^ m ^ 9 $? 5- 'T>L- -ä- W- ^ m- THE JOURNAL OF THE COLLEGE OF SCIENCE IMPERIAL UNIVERSITY, JAPAN. VOL. III. ^ m * # tfj if BJ Vf? * H ^ PUBLISHED BY THE UNIVERSITY. TOKYO, JAPAN. 1890. ^7^3 ['.'-.-. CONTENTS. Vol. III. Jurassic Plants from Kaga, Hida, and Echizen. By Matajieö Yokoyama, BigahusH. (Plates I. -XIV.) x On Pyroxenic Components in certain Volcanic Rocks from Bonin Island. By Yastjshi Kikuchi, Eigalcushi, Assistant Professor of Geology, Imperial University. (Plate XIV bis.) 67 The Eruption of Bandai-san. By S. Sekiya, Professor of Seismology, and Y. Kikuchi, Rigalaishi, Assistant Professor of Geology, Imperial University. (Plates XV. -XXIV. ) 91 On Magnetic Lagging and Priming in Twisted Iron and Nickel Wires. By Cabgill G. Knott, D.Sc. ( Ediu. \ F.E.S.E., Professor of Physics, Imperial University. (Plates XXV. -XXV II.) 173 Effect of Twist on the Magnetization of