Discussion and conclusions. — The micrographs ob- tained show that the electron microscope can, with the use of a formvar replica technique, reveal small scale features of interest in the study of wear and can yield experimental evidence for postulated theo- ries of the mechanism of wear. With very hard steel there is little plastic deforma- tion, and overstressing of the surface appears to initiate damage by fine cracks and the displacement of material by microscopic adhesion of the contacting asperities (fig. 2), possibly in the manner suggested by Bowden and Tabor (2). With carbon steel in the annealed or softened state considerable plastic rough- ening of the contacting surface is evident (fig. I), resulting in a pattern of ridges and grooves with mating surfaces probably conforming exactly, as suggested by Ming- Feng (6). Although named freiling corrosion, presumably owing to the accumulation of oxidised debris, the initiation of surface damage appears to be similar to the onset of wear of rubbing surfaces, fine debris produced by the initial damage being the start of cumulative abrasive action. The authors wish to acknowledge the assistance they have received from their colleagues in the Mechanical I ngincering Research Laboratory in providing, for elec- tron microscope study, specimens which have been sub- jected to various forms of wear test. In particular, Mr. A. A. Milne has provided guidance regarding the test conditions and the bearing of the results on current knowledge of friction and wear. I he work described has been carried out as part of the programme of the Lubri- cation and Wear Division of the Mechanical Lngineering Research Laboratory and the paper is published by permission of tlie Director. References 1. Barwell, F. T. et al.. Proceedings Fourth World Petro- leum Congress. Sect. VI C Preprint 3. 1955. Brit. Patent No. 732447. 2. BowDHN, F. p. and Tabor, D., The Friction of Lubrica- tion and Solids. University Press, 1950. 3. Bradley, D. £., J. Inst. Metals 83, 35 (1955). 4. Halliday, J. S., Proc. Inst. Mecli. Eni;rs. 169, No. 38 (1955). 5. Menter, J. W., J. Inst. Metals 81, 163 (1952). 6. Ming-Feng, I., J. Appl. Pins. 23, 1011 (1952). 7. MooRF, A. J. W., Mctallingia 38, No. 224 (1948). 8. Wright, K. H. R., Proc. Inst. Mech. F.ni^rs. (B) IB, No. 11, 566-574 (1952-53). Anwendungen (der Mikrotomschnitt-Technik auf elektronenmikroskopische Minerakintersuchungen G. Pfefferkorn, H. Themann und H. Urban Medizinische Elektionenmikroskopie, Universitcit Miiiisterj Westf. JjLSHER fiihrte man elektronenmikroskopische Mine- raluntersuchungen an kleinen Einkristallen bzw. Aggregaten im Durchstrahlungsverfahren durch, wiihrend Kristalloberfiachen mit verschiedenen Ab- druckverfahren studiert wurden. Es fehlte bislang die Moglichkeit, Mineralaggregatc in Querschnitten zur Feststellung der dreidimensionalen Textur zu untersuchen. Aus diesem Grunde haben wir versucht, das bei biologischen Objekten gebriiuchliche Verfahren der Einbettung und Schnitt-Technik auf kristalline Kor- per anzuwenden. Es zeigte sich, daB porose Aggregate, in deren Hohlraume das Einbettungsmedium eindringen kann, nach der Polymerisation schon mit Glasmes- sern schneidbar sind. Noch giinstigere Erfolge erziel- ten wir mit dem Diamantmesser cines Ultra-Mikro- toms nach Moran der Firma Leitz. Die Kristallag- gregate wurden vor der Einbettung zur Besciligung von Feuchtigkeitsspuren im Vakuumtrockenschrank erwarmt oder mit absolutem Alkohol \orbehandelt. Die bisherigen Untersuchungen fiihrten zu folgen- dcn Ergebnissen: Glaukonit, ein charakteristisch griines kalium- und magnesiumfiihrendes Eisenaluminiumsilikat mit Glimmcrstruktur, liegt in der Natur in gerundeten, hiiufig eiformigen Kornern vor. die, wie die licht- mikroskopische Untersuchung im Gesteinsdiinn- schlifT zeigt, aus polykristallinen Aggregaten beste- hen. Mit der Schnitt-Technik kann man jet/t den Aufbau einzelner Korner genauer studieren. Schon lichtmikroskopisch zeigen die Schnittc in verschic- 334 H. C. CORBET AND J. WOLFFES Bild 1. Glaukonit. Bild 2. Knochen. Bild 3. Kaolin (Zettlitz) aus 5 mm Hohe sedimentiert. Bild 4. Kaolin (Zettlitz) aus 10 cm Hohe sedimentiert. denen Bereichen eine gewisse Orientierung. Elektro- nenmikroskopisch dagegen ist man auch in der Lage, die Form und Aneinanderlagerung der einzelnen Bliittchen genauer zu untersuchen. So findet man verschiedene Formen und Aggregatbildungen. Uber- wiegend lagern sich die Einzelteilchen zu biiumchen- formigen Aggregaten zusammen. Hierbei ist es noch fraglich, ob dies eine zufilUige Aneinanderkigerung oder ein dendritisclnes Wachstum ist. Die Aggregate bilden ein verfilztes Netzwerk (Bild I). Um Hohl- raume finden sich sowohl konzentrische wie auch radialstrahlige Anordnungen der einzelnen Aggre- gate. In groBen Bereichen ist eine anniihernd paral- lele Anordnung der Glaukonitteilchen zu beobach- ten. Neben diesen natiirlichen Mineralaggregaten kon- nen wir kiinstliche Aggregatbildungen, z. B. die Se- dimentation von Tonmineralien verfolgen. Wir ha- ben einen derartigen Versuch mit Zettlitzer Kaolin durchgefuhrt. Sedimentation aus geringer Hohe von einigen mm ergibt eine fast regellose Lagerung der einzelnen Kaolinitplattchen (Bild 3), wahrend eine Sedimentation aus groBerer Hohe (iiber 10 cm) eine Paralleltextur nach sich zieht (Bild 4). Als poroses kristallines Haufwerk sind auch die Knochen aufzufassen, deren anorganischer Anteil aus Hydroxylapatitkristiillchen besteht. Dieser wurde durch Behandlung der Knochen mit HjO.. und Extraktion in organischen Losungsmitteln isoliert. Danach wurde eingebettet und geschnitten. Aus friiheren Untersuchungen mit Durchstrahlung klein- ster Kornchen oder deren Kohleabdriicken sowie von Oberfliichenuntersuchungen kennt man die Anordnung der einzelnen Kristallchen. Mittels Dunnschnitten laBt sich der lockere Bau der anorga- nischen Substanz mit verschiedener sich kreuzender Textur in benachbarten Bereichen bestatigen ( Bild 2). In den jetzt von Einbettungsmaterial erfiillten Hohl- raumen lag friiher der organische Knochenbestand- teil. So laBt sich die Raumbeanspruchung von organi- schem und anorganischem Anteil leicht feststellen. An diesen Schnittbildern erkennt man deutlich, wie eng der organische und anorganische Anteil in klein- sten Bereichen miteinander verfiochten sind. Durch orientierte Schnitte in verschiedenen Richtungen laBt sich so ein dreidimensionales Bild des Knochen- aufbaus gewinnen. Es steht zu erwarten, daB sich die Anwendung von Schnittuntersuchungen auf kristalline Objekte zu einem wesentlichen neuen Hilfsmittel fi.ir Minera- logie und Technik entwickeln wird. The Use of a Freeze-(drying Technique in the Investigation of Sodium-Montmorillonite by Electron Microscopy H. C. Corbet and J. Wolffes Koiiiiiklijkc'j Shell- Laboratoriiini, Aiusteickini (N. V. De Bataafsche Petroleum Maatschappij) 1 HE electron microscope has proved to be a valu- able tool for investigations into the properties of systems consisting of montmorillonite and water. Notably the particle form and size as well as the cohesion of the particles have been studied by elec- tron microscopy. At first, however, the nature of the clay mineral under investigation, montmorillonite, caused some difficulties, for this mineral consists of articles which have a large extent in two directions (several thousand A), whereas the dimension in the third direction is very small (some tens of A). In the examination of such clay-water systems the technique of preparation normally followed consists in placing a drop of the suspension on a supporting film and allowing the water to evaporate in air. This method has several consequences that exert an unfavourable influence on the investigation. Freeze-drying Technique in Investigating Sodiiim-Montinorillonitc 335 In the first place it is impossible under these circumstances to maintain the proper dispersion and concentration of the montmorillonite particles. The latter are frequently present in the suspension in the form of very thin sheets. When the drop dries, these sheets tend to be fiattened or distorted. Further, the actual distribution of the particles tends to be obscured. In air-dried preparations the particles are sometimes observed as separate units, but sometimes also as sheets of considerable dimen- sions. The latter might have been composed of a number of smaller particles, but it is impossible to distinguish these clearly. If, moreover, we are interested in the structure of the clay skeleton in the suspension, the investiga- tion by electron microscopy becomes extremely diffi- cult. The above considerations induced us to try and find a new technique of preparation that would obviate these objections. Application of the principle of freeze-drying has turned out to lead to good results. In this technique the specimen is rapidly frozen, after which the water is removed by evaporation of the ice formed. Any change in the cohesion of the clay particles is minimized by freezing as quickly as possible. By pumping the water vapour from the solid phase a slow and even removal is ensured. Consequently, here again a good retention of the cohesion of the clay particles may be expected. For the application of this technique there are several possibilities. For instance, the drop of suspen- sion may be either placed on a specimen carrier covered with a supporting film or dried on a narrow slit, without supporting film. The cohesion of the particles has proved to be such that a 200 /< slit can be bridged by the clay skeleton, if the concentra- tion is not too low. Freezing, too, can be effected in different ways provided it is done quickly. In our experiments the sample was immersed in liquid ni- trogen. The ice is evaporated in vacuo, for instance in the shadowcast installation. Usually this can be done without any special low-temperature provision. The solid phase is then maintained by withdrawing heat of evaporation of the ice from the specimen. If required, mounting on a refrigerated metal block may assist in maintaining the solid phase for a longer time. The specimen thus freed of water cannot be sub- mitted to any mechanical treatment, but can be in- vestigated directly in the electron microscope. A number of suspensions of montmorillonites have been investigated by this method. The con- Fig. 1. Montmorillonite, prepared by freeze-drying. centrations varied within the range of 1-0.005 "o clay mineral in water. Concentrations of 1 ",, and higher make specimens opaque to the electron beam. At the lowest concen- trations the supporting film cannot be dispensed with because there is not enough material to span the slit in the carrier. The results obtained with this method can be seen in fig. 1. To obtain maximum information the slereotech- nique has been employed. From these stereophoto- graphs the spatial structure of the montmorilU^nite skeleton can be clearly visualized. The conclusion to be drawn from these photomicrographs is that the clay particles form a three-dimensional network ol sheets and ribbons. This conclusion is in agreement with the impression pre\iotisl\ obtained on the basis of the macroscopic behaviour of these suspensions. Furthermore, we can conclude from these photo- graphs that difTerent types of bonds between the particles occur and that an edge-sheet bonding as was postulated by other investigators is certainl\ not very conspicuous. The authors thank the management of the Konink- lijkc Shell-Laboratorium, Amsterdam, for permission to publish the above results. Zur Kenntnis der Glasoberflache E. Bruche und H. Poppa Physikalisches Laboiatoiiiim, Mosbach In den letzten Jahren hat sich unser Laboratorium u. a. mit der Oberfliiche und Oberflachenschicht des Glases sowie den sich daran kniipfenden wissen- schaftlichen und technischen Problemen beschaftigt. Dazu erwies sich das Elektronenmikroskop als sehr niitzlich, besonders wenn man es bei kleinen Ver- groBerungen und groBem Gesichtsfeld benutzt. Wir haben nach Ausbildung eines besonderen Abdruckverfahrens mit Abdruckfolien bis zu V2 mm Durchmesser teils in weniger als lOOfacher Ver- groBerung gearbeitet, wobei die groBe Schiirfen- tiefe des Elektronenmikroskops mehr Aussagen als das Lichtmikroskop erlaubt. VergroBerungen iiber lOOOOfach wurden kaum benutzt. Zur Kontrolleund quantitativen Erganzung wurde neben dem Licht- mikroskop das Interferenzmikroskop herangezogen. Die Ergebnisse, die von uns (zum Teil auch von G. Schimmel) erzielt wurden, sind in einer Reihe von VerofFentlichungen niedergelegt (4-9); weitere Arbeiten sind in Vorbereitung. Die Untersuchungen betreffen dieFragen der Plastizitat desGlases( 10), der Erweichung bei hohen Drucken (3), der Spriinge und ihrer Bedeutung fur den SchieifprozeB (11). Es wurde untersucht: die Harte des Glases im Kern- material und an der verwitterten Oberflache, der PolierprozeB mit seinen teils chemischen Vorgangen sowie manche andere Frage. Es seien hier einige Punkte herausgegriffen und nach dem Gesichtspunkt der Kraftwirkung und des Verhaltens des Glases kurz dargestellt. Vorgcinge bei grofien Kraften. — Glas kann im Gegensatz zu Metallen keine groBen Spannungen aufnehmen. Unmittelbar im AnschluB an den linea- ren Teil im Spannungs-Dehnungs-Diagramm tritt nach einem nur sehr schmalen Obergangsbereich Bruch ein. Besonders klein ist die Zugfestigkeit, wahrend die zulassige Druckfestigkeit mit ■^ 9000 kg/cm- mehr als zehnmal so groB ist. Wenn z. B. eine Glasflache durch einen aufgesetz- ten und unter Druck fortgeschobenen Glasstab kraftig beansprucht wird, so reiBt das Glas hinter dem sich bewegenden Stab infolge der hohen Zug- spannung auf und es bilden sich Risse quer zur Bewegungsrichtung (Abb. 1). Fuhrt man dagegen eine Diamantspitze leicht iiber eine Glasflache, so dringt sie ein und zieht einen Graben. Dabei driickt sie das Glas nach beiden Seiten auseinander. So entsteht an der Grundlinie des Grabens der fur das Glasschneiden wichtige Sprung ins Innere senkrecht zur Oberflache. Bei starkerer Beanspruchung treten in den Seitenbezirken schollenartige Ausbriiche durch die Wirkung von Spriingen auf, die unter der Glasoberflache fort- laufen und mehr oder minder entfernt von dem Graben die Oberflache erreichen (Abb. 2). Eine ahnliche Beanspruchung wie bei dem Ziehen Abb. 1. Querrisse in einer Spur bei kriiftiger Oberflachen- beanspruciiung. Vergr. 6200 . Abb. 2. Ausspriinge am Rande einer tiefen Ritzspur. Vergr. 5700 X . Abb. 3. Geschliffene Oberflache. Vergr. 8100 . Abb. 4. Eindruck eines Vickers-Diamanten mit Sprungen (Prap. R. Schulze). Vergr. 1600 . Ziir Kenntnis der Glasoherftachc 137 4 Abb. 5. Glasbruch mit Lanzcltspriingen. Abb. 6. Ritzspiir mit lierausgedriicktem Glas. Abb. 8. Kreissysteme mit Diamantspitzc bei verschiedener Kraft auf bcdampflc Giasflachc izcsciiricbcn. eines harten Gegenstandes iiber die Glasflache tritt bei dem AbroUen kugelformiger barter Gebilde auf der Glasobcrfiache auf, wie sie beim SchleifprozeB stattfindet. Es bilden sich Risse im Glas, die die Oberflachenschicht zerfurchen und zu Ausbriichen fiihren, so daB ein ausgepragtes Gebirge die Folge ist (Abb. 3). Sehr deutlich zeigt sich die Empfindlichkeit des Glases gegeniiber Eindriicken auch bei der Hiirte- messung nach Vickers. Selbst bei kleinen Kraften konnen leicht Uberbeanspruchungen eintreten, die zu Rissebildungen fiihren, so daB die MeBergebnisse von zweifelhaftem Wert sind (Abb. 4). Smekal hat daher auch gegeniiber den neueren Messungen von Ainsworth (1) Bedenken erhoben. SchlieBlich seien noch die interessanten Lanzett- sprunge ervvahnt, die beim Bruch von Glas auftreten und ein Studienobjekt fur die Kraftwirkungen und Spannungsverhiiltnisse im Glas sind (Abb. 5). Es zeigen sich auBer den ausgepriigten Lanzettspriingen noch vielerlei andere aufschluBreiche Formen. Vorgcinge hei kleinen Kiiiftcn. — Achtet man darauf. daB keine Uberbeanspruchung dcs Glases auftritt, d.h. daB Risse vermieden werden, so tritt die Plasti- zitat des Glases in den Vordergrund. Diese plastische Deformation bildet die Voraussetzung von einwand- freien Hiirtemessungen nach Vickers wie von Ritz- hiirtemessungen. Fahrt man mit einer Diamantspitze oder mit Dia- mantstaub mit kleiner Kraft iibcr cine Glasober- fliiche, so treten unter der Diamantspitze nur Drucke auf, die das plastisch werdende Glas fortdriingen, ohne auBerhalb dieses Erweichungsgebietes zu hohe, d. h. sprungerzeugende Spannungen hervorzurufen. Das Glas wird seitwarts aus den Spuren gedriickt 22 — 568204 Electron Microscopy (Abb. 6). Bei sich kreuzenden Spuren zeigen sich Durchdringungsfiguren der Ritzgriiben, die die Querschnittsformen zu bcurtcilcn eriauben. Die Glasplastizitiit und das bcsondere Verhalten bei Ritzbeanspruchung erlaubt, Ritzhiirtemessungen relativ weit in das Makrogebiet bis zu Kriiftcn von einigen 10 g durchzufiihren, ohne daB die Gefahr des Springens besteht. Man erhiilt charakteristische parabelformige Kurven bei Auftragung der Ritzspur- breite als Funktion der Kraft, die auf den Diaman- ten wirkt. Tragi man die Ritzspurbreite im doppelt logarithmischen Diagramm iiber der Kraft auf. so erhiilt man eine Gerade (Abb. 7 rechts). Besonders aufschluBreich werden solche Messungen, wenn man zu sehr feinen Kraften von 50-500 mg iibergeht. Die erforderlichen feinen Ritzspuren lassen sich mit einem Mikroschreibcr durchfiihren (Abb. 8) und schlieBen sich den Messungen im Makrogebiet an (Abb. 7 links). Ihre genaue Analyse erlaubt Schliisse m A c (/) ■s s- i 0.1 .>'J>^ --'' -- ^ .ei,.^ j Oi 10 Kraft Ping IOC Abb. 7. Spurbreilc s in Abhangigkcit von der Rilzkralt P fiir Fenster- bzw. Tafelglas. Sprungfrcic Ritzspuren links im Mikrogebiet, rechts im Makrogebiet. 338 E. BRUCHE UND H. POPPA f* V ^i\-. i 0 Abb. 9. Weitgehend poliertes Glas mit resllichen Schleif- griibchen. Vergr. 600 . Abb. 10. Sprungsystem bei anpoliertem grobgeschliffenem Glas. Vergr. 450 x . iiber die Harte in verschiedenen Tiefen unter der Oberfliiche. So wurde bei einem mehr als 10 Jahre alten Glas gefunden. daB die Harte bei 60 nyt Tiefe unter der Oberflache sprunghaft in den dop- pelten Wert iiberging. Es bedeutet das, daB die oberste Schicht infolge der atmospharischen Ein- fiiisse erweicht war. Bei noch kleineren Kraften spielen elastische Er- scheinungen eine Rolle, wie sie auch zu der Deutung von gewissen Unstimmigkeiten bei Vickers-Mes- sungen schon herangezogen worden sind (Schulze). Dei EinfiuB von Feuchtigkeit und chemischen Einfliissen wird besonders deutlich bei der Behand- lung des Glases mit FluBsaure. Auch hier wurden Untersuchungen angestellt, die die Grubchenbildung und ihre Unterschiede bei verschiedenen Glassorten (Libbey-Owens-Glas, Fourcault-Glas) zeigten. Chemische Einflusse an der Oberflache sind auch fiir den technisch wichtigen ProzeB des Glaspolierens von groBer Bedeutung. Hier wurden zunachst einge- hende Untersuchungen iiber den PolierprozeB an Proben aus der optischen und Spiegelglas-Industrie durchgefiihrt. Es konnte gezeigt werden, daB das Glas laufend wahrend des Polierprozesses abgetragen wird (Abb. 9). Sind die untersten Tiefen der Schleif- griJbchen erreicht, so wird trotzdem in der Praxis die Abtragung dariiber hinaus fortgesetzt, bis auch die unter diesen Tiilern herunter reichenden Spriinge, die von der hohen Beanspruchung des Schleifprozes- ses zuriickgeblieben sind, beseitigt sind (Abb. 10). Es wurde auch nach der besondersartigen Ober- fliichenschicht gesucht, die eine bewegliche Ober- flachenhaut darstellen soil (2). Unter der Annahme, daB das Glas nach der Druckentlastung eineZeit von 10^* Sekunden zur Erstarrung braucht, kann man sich eine solche bewegliche Schicht plausibel machen (7). Ob sie in dieser Art wirklich vorhanden ist, konnte nicht entschieden werden. Dagegen ist aus vielen Einzelbeobachtungen sichergestellt. daB sich an der Oberflache eine besonders geartete Schicht befindet, die vermutlich mit chemischen Vorgangen zwischen Glas, Poliermittel und Wasser in Zusam- menhang steht. In einer neuen Untersuchung, die noch nicht abgeschlossen ist (Poppa), werden einzelne Schleif- griJbchen wahrend des Polierprozesses mit dem Elektronen- und Interferenzmikroskop in ihrer Ge- staltsanderung verfolgt. Es gelingt so, die allmahliche Abtragung quantitativ zu erfassen. Die verschiedenen Untersuchungen haben gezeigt, daB das Elektronenmikroskop — in seiner Aussage durch das Interferenzmikroskop erganzt — ein wert- volles Instrument zum Studium der Eigenschaften und Prozesse ist, die bei Glas zu beobachten sind. Sie fuhren damit zu einem tieferen Verstandnis des stofflichen Verhaltens sehr barter, sproder Korper und geben niitzliche Hinweise zu technisch wichtigen Prozesser. LiTERATUR 1. AiNSWORTH, L., Trans. Soc. Glass Techiwl. 38, part I 479-500, part II 501-535, part III 536-547 (1954). 2. Beilby, G. T., Proc. Roy. Soc. 72. 218-225 (1903). 3. Bridgman, p. W. und Simon, I., /. Appl. Phys. 405-413, 24(1953). 4. Bruche, E. und Poppa, H., Glaslech. Ber. 28. 232-242 (1955). 5. — ibid. 29, 183-192 (1956). 6. — Silikatteclmik 6, 378-384 (1955). 7. — Z. angew. Physik 8, 486^92 (1956). 8. Bruche, E. und Schimmel, G., Glastech. Ber. 11, 239- 247 (1954). 9. — Z. angew. Physik 7, 378-385 (1955). 10. Klemm, W. und Smekal, A., Naturwiss. 29, 688-690 (1941). 11. Preston, F. W.: Trans. Opt. Soc. London 23. 141-164 (1922). Elektronenoptische Untersuchung natiirlicher Opale in Verbindung mit dinerenlialthermischen und ronlgenographischen StLidien Liber die Polytypie des SiO^* A. Maas MiiH'ialogisc/i-Pclrologisc/u's Inslilul der Uiiiveisitiit Bonn iinil Zentiallaboraforiiim fiir angewanihe (Jbermikroskopie der Univcrsitiit Bonn Gegenstand der Untersuchung sind die in der Natur vorkommenden in verschiedenem MaBe in Eintrocknung und Kristaliisation betindlichen Geie der Kieselsiiure bis hin zum Chalcedon. Der struktu- rclle Ordnungszustand der eigentlichen Opale ergab sich nach den alteren lichtmikroskopischen Unter- suchungen als ijberwiegend amorph, der des Chalce- dons und Achats als ijberwiegend kristallin (1, 10). In neuerer Zeit konnte dann mit Hilfe rontgeno- graphischer Methoden, erstmals von Levin und Ott ( 1 2), in einer groBeren Anzahl Opalvarietiiten das Vor- liegen verschiedener kristallisierter SiO.-Modifika- tionen neben einem offenbar groBen amorphen An- teil nachgewiesen werden (4, II. 14). Die erhaltenen Rontgen-Diagramme lieBen jedoch weder beini nor- malen Debye-Scherrer-Verfahren noch bei den be- deutend hoher aufiosenden und empfindlicheren Methoden nach Guinier und Bragg- Brentano (mit Ziihlrohrregistrierung) eine eindeutige Unterschei- dung der verschiedenen SiO.-Modifikationen zu. Ungekliirt blieb auch der Umwandlungsmechanis- mus der SiO^-Modifikationen in den natiirlichen Opalen. Die Aufgabe der vorliegenden Untersuchungen (20) bestand darin, die Natur der in den verschiedenen Opalvarietiiten vorliegenden SiO.-Modifikationen und ihren Bildurgs- und Umwandlungsmechanismus mittels neuerer, leistungsfiihigerer Verfahren zu untersuchen. DasZustandsdiagramm des SiO... — Nach dem idea- len Zustandsdiagramm von Fenner (2, 3, 4, 5) fiir das System SiO. (5). wonach 8 verschiedene Modifi- kationen auftreten, und zwar a- und /i-Quarz, a-. /3- und }'-Tridymit, a- und /^-Cristobal it. Schmelze. Die Umwandlungstemperaturen im System SiO-.. die in neuerer Zeit insbesondere fiir die keramische Industrie ein groBes Interesse erlangt haben. wurdcn von Fenner durch Messung der beim Erhitzen auf- tretenden Wiirmetonungen nach der kurze Zeit vor- her von W. C. Roberts-Austen (17) eingcfuhrten differentialthermischen Methode bestimmt. In zahl- reichen spateren VerofTentlichungen anderer Auto- ren (15. 21) konnten die Ergebnisse Fenners zwar im wesentlichen bestiitigt werden, jedoch crgaben sich oftmals bei den natiirlich vorkommenden SiO,- * Teilergehnis einer Dissertationsarbeit, Mineralog. Petro- log. Inslitut der Universiliit Bonn (1956). — Herrn Professor Dr. A. Neuhaus sei fur die tJberlassung des Themas sowie wertvolle Anregungen und Diskussionen gedankt. Modifikationen erhebliche Abweichungen von den erwiihnten Umwandlungstemperaturen. insbesondere beim Tridymit und Cristobalit. Auf das anomal breite Intervall der a-//-Umwandlung des Crisloba- iits (180-270 C) hatte bereits Fenner hingewiesen (4). In neuesten Arbeiten hat O. W. Florke (7, 8, 9) nachgewiesen. daB Trid\mit und im ailgemeinen auch Cristobalit I'ehlgeordnete Strukturen sind und daB die bei den SiO.-Modifikationen auftretenden Anomalien hierauf zuriickzul'uhren sind. Tnd\mit ist wahrscheinlich nur beim Vorhandensein von Fremdionen realisierbar (6). Wiihrend man bei den vorstehend nur kurz skiz- zierten Untersuchungen zur Uberpriifung des SiO..- Zustandsdiagramms mciglichst rcinc amorphe oder kristallisiertc Phasen des SiOj zu Grunde legte, sind die Verhiiltnisse bei den natiirlichen Opalen wesent- lich komplexerer Natur. Es konnte erwartet und durch eigene Voruntersuchungen bestiitigt werden. daB der Kristallinitiitszustand und der Anteil der einzelnen SiOo-Moditikationcn in Opalen von ihrer Vorgeschichte abhiingig sind und daB fiir ihren differenzierten Nachweis die Empfindlichkeil der bisher benutzten Untersuchungsverfahren gesteigert werden muB. Diffeicntialthentio- Analyse. — Die Wiirmetonungen sind bei den Umwancikingcn der SiOj-Mndifikationcn \on auBerordentiicher KIcinlicil (maximal cinigc cal g). Fs muBte also cine apparative Anordnung gefunden werden. welche die thermischen EtTekte geniigend emplindlich registriert. Die da/u enlwickelte DIA-Anlage bestehl aus einem elektrisch behci/lcn \criikal slehcnden Ofen, in desscn Tcmperalurfeki in radials>mmetrischer \'crtei- lung 13 sehr dunnwandige Pt-Zylinder als Probcnbchiillcr angcordnet sind. Fs konnen so iinter \ollig gleicharligen Vcrliiiltnissen bis /ii 10 Proben gleicli/cilig untersiicht werden. Eine Rcgelaniage nach dem Transduktor-Prin/ip (\ormagnetisierte Drosselspulc. Siittigungswinkclstcue- ning) mil eingebauter getrcnnter Spannungs- und Strom- stabilisierung ermoglicht eincn sictigen und sircng repro- du/icrbaren Tcmpcraluranstieg des Ofcns. Die Registrie- rung der mit Thermo-Flementen gemessenen Warmeto- nungcn der Probe erfolgt nach \erslarkung der Thernio- spannungen mittels Phott>/ellenkonipensator dureh einen Punktschreiber. der in zweisekiindlichcr Punktfolge die Kurven \on 6 McBstcllen glcich/eitig auf/eichnet. Die Kmptindlichkeit der Regislrierung betriigi 0.04 C • 0.01 C" pro mm Skalenliingc (13). Rd/ilaen/einstrnkliir - Lntcr.staliioinen. Hier/u wurde ein Bragg-Brentano-Goniomeler nach Pmf. Berthold mit Ziihlrohrregistrierung der intcrfcrcn/en benut/l. Durch Vcrwcndung einer Feinfokusrohre in \erbindung mit einem hochspannungs- und heizspannungsscitig stabili- siertcm Netzgeriit und durch spe/ielle apparati\e Mali- .568204 340 A. MAAS Abb. 1. Holzopal (Rosenau/Siebengebirge). Kollodium- abdruck, Bedampfung SiO, VergroBerung elektronenoptisch 11300 , GesamtvergroBerung 50 000 . nahmen konnte die Empfindlichkeit der Anordnung so weit gesteigert werden, daB die sehr intensitatsschwachen Interferenzen der Opale auch bei groBen Glanzwinkeln noch legistriert werden. Es wurden folgende Opalvarietaten untersucht: Holzopal — Stenzelberg/Siebengebirge und Rose- nau/Siebengebirge; Feueropal — Zimapan/Mexiko; Milchopal — Baldiestro/Piemont; ferner Chalce- don — Island. Die differentialthermischen Untersuchungen erga- ben, daB bei den genannten Opalen stets mehrere SiOz-Modifikationen nebeneinander vorliegen, und zwar mit verschiedenem Kristallisationszustand (KorngroBe, Fehlordnung) und unterschiedlichem Anteil der Modifikationen. Kristallisationsgrad und Art der vorliegenden Phasen lassen sich durch geeignete thermische Behandlung verandern. — Ausfiihrliche Veroffentlichung erfolgt in Kiirze (s. (20)). Die rontgenographischen Ergebnisse konnten die difFerentialthermischen Befunde bestatigen. Bezeich- nend ist die starke ,,Verbreiterung" der Interferenzen und die ungewohnliche Hohe der Untergrund-Kurve, die auf den erwarteten hohen Dispersitatsgrad hin- weist und den Einsatz elektronenoptischer Metho- den nahelegt. Die elektronenoptischen Untersuchungen wurden mit dem elektromagnetischen Siemens-Ubermikroskop 100 d durchgefuhrt. Feueropal und Milchopal, die unter glei- chen Bedingungen im Edelstahlmorser zerkleinert wur- den, lassen in elektronendurchlassigen Randzonen groBer Bruchstiicke kleine Partikel erkennen, die in eine struk- turlos erscheinende Masse eingebettet sind. Urn diese Strukturen //; situ ohne vorangegangene mechanische Beanspruchung zu erfassen, wurden von sehr gut polierten ebenen Oberflachen verschiedener Opale, die zuvor mit Hilfe der DTA und der Rontgenfeinstruktur-Analyse untersucht worden waren, KoUodiumabdriicke herge- stellt, deren Negativ-Seiten kombiniert senkrecht und unter einem Winkel von 30° mit SiO bedampft wurden. Durch die Senkrechtbedampfung konnten Artefakte ver- mieden werden, die auftraten, wenn die Abdriicke nur schrag bedampft wurden. Die Oberflachen des untersuchten Feueropals, Milchopals, Chalcedons und Holzopals erwiesen sich im allgemeinen elektronenoptisch, abgesehen von vereinzelten Schleifspuren sowie beim Schleifen angeschnittenen Hohlraumen, als weitgehend struk- turlos, wobei der Chalcedon ein Extrem an Feinheit darstellt. Nur bei dem stellenweise bereits licht- mikroskopisch inhomogen erscheinenden Holzopal konnten vereinzelt hexagonal ausgebildete Partikel von ca. 250-max. 1 000 A GroBe festgestellt werden. Bedingt durch die im Vergleich zu den ubrigen Opa- len geringe Harte des Holzopals treten ausgepragte Schleifspuren auf, in denen Partikel der gleichen GroBenordnung sichtbar werden (Abb. 1). Da die differentialthermischen und rontgenographischen Untersuchungen der Opale ergeben hatten, daB in ihnen verschiedene SiOa-Modifikationen nebenein- ander vorliegen, wurde versucht, diese durch diflfe- rentielle Atzung sichtbar zu machen. An Einkristallen wurden solche Atzversuche von PfefTerkorn (16) beim Kalkspat vorgenommen. Seine Untersuchungen iiber den Realbau dieser Kristalle zeigen, daB bei geeigneter Anatzung in den Atzgru- ben von Kalkspat, die durch Storstellen des Kristall- gitters hervorgerufen werden, scharfkantig idio- morph ausgebildete Rhomboeder-Blocke von ca. 300 A GroBe als kleinste Bausteine des Realbaues sichtbar werden. In Voruntersuchungen wurde eine geeignete Kon- zentration und Einwirkungsdauer der als Atzmittel verwandten FluBsaure HE ermittelt. (Uber die Los- lichkeit von SiOa-Modifikationen in HE siehe (19).) Sie ergaben 0,5 "o HE bei 30 sec. Atzdauer. Die hochglanzpolierten, mittels Kollodiumab- driicken gereinigten und anschlieBend mit 0,5°oiger HE geatzten AnschlifFe von Feueropal, Milchopal und Holzopal zeigen iibereinstimmend Partikel mit mehr oder weniger morphologisch ausgebildetem Habitus, die in eine mikroskopisch strukturlos er- scheinende Substanz eingebettet erscheinen und oft- mals zu Nestern gruppiert auftreten. Die in den Aufnahmen schwarz erscheinenden Partikel wurden durch den Atzvorgang aus der Oberflache herausge- lost und mit der Abdruckfolie abgehoben. Eine gleichartige Erscheinung wurde von A. Schrader (18) bei angeatzten StahlschlifTen beobachtet. Die GroBe dieser herausgelosten Partikel betragt in Ubereinstimmung mit der GroBe der in der Ober- flache verbliebenen Partikel zwischen ca. 300 und 500 A. Der untersuchte Chalcedon wurde bei dieser schwachen Atzung kaum angegriffen und lieB ledig- lich vereinzelt groBe ovale Atzstellen von einigen /^ GroBe erkennen, iihnlich den elektronenoptischen Bildern von mit HE angeatzten Glasern. Dieser Diamond Cleavage Surfaces 341 Befund steht in Ubcreinstimmung mit den rontgeno- graphischen Ergebnissen, wonach der untcrsuchte Chalcedon wesentlich aus a-Quarz besteht. Beim Erhitzen der Opale tretcn. wie bereits cr- wiihnt, Strukturveriinderungen auf, die differcntial- thermoanalytisch und rontgcnographisch verfolgt wurden. Es wurde nun versuchl, auch elektronenop- tisch einen Einblick in die sich hierbei vermutlich vollziehenden mikroskopischen Veriinderungcn zu erhalten. Hierzu wurden die AnschlifTe unter gleich- artigen thermischen Bedingungen wie die DTA- Pulverpriiparate im DTA-Ofen erhitzt und abge- kiihlt. Beim Vergleich der elektronenoptischen Auf- nahmen der thermisch vorbehandelten Feuer-, Milch- und Holzopale mit den Aufnahmen der entsprechen- den unbehandelten Proben sind deutliche Habitus- und GroBenveranderungen der Partikel zu erkennen. Die Untersuchungen iiber diese Strukturveriinde- rungen von Opalen nach thermischer Behandlungsind noch im Gange. Auf Grund des elektronenoptischen Befundes diirften sich die AnomaUen des Rontgendiagramms in der Weise erkliiren lassen, daB die Verbreiterung der Interferenzen durch die extreme Kleinheit der kristallisierten Partikel bedingt und die anomale Hohe der Untergrund-Kurve auf einen groBen An- teil amorpher bzw. stark fehlgeordneter Substanz zuriickzufiihren ist. Dem Leiter des Zentral-Laboratoriums, Herrn Dr. K. E. Wohlfarth-Bottermann danke ich flir die Betreuung und stete Unterstiitzung der elektronenoptischen Unter- suchungen. — Herrn Prof. Dr. H. Ginsberg und Herrn Dr. Hiiltig bin ich fiir ihr 1 ntgcgcnkomnicn b/gl. der Benutzung eincr neucn Kontgcnfcinstruktur-Apparatur der Vereinigten Aluminiumv\erke AG, Bonn, zu Dank verpflichtet. LiTERATUR t. BuTsrmi, O., \'erh. luiturh. meet. Ver. Heidelberg, N. F. 6, 287 348 (1899). Mit .3 Tafcln. 2. Fenner, C. N., J. Wash. Acad. Sci. 2. 471 (1912). 3. — Amer. J. Sci. 36, 331 (1913). 4. — Z. anorg. Chem. 85, No. 1/2, 133 (1914). 5. — Diagramm entnommcn aus Eitel, E., Physik. Chemie der Siiikatc, AuH. 1941. 6. Florke, O. W.. Fortschr. Mineral. 32, 33 (1953). 7. — Naturwis.s. 41, 371 (1954). 8. — Neue.s Jalirh. Mineral., Monatsh. 217 (1955). 9. — Ber. deut. keram. Ges. H. 12, 369 (1955). 10. Hein, H., Neue.s Jahrh. Mineral., Monatsh. Bcilagcband XXV, 182-231 (1908). 11. HoFMANN, U., WiLM, D., und Endi I I , K., /.. (ingew. Chen,. 47, 539 (1934). 12. Levin, J. und Ott, E., Z. Krist. 85, 305 (1933). 13. Maas. a., Forl.schr. Mineral. 34, 54 (1956). 14. Marchet, a., Silznngsher. Akad. \Mss. Wien, niaih.- naturw. Kl. I, 562 ff. (1939). 15. Van Nieuwenburg, C. J., Ber. deut. keram. Ge.s. 9, 228 (1928); Rec. trav. chint. Pays-Bas 47, 627 (1928); ibid. 48, 402 (1929). 16. Peefferkorn, G., Unischau 18, 557 (1951). 17. Roberts- Austen, W. C, Proc. Roy. Soc. 49, 347 (1891). 18. SrHRADFR, A., Z. H/.V.S. Mikroskop. 60, H. 6 7 (1952). 19. ScHWARZ, R., Z. anorg. Chemie 76, 422 ( 1912). 20. Teilcrgebnisauseiner Disscrtationsarbeit, Mineralogisch- Petroiogisches Institut der Universitat Bonn (1956). 21. Tool, A. Q., Phy.s. Rev. 53, 945 (1938). A Reflection Electron Microscope Study of Diamond Cleavage Surfaces M. Seal Research Laboratory for the Physics and Chemistry of Surfaces, Department of Physical Chemistry, University of Cambridge The existence of two types of diamond, which differ in a number of physical properties, was first reported by Robertson, Fox, and Martin (7). These workers classified diamonds as type 1 or type II on the basis of their ultra violet and infra red absorptions, photoconductivity, and birefringence. Type I diamonds have an absorption at 8 // in the infra red, absorb strongly below about 2900 A in the ultra violet, show but little photoconductivity, and are birefringent. Type II diamonds have no 8 n absorption, are transparent down to about 2200 A, have good photoconductivity, and are optically iso- tropic. The problem of the two types of diamond has been studied extensively in the last twenty years, and the two types have been sub-divided into further classes. The present position of this work is described in articles by Sutherland, Blackwell. and Simeral (8), and by Champion (3). It has also been found, using optical multiple beam interferometry, that there are difTerences in the nature of the cleavage surfaces of diamonds of the two types (9, 10). The cleavage surfaces of type I diamonds are, on a micro-scale, rough and almost conchoidal in nature; those of type II diamonds are very much smoother. In this paper an account will be given of results obtained from a reflection electron microscope exa- mination of cleavage surfaces of diamonds of both types. Diamond cleaves along octahedron {ill} planes: it was checked by electron diffraction that the planes studied had this orientation. The classification of the diamonds was made on the basis of their 342 M. SEAL Fig. 1. Reflection electron micrographs of a type 1 diamond cleavage surface, m^ ~ 2000 : m^ 250 Fig. 2. Reflection electron micrograph of a type II diamond cleavage surface. /«i ~ 3000 : /^u ~ 400 ultra violet absorption. The diamonds were cleaned in nitric acid, followed by distilled water. A layer of silver about 500 A thick was evaporated onto the surface to prevent charging of the specimen in the electron beam. The examination was carried out using the Metropolitan Vickers EM3 electron micro- scope modified for use in reflection (4). Reflection electron micrographs typical of dia- monds of the two types are shown in figures I and 2. The conchoidal nature of the type I cleavage and the much smoother nature of the type II cleavage are apparent. The surface roughnesses may be estimated from measurements on the micrographs. The type I cleavage surfaces are so rough, however, that it is difficult to give an accurate value of the change in height which corresponds to a particular shadow, since the local slope of the surface on which the shadow falls is not known. Certainly, there are fairly abrupt changes in height of a micron or more on the part of the surface shown in figure I. Other parts of the surface of this diamond were consider- ably rougher and were too rough for examination by reflection electron microscopy. The type II sur- face shown in figure 2 is considerably smoother. The steps seen here are probably all less than 1500 A in height, and many of the regions between the larger steps are flat to within 300 A. Other regions of type II diamond cleavage surfaces are similar in appear- ance, but there are occasional large steps of perhaps a micron in height. These large steps differ from those on type 1 diamonds in that they separate relatively smooth regions. The lines running in the direction A on figure 2 are of interest. It is believed that they correspond to lamellae in the diamond. Type II diamonds are known to have a lamellar structure. Robertson, Fox, and Martin (7) reported the existence of fine laminations parallel to (ill) planes on type IT cleav- age surfaces. These laminations were about 10 /t apart, but some may have been as narrow as I //. Such laminations have since been observed by Ramachandran (6) (thicknesses 10 to 100//) and by Custers (5) (thicknesses of the order of 5 //). The lines A on figure 2 are also spaced about 5 // apart. Similar lines are visible on most of the other micro- graphs of type II diamonds. Type I diamonds do not, however, always have an appearance similar to that of fig. 1. Regions of a type I cleavage surface have been found which appear in the reflection electron microscope indistinguish- able from a type II cleavage. Micrographs have also been obtained showing regions where the two types of cleavage join along quite a sharp and well defined boundary. It is suggested that, in these cases, the type I diamond has regions of type II material within it. It is probable (2) that type I diamonds have a mosaic structure and are for the most part highly dislocated, whereas type II diamonds consist of lamellae of relatively perfect material. It would then be quite reasonable for there to be inhomo- geneity in the perfection of individual diamond crys- tals. Indeed. Ahearn ( I ) has shown that there may be inhomogeneities in the ultra violet (2550 A) trans- mission and alpha particle bombardment conduction of individual diamonds and suggests that some dia- monds may contain regions of both type I and type II material. Furthermore, inhomogeneity in the ultra violet transmission (at 2850 A) of a diamond has recently been observed by the author. The observa- tions on diamond cleavage surfaces thus support Ahearn's views. Conclusion. — Cleavage surfaces of diamonds of types I and II appear difterent in character and roughness when examined in the reflection electron Chemical Decomposition of Crystals of Explosive Materials 343 microscope. Type I diamonds show a conchoidal type of cleavage surface with local changes in height of several microns: type II diamonds show much smoother cleavage surfaces which do, however, have a considerable amount of tine structure ranging up to perhaps 1500 A in depth, with a few larger steps of a micron or more crossing the surface. Regions similar in appearance to type II cleavage surfaces have been found on a type I diamond. It is believed that this is due to real inhomogeneity in the diamond. The results are in agreement with those obtained by other workers and, in particular, with those obtained by Wilks (II) in a recent interferometric study of diamond surfaces.' ' The above results are in agreement with some observa- tions made by Custers (unpublished) on the inhomogeneity in the optical properties of diamonds. He observed that, under ultra violet irradiation, some diamonds fluoresced strongis in parts and were non-fiuorescent in others. These diamonds showed local ditVerences in transparency below 3000 A. One diamond in particular showed inhomogeneity in the visible part of the spectrum, being partly blue and partly white with a clear boundary between the two regions. The author thanks Dr. F. P. Bowden, C.B.E., F.R.S., for his advice and encouragement, and Messrs. Industrial Distributors (1946) Ltd. for a research grant and for the loan of the diamonds. RtFERENCES 9. 10. AiitAKN, A. J., Phys. Rev. 96, 828 (1954). Champion, F. C, Proc. Roy. Sue. A 220, 485 (1953). — ibid. 234, 541 (1956). CottRiNE ^-Pkai r, J. S., Mfntfr, J. W., and Sial, M., Proc. 3rd Intern. Conf. Elect. Microsc. London 1954. CusTFRS, J. F. H., Re.secirch 4, 131 (1951). Ramachandran, G. N., Proc. Iiulian Acad. Sci. A 24, 65 (1946). Robertson, R., Fox, J. J., and Marmn, A. L., P/iil. Trans. A 12,1, 463 (1934). SuTHFRLAND, G. B. B. M., BiA( KvvrLi, D. E., and Simeral, W. G., Nature, Loud. 174, 901 (1954). ToLANSKY, S., Microstructures of Diamond Surfaces. N.A.G. Press. London, 1955. Wilks, E. M., Dissertation, University of London. 1952. — Private communication (1956). The Direct Observation in the Scanning Microscope of Chemical Reactions J. H. L. McAusLAN and K. C. A. Smith P.C.S. Laboratory, Depart iiient of Physical Cheniistry, and Engineering; Laboratories. University of Cambridge This paper contains an account of the application of the scanning electron microscope (2) to an explo- ratory study of the slow chemical decomposition of single crystals of explosive materials. The materials so far investigated in this way have been the metallic azides and lead styphnate. Although normally considered explosive these materials will decompose slowly if the temperature is kept below about 200 C. Their explosive properties have been extensively studied and it is known that reaction starts from small centres of I0~-'-10"^ cm, termed hot-spots, and grows to burning and detonation by a self-heating mechanism. It was considered that an electron-microscopical study of the slow thermal decomposition would give information on reaction mechanisms possible in the hot-spot size range. Study in the light microscope had already shown large scale break-up effects but until Bowden and Singh (1), with a standard replica method, and Sawkill (3). with direct viewing of very thin crystals, there was little work done in this field. The problem of specimen heating in an electron beam is a well known one to electron microscopists and Sawkill in fact used this effect to decompose crystals during direct observation. This was not entirely satisfactory as he suspected from his results that the primary action of bombardment by high- energy electrons was responsible for some of the change. Attempts have been made to observe explo- sive crystals in the conventional reflexion instrument without success, the high intensity of bombardment resulting in premature and uncontrolled decomposi- tion. Because of the low intensity obtaining in the scanning instrument the above difficulties may be largely overcome and it has been found possible to observe the genuine effects of thermal treatment SILVER PLATE. NICHROME ELEMENT. MICA 5INDANY0, Fig. I. Hotplate for scanning electron microscope. 344 J. H. L. MCAUSLAN AND K. C. A. SMITH Fig. 2 a. Needle crystal of silver azide — partly decomposed by heat. Fig. 2 b. End of needle crystal of lead styphnate. Note crystal- lographic break-up due to dehydration. directly in the microscope by means of a hot stage (fig. 1). Bombardment of the specimen in the conventional and scanning,' microscopes. — In order to extract a given amount of information from the image in the electron microscope a certain minimum number of electrons, as determined by quantum and fluctuation considerations, must interact with the specimen. This minimum number will be determined, among other factors, by the quantum efficiency of the trans- mission process between the specimen and the brain of the observer; the quantum efficiency being defined as the ratio of the number of electrons or quanta associated with the point of minimum quanta trans- fer to the number falling upon the specimen. The bombardment of the specimen is smaller in the scan- ning instrument mainly because of its superior quan- tum efficiency. The conventional instrument operating in trans- mission will have a quantum efficiency, during re- cording of the image, not far short of unity since virtually all of the electrons passing through an element of the specimen will fall on the correspond- ing element of the plate and be recorded. However, when observing the image directly the quantum level falls because of the low efficiency of the fluorescent screen and the small angle subtended at the screen by the eye. It may be estimated that under these conditions the quantum efficiency is about 4 10~^ (4), that is, for every single visual stimulus which the observer receives, about 250 electrons must pass through the specimen. The quantum efficiency of the conventional in- strument operated in reflexion is very much worse because, under the usual conditions of operation, only about 1 in 10^ of the electrons hitting the speci- men will pass through the objective aperture. This factor will, of course, vary widely according to the nature of the specimen, the observation angle and the objective aperture, but for the purposes of this paper will be assumed constant at 10"-'. Thus the quantum efficiency will be of the order of 10"* when recording and lO^-lO"** when viewing directly. In the scanning instrument an electron multiplier is used to collect the electrons passing from the specimen and the input of the multiplier may be arranged to collect electrons of all energies over a very large solid angle. Since the total emission ratio may, under certain conditions, exceed unity the transfer efficiency at this stage of the process may well approach unity. The high gain of the electron multiplier ensures that the quantum level is Chemical Decomposition of Crystals of Explosive Materials 345 maintained at all subsequent stages so that the quantum efficiency in the scanning instrument may approach unity for all modes of operation. For reflexion operation, therefore, the total num- ber of electrons which the specimen must receive will be greater in the conventional instrument by roughly a factor of I0-' for recording and 10^-10" for direct observation at ecjual integration times of observation. In the conventional instrument the integration time is usually that of the eye (0.2 second) whereas in the scanning instrument a long persistence screen is used. This gives an additional factor of 10 or so in favour of the scanning instru- ment which means that the total current passing to the specimen during direct viewing in reflexion may well be higher in the conventional instrument by a factor of 10'. So far, only the total dosage of specimen and the current passing to the specimen have been consid- ered; the current density incident on the specimen is also of importance. The maximum current density which can be brought to bear on the specimen is limited by the electron gun and will be the same in both instruments. Other things being equal, there- fore, the rate of transfer of information in the scanning instrument in which the elements are imaged sequentially, will be much lower. However, for reflexion operation, the improved quantum effi- ciency and longer integration time of the scanning instrument offsets the reduced time efficiency, and the information content in the directly observed pictures will therefore be much the same. An additional factor which reduces the heating effect is that in the scanning instrument only the area actually observed is irradiated by the electron beam whereas in the conventional instrument the irradiated area will be appreciably larger than the field of view, even with a specimen screening aper- ture. Also, in the scanning instrument, the primary effects of electron bombardment are reduced by the use of a comparatively low accelerating voltage (-15 kV). Results. — Silver azide melts at 230 C and detonates at 350 C but at temperatures between 120 C and 230°C it decomposes slowly, giving off nitrogen and leaving metallic silver. It is soluble in ammonium hydroxide and specimens were prepared by recrys- tallisation from this solution on to silver discs which screw into the hot stage. Fig. 2 a shows a needle shaped crystal of silver azide which has decomposed from one end, because of the better contact with the hot plate at that end. Decomposition is not considered to be complete at this stage, but it can be seen that the crystal is break- ing up into pebbles of approximately 0.3 f^i in size. Complete decomposition would reduce this size to about 0.2 // which is the size most common in other larger crystals heated in this way. It is not clear yet whether these small pebbles are formed by aggregation of silver by a diffusion process into preferred sizes, as found by Sawkill, or whether the initial size is determined by a fine structure of dislocations and decomposition to pure silver then follows. It is evident, however, from later work that free surfaces assist in the mechanism of decomposi- tion. This is shown by the fact that the top and side faces of a larger crystal decompose more rapidly than does the bulk, when heated from the underside. A break-up of larger dimensions I -^ 2// has been observed which is attributed to a polymorphic phase change known to take place at 180 C (Sawkill and Duke, independently, private communications). The pieces produced in this way are angular in shape and very different from those in Fig. 2 a. When this has occurred it is difficult to decompose the crystal further because of the poor thermal conduc- tivity through the porous mass, and high radiation losses from the large surface area. The second material, lead styphnate monohydrate, was examined to determine the effects of dehydration which takes place above 120 C. Chemical decom- position at this temperature is very slow and the major effects can be attributed to the loss of water which is tightly bound in the material. Fig. 2 b shows one stage of the dehydration process, the crystallographic nature of break-up is very apparent. Break-up in this way by polymorphic change or dehydration, which can be considered a physical rather than a chemical change, has some bearing on the study of reaction kinetics by pressure-time curves. Since the rate of reaction normally depends on surface area any mechanism not initially related to the chemical change but which causes an increase in surface area will assist in the later stages of the reaction, and must be considered in the final analysis. References 1. BowDEN, F. P. and Singh, K.. Proc. Roy. Soc. A 111, 22 (1954). 2. McMuLi.AN, D., Proc. Inst. Elec. Engr.i. 100 (II). 245 (1953). 3. Sawkill, J., Proc. Roy. Soc. A 229, 135 (1955). 4. Smith, K. C. A. and Oatlev, C. W., Biii. J. Appl. Phys. 6, 391-399 (1955). liber das Teilchenwachstum sublimierbarer Stoflfe, dargestellt am Beispiel des Zinksulfids W. MiJLLER und W. Jaenicke Lahoratoriiim fiir Elektioneiimikroskopie und histilut fiir physikalische und Elektiochemie, Technische Hochschule, Karlsruhe Chemisch gefalltes ZnS besitzt fiir industrielle Ver- wendung (z. B. fiir Leuchtschirme) im allgemeinen eine zu kleine mittlere KorngroBe (Leverenz, 1949). Die Keimbildungsgeschwindigkeit fiir die Entste- hung des ZnS aus der chemischen Fallung ist nam- lich sehr groB, so daB relativ kleine Primarkristallite entstehen. Fiir verschiedene Faliungen ergaben sich 21, 25 und 16 m// als mittlerer Korndurchmesser. Sie wurden aus Zinksulfatlosung mit Natriumsulfid- losung gefallt. Die Bedingungen fiir ein Kornwachs- tum wurden untersucht und auch versucht, den Wachstumsmechanismus aufzudecken. Prinzipiell liiBt sich ein Wachstum durch eine Warmebehandlung erreichen. Reines Zinksulfid (er- halten durch mehrtagige Dialyse) wachst beim Glii- hen bei 600 C nach 30 Minuten von 20 auf 47 m// mittlerer KorngroBe an. Verlangerung der Gliih- dauer bringt kein wesentliches Wachstum mehr. Versetzt man das dialysierte Zinksulfid mit NaCI, so wachsen die Kristallite unter genau den gleichen Bedingungen betrachtlich starker an. Die mittlere KorngroBe betriigt nun etwa 200 m//, also etwa das 4,5fache wie beim reinen ZnS. Nach dem Gliihen wurde der Fremdsalzgehalt wieder herausgewaschen. Gegliiht wurde immer in StickstofTatmosphare, um Oxydbildung zu verhindern. Im einzelnen hiingt das Teilchenwachstum und auch die Teilchenform von einer Reihe von Parame- tern ab wie z. B. von den Ausgangssubstanzen fiir die Fallung, den Fiillungsbedingungen, der Art und Konzentration des Fremdsalzes und von der Tempe- ratur und Dauer der Gliihbehandlung. Diese Para- meter wurden variiert. Es wird im folgenden auf einige Teilergebnisse naher eingegangen. PnttS) «' m Abb. 1 zeigt einige Verteilungskurven iiber die KorngroBen der ZnS-Kristalle in Abhiingigkeit von der Gliihzeit. Die ZnS-Proben wurden hierzu aus ZnS04 mit Na.S gefallt, dialysiert und dann mit NaCl versetzt. Zur Ermittlung der Verteilungskurven wurden jeweils einige Hundert bis Tausend Teilchen vermessen. Die Kurven sind auf gleiche Massen normiert. Als Gliihzeiten wurden 3, 10, 15. 20, 30, 40 und 60 Minuten gewiihlt. Die Gliihtemperatur betrug 600 C. Von 20 Minuten Gliihzeit ab bleibt bei dieser Temperatur die Verteilungskurve prak- tisch die gleiche. Abb. 2 zeigt die mittleren Korndurchmesser, welche aus diesen Verteilungskurven gewonnen wurden. Zum Vergleich sind die Werte fur reines Zinksulfid mit eingetragen. ZnS existiert in 2 Modifikationen, der kubischen Zinkblende und dem hexagonalen Wurtzit. Die Um- wandlungstemperatur fiir den Ubergang der Blende in Wurtzit liegt oberhalb 1000 C. Wie die geringen Unterschiede in den Madelungskonstanten (1,6381 und 1,641) zeigen, liegen jedoch die Gitterenergien beider Modifikationen nahe beieinander. So er- scheint es moglich, daB Fremdsalzzusatz wie z. B. NaCl bewirken kann. daB die Kristallite wahrend ihres Wachstums bei wesentlich tieferer Temperatur zum Teil Wurtzitstruktur annehmen. Dies gilt je- doch nicht allgemein. So wird bei LiCl oder NaaSOj (D ® Hacflstum der miHleren Korngro/ie you InS * CI" in Hbh van der dluhieit I© lunahme des Wurzitanteils ♦,• relativ lur Zinkblende (D mittlere Korngrofie von reinem Zn S (J) mittlere KorngrdPe von ZnS * SO,," ® (D (5) ^00 250 rjm^ Abb. 1. Verteilungskurven, auf gleiche Massen normiert, tiir die Gluhzeiten 3, 10, 15 ,20, 30, 40, 60 Minuten. 10 EO 30 kO 50 bO Minuten k 8 12 16 20 ?it Abb. 2. Mittlere Korndurchmesser in Abhangigkeit von der Gluhzeit und Wurtzitanteil zu Blendeanteil in Abh. von der Gliihzeit. (Gestrichelte Kurve gilt fur untere Zeitskala.) Calcium Hydroxide and Calcium Carbonate 347 Abb. 3. ZnS in Gcmisch aus 30 "„ LiCl 70 "„ NaCI gegliiht. a) 5 Minuten, b) 10 Minulen, c) 22 Minulen. Vergr. 60000 . als Fremdsalz oder bei einem Gemisch aus NaCI und LiCl in weit geringerem Masse Wurtzit gebildet als bei NaCl-Zusatz allein. Bei reinem ZnS macht sich erst oberhalb 900 C ein (Jbergang von der kubisch dichtesten zur hexagonal dichtesten Kugel- packung bemerkbar. Fiir die Deutung des Mechanismus der Wurtzit- bildung ist es wichtig, dali der Wurtzitanteil (neben der Blende) nur solange zunimmt, als noch die mittlere KorngroBe wiichst. Das bevorzugte Wachstum in kubischer oder hexagonaler Form driickt sich auch in der Mor- phologie der Kristalle aus. Blende entsteht bei im wcsentlichen zweidimensionalen Wachstum (Zusatz von LiCl oder einer Mischung von 30 % LiCl + 70 % NaCI Oder Na.SO, -Zusatz). Wurtzit tritt neben Blende bei drcidimensionalcm Wachstum auf. Das Stereobild zeigt dcutlich die kugelige Gestalt der Kristallite zum Teil mit ebcnen Begrenzungen. Um zu kliiren, ob das Wachstum durch Sublimation erfolgt, wurde mit NaCI versetztes ZnS direkt auf den Objckttriigerblenden bei 600 C mehrere Male hintereinander gegliiht. Zwischendurch wurdcn Je- wells eiektronenmikroskopische Aufnahmen herge- stellt. In cincr Parallelversuchsreihe wurden fiir die gleichcn Gesamtgliihzeiten von 0, 5, 10 und 22 Mi- nuten jeweils frisch priipariertc Proben verwendet. Die Ergebnisse beider Versuchsreihen sind identisch und lassen praktisch kein Wachstum erkennen. Man kann daraus schlieBen, daB das Wachstum nicht iiber die Dampfphase geht. Bettet man dieselbe Substanz in cine Schmelze aus NaCI ■ LiCl ein, deren Schmelzpunkt bei 580 C liegt und gliiht wie vorher, so vergroBern sich die Kristalle erheblich (Abb. 3). Es ist demnach anzunehmen, daB das Wachstum mit einer Diffusion des ZnS oder der beiden lonen durch die Fremdsalzschicht verkniipft ist. Dafiir spricht auch der Befund, daB erst bei mindestens monomolekularer Bedeckung derZnS-Kristallite mit Fremdsalz das Wachstum fiir eine gegebene Gliih- temperatur optimal wird. Die Schmelztemperatur des NaCI liegt mit 800 C betrachtlich iiber der Gliihtemperatur von 600 . Man kann jedoch annehmen, daB in einer Schicht von wenigen Molekiilen Dicke bei 600 das Kristall- gitter bereits bis zu einem quasifliissigen Zustand gestort ist. Unter Umstanden liegt auch eine Schmelz- schicht aus einem Gemisch von NaCI undZnCL, vor. Es kann wohl auch hier angenommcn werden. daB das Wachstum durch Diffusion der Partikel durch eine geschmolzene oder doch stark gitter- gestorte Fremdsalzschicht erfolgt. Ausfiihrliche Ergebnisse soUen spiiter veroffent- licht werden. LiTERATUR Leverenz, H. W., Science 109, 183 (1949). Electron-microscopical Investigations of Calcium HycJroxide and Calcium Carbonate G. SCHIMMEL Baitcllc I list it lit, Frankfurt a. M. The Battelle-Institut, Frankfurt /Main, has been requested to examine the structure of calcium hydroxide and calcium carbonate by means of the electron microscope. There are only two older publi- cations regarding this Held by O. E. RadczewskI, H. O. Muilcr, and W. Eitel (1939) which belong to the earliest papers on electron-microscopical in- vestigations. The authors stated that calcium hy- 348 G. SCHIMMEL Fig. 1. Monocrystalline calcium carbonate. Magnification : 90,000. Fig. 2. Calcium liydroxyde obtained from aqueous suspen- sion. Magnification 80,000. droxide precipitates from clear calcinated water in the form of smooth, compact spherulites (hemis- pheres). During the test they permitted the calcinated water to dry in the air on collodion films. Since then, these results have become the general scien- tific property of the lime industry; they are discussed, for instance, in the third volume of the book Zement-Cheniie by Dr. Hans Kuehl. We re- peated the investigations and obtained, at first, exactly the same photographs. When heating their prepara- tions by an intensive irradiation, the above-men- tioned authors observed a change in the spherulites, which they interpreted to be a transformation to calcium oxide. It has to be borne in mind, however, that it was not possible with the equipment used at that time to make electron diffraction diagrams. We, too, have observed corresponding changes, which we have analyzed by means of diffraction. Hardly any differences are detectable if the same spot is examined prior to and following an intensive irradiation. The proper diffraction patterns demon- strate, however, that the precipitations prior to the irradiation were of an amorphous nature and fol- lowing the irradiation of a crystalline one. Surpris- ingly, however, an evaluation of the diffraction pattern revealed calcium carbonate in a calcite structure to be present, rather than oxyde. Fig. 1 shows a monocrystalline section at which the hex- agonal symmetry of the calcite is clearly noticeable in the diffraction pattern. Radczewski, Miiller, and Eitel (1939) obtained spherulitic calcium carbonate as a residuum from dried calcium-bicarbonate solutions; however, they did not recognize the identity with the precipitations of calcinated water form.ed during drying in air. A light-microscopical investigation of spherulites up to a diameter of 4 //, which had been skimmed off from the surface of a calcium-hydroxide solution, shows that the amorphous modification of the carbonate is stable even if it is present in a larger particle size in contrast with the above publication. I have to thank Dr. Ney of the Technische Hoch- schule MiJnchen for informing me of the existence of such large spherulites on the surface of liquids. When solutions still containing undissolved hy- droxide in suspension were dried, crystalline calcium carbonate, exhibiting characteristic growth phenom- ena, always formed during air drying. To avoid the hydroxide to transform into the car- bonate during the drying process, the specimen grid was placed into the vacuum immediately after it was covered with the calcium-hydroxide solution. During this process, the hydroxide precipitated while drying to very small crystals of different shapes, but it was always crystalline. An amorphous modification of the hydroxide has never been found. Of industrial importance is mainly the question in which form the hydroxide is present in the commercial dry hydrate and the lime paste. The amount of lime contained in the mortar is determined by the plas- ticity of the latter, rather than by the desired strength of the solidified material. This plasticity, however, will depend on the shape and size of the crystals. The dry hydrate prepared from fog obtained by means of ultrasonics contains numerous hexagonal scales and fragments thereof. Almost without any excep- tion, these crystals in their commercial form are strongly agglomerated. A photograph of crystals of lime paste dried in vacuo is shown in fig. 2. The same oblong crystals are found if the water of the lime paste is substituted by alcohol and the alcohol suspension is dried. The following preparation technique proved to be very advantageous: The lime paste spread out in a thin manner is frozen rapidly by means of liquid nitrogen; subsequently it is dried in vacuo in which a carbon replica is prepared immediately after drying. This method is particularly suited for precipitation proc- esses aiming at a carbonation. All photographs were taken at the Battelle-Institut, Frankfurt/ Main, with the Elmiskop I of the firm Siemens, using a voltage of 80 kV. I am very grateful to the "Bundesverband der Deut- schen Kalkindustrie" (Federal Association of the German Lime Industry) for suggesting these studies, for supplying the lime samples, and for the permission to publish the photographs made in the course of these investigations for the "Bundesverband". Further Investigations of Photographic Development by Means of the Electron Microscope E. Klein Wissemchaftlich Photographisches Lahoiatoviiini, Agfa Aktiengesellschaft fiir Photofabrikation, Leveikiisen The object of the present paper is a study of the fine structure of silver halide crystals in photo- graphic emulsions and of developed silver visible by examinations with an electron microscope. It is impossible to do this by making direct electron micrographs of silver halide or silver, for the follow- ing two reasons: ( 1 ) Scattering of the electrons on the crystals (silver halide or silver) is so great that, in the crystal thicknesses d encountered in practice (100 A * d •=- 20,000 A) no electrons are able to penetrate the specimen in unscattered form and thus it is impossible to determine surface quality or condition. The entire object is reproduced in uniform high opacity com- pared to the surrounding area and no deviations in contrast are apparent within the object. (2) The risk of change in the objects during their exposure to the electron beam is especially great with direct observation of silver halide or silver. For one thing the silver halide undergoes severe pho- tolysis, i.e. decomposition into silver and free halide, and for another the mobility of the lattice units in the silver halide and silver is extremely high in an intense beam of electrons, leading to a process similar to melting and frequently to partial evaporation as well. Very intense radiation and the marked rise in temperature associated with it can force the reac- tions to the point where the crystals are completely evaporated (and partial condensation on cold parts in the environment of the crystals ensues). The phenomena mentioned under point 2 have already been discussed in detail in an earlier report, with references to the literature already available in this respect (5). For the purpose of the present investigations, it was absolutely necessary to employ a replica method since the actual object is not then exposed to the electron beam. Our experience has shown that the best method is that of direct carbon evaporation on to the objects devised by Bradley (I, 6). Exposure of the silver halide crystals during the brief period of evaporation with carbon (a few sec- onds) causes no recognizable change in the struc- ture of the crystal, as has been established by com- parison with much greater exposures (contrary to our earlier assumptions (5)). In most cases carbon evaporation is carried out at an angle of 45 with a rotating object which, admit- tedly, results in loss of shadow effect but the carbon replicas are more uniform and stable, i.e. they can also be kept thinner (6). Previous tests by Konig and Helwig's (7) carbon replica process and that of Konig and Knoch (8) also produced good results. However, if it is desired to use this process (in which the carbon replica is formed by glow discharge in a hydrocarbon atmosphere) to obtain a carbon rep- lica which is stable without being hardened by electron bombardment, the replica must be kept relatively thick. On the other hand, hardening by the electron beam would produce just the effects mentioned above in the case of silver halide and sil- ver. Thus the Bradley process would appear to be more suitable for examination of photographic grains. Fig. 1 illustrates a carbon replica of undeveloped silver bromid crystals. In this case evaluation of the stereo photographs reveals mainly a rounded cube shape of the grains. In fig. 2 the form is that of plate-shaped triangles, hexagons and isolated tetra- hedrons. The latter shapes in particular can only be detected by the replica method. Generally speaking, silver chloride crystals are mainly cube-shaped. On the other hand the silver iodide has a quite different appearance and crystallizes in a hexagonal lattice as /^-AgJ. In general hexagonal pyramids are found (fig. 3). It is not possible to detect any characteristic surface structure or other sub-structures in the crystals — a point mainly of importance with the silver bromide and chloride crystals. It is true that steps are found on individual crystals but in the main the crystals possess a well-formed, smooth surface and any large clefts or fissures are not to be seen. If the angle of evaporation is very flat it is possible for so-called "dune effects" to occur when the object is not rotated. A tine structure can be seen, which, however, only occurs on the outside of the replica, as could be established from evalua- tion of the stereo micrographs. This is, therefore, only an apparent structure. In order to determine varying quality of crystal structure within a crystal, different kinds of etching were carried out, of which only one with sulphite solution will be mentioned. In this case a terrace-like structure of the pyramid-shaped crystal (fig. 4) is revealed. Thus, by means of etching, it is possible to estab- lish those zones of a crystal which tend to be of better solubility due to their disturbed lattice struc- ture, although no sub-crystallites could be found. Due to the high resolving power obtained by the carbon replica process (approx. 50 A) it must be concluded that the sub-structure discovered by Hed- 350 E. KLEIN Figs. 1-3. Carbon replicas of silver halide grams. /Ig. J, ammonia emulsion, silver bromide (carbon replica). Fig. 2, boiling emulsion (non-ammonia), silver bromide (carbon replica). Fig. 3, silver iodide emulsion (carbon replica). Fig. 4. Slightly etched silver bromide grains. ges and Mitchell (4) with a light microscope on large silver bromide single crystals is quite definitely brought about by clefts of less than 50 A in size; in other words, that the inner surfaces making their appearance in this respect are less widely spaced. It would appear, therefore, that the sub-structure is only the result of displacements and spaces in the magnitude of one or a few lattice intervals.' Of course it is possible for sub-structure also to be present in the silver bromide crystals of normal photographic emulsions. Usually it is assumed (3) that inner surfaces in the crystal could arise from crystal growth due to grain conglomeration. However, examination of va- rious types of emulsion precipitations with an elec- tron microscope have confirmed that conglomeration is of importance only in the first stage of the preci- pitation, i.e. with the extremely fine grains. The growth of grain is due entirely to accumulation of the newly forming silver halide on grains already present and to Oswald ripening." Figs. 5-6 show a comparison of developed grains from the same silver bromide emulsion. Fig. 5: normal metolquinone developer; fig. 6: p-phenylene- diamine developer. The fine structure of the devel- oped silver is considerably coarser in the case of the /7-phenylenediamine development.'' Experiments on initial development show the for- mation of a single silver filament from a silver bro- mide crystal and also three etching pits alongside each other at angles of 120 to the centre of the crystal (2, 6) (fig. 7). This phenomenon has already been investigated by Evans and Mitchell (2) with a light microscope. The silver filament forming on the developing nucleus probably arises from silver ions which move from the etching pits over the surface or possibly right through the crystal to the growing filament. When developing pure silver iodide crystals (Fig. ' Mr. Mitchell confirmed this view during a discussion held with him. - From a combined report with Dr. Saleck, Agfa AG, Leverkusen. ^ The effect upon the "graininess" of a photographic emulsion (medium variation in density) will have to be discussed on another occasion. Figs. 5-6. Differently developed silver bromide grains. Fig. 5, metolquinone developer (carbon replica). Fig. 6. p- phenylenediamine developer (carbon replica). Figs. 7-8. Initial development results. Fig. 7, fiat silver bro- mide crystals. Fig. 8, silver iodide crystals. Gelatin in the Photographic Process 351 8), very long and thin silver filaments occur and at the same time pronounced surface structures appear on the crystal. The results show that it is possible by the carbon replica method to render tine structures of undevel- oped and developed silver halide crystals visible which were previously inaccessible to investigation. I should like to thank Prof. Friescr for the detailed discussions he held with mc upon the suhjcct in question. The electron micrographs were made by Dr. Kirchcr in the inorganic analytical laboratory of the Farbcnfabriken Bayer, Leverkusen, and I should also like to express m\ thanks to him. In addition, i am indebted to Miss ina Lehnert and Mr. Hans-Jorg Metz for valuable assistance during the experiments. Prof, von Borries always devoted his special attention to these problems and generously assisted us in dealing with them. References 1. Bradley, D. E., Brit. J. Appl. Pliys. 5, No. 2, 62 (1954). 2. Evans, T. and Mik mfll, J. W., Report ol" Bristol C"on- lercnce 1954 (409). ndects in Crystalline Solids. 3. pRrESiR, H. and Kiiin, E., Mitt. Agfa 1, I. Springer Verlag, Berlin, 1955. 4. HiixiEs, J. M. and Mit( iiii i , J. \V., Pliil. Mai;. (7), 44, 223, 357 (1953). 5. Klein, E., Mitt. Agfa 1, 10. Springer Verlag, Berlin, 1955. — Phot. Korr. 91, \19, 199 (1955). 6. Biinsentagung 1956 Freiburg. Z. Elektrocheni. 60, 998 (1956). 7. KoNiG, H. and Helwiu, G.,Z. Physik 129, 491 (1951). 8. KoMCi, H. and Knoch, M., Phys. Vi'iluimli. 2. 14 (1955). Gelatin in the Photographic Process G. Vandermeerssche, C. Maertens and G. Lion Centre de Microscopie Electroniqiie, Medical, Industriel et Agricole, Briissels-Uccle As early as 1935 Russian scientists have tried to find a practical classification for photographic gela- tins (5). This classification was based only on the value of gamma and on the sensitivity of emulsions prepared with the diflferent gelatins but American workers proved this classification to be impractical (12). It is the aim of the present paper to show the marked differences which exist when emulsions are made under completely identical conditions except for the fact that different gelatins are used. The raw material which had been used to make the six different types of gelatin can be classihed in two main groups, ossein and skin. The duration of liming for the skin gelatin has been similar for two of the specimens but there was a difference in treatment, one skin-gelatin being prepared with nitric acid, the other with hydrochloric acid. Fig. 1. Gelatin 1: ossein from hard bone, 1st extraction. Fig. 4. Gclaiin 4: ossein from sinews. Fig. 2. Gelatin 2: ossein from hard bone, 3rd extraction. Fig. 5. Gelatin 5: skin, nitric acid treatment. Fig. 3. Gelatin 3: ossein from soft bone. Fig. 6. Gelatin 6: skin. Indrochloric acid treatment. The emulsions were exposed to light (one minvitc) and then developed. The ossein gelatins were different with respect to their extraction-temperatures and also with respect to the fact that one sample came from hard bones, another one from soft bones and a third one from sinews. The emulsions were prepared in the classical way as indicated in the literature on photographic emulsion techniques (3, 13). As developers we have used the for- mulas given in the Gevaert handbook. A first series of emulsions has been exposed to light, developed and then washed in order to remove most of the gelatin present on the specimen grid. A second series has been treated in the same way but the gelatin has been remo\ed before exposure to light. Use has also been made of the specimen preparation technique described by Hamm and Comer (7), which consists in utilizing the thin tilm of the sii\cr- gelatin complex which exists as a tightly fitting en\elopc around the photographic grains. In that case the gold development has been applied. 352 A. FELTYNOWSKI, I. GLASS AND L. GRELEWICZ Figs. 1-6 show the difference in grains obtained when six different gelatins are used in emulsions which are otherwise completely similar. From this investigation it seems to be impossible to draw a definite conclusion about the relation between silver-grain size and silver-grain colour without taking into account the kind of gelatin used. For the ossein gelatin the grains are more compact and more dense in the blue-black pictures, while for the skin-gelatin the effect is just opposite. Further work is in progress in order to interpret these different findings while taking into account the physical and chemical properties of the gelatins used. We wish to express our sincere appreciation to Messrs. L. de Barcy and P. Frottier of the Entreprises Chimiques et Electriques at Vilvorde for permission to publish this paper. References 1. Ammann-Brass. H., Colloque de Science Photogr., Paris 1951. 2. Arens, H., Z. wiss. Pilot. 43, 120 (1948). 3. Baker, Th., Photographic Emulsion Technique. Mc Millan. 4. Bradley, D. E., Brit. J. Appl. Phys. 5, 96 (1954). 5. Chibisov, K. V. and Mikhailova, A. A., Kino Photo Chem. Incl. 3, 24 (1937). (In Russian.) 6. Hall, C. E. and Schoen, A. L., J. Opt. Soc. Am. 31, 281 (1941). 7. Hamm. F. a. and Co.mer, J. J., /. Appl. Phvs. 24, 1495 (1953). 8. JOHANN, I. and Klein, E., Sc. et Ind. Phot. 27, 91 (1956). 9. Karpova, a. L. and Moshkovskii, Sc. et liul. Phot. 11, 346 (1956). 10. Levenson, G. I. p. and Tabor, J. H., Sc. et Ind. Phot. 23, 295 (1952). 11. Pouradier, J., Chim. et Ind. 74, 1175 (1955). 12. Trivelli, a. p. H. and Smith, W. F., Z. wiss. Phot. 2,1, 123 (1938); 37, 140 (1938). 13. Wall, E. Photographic Emulsions. The Microstructure of Photoconductive PbTe Layers A. Feltynowski, I. Glass, an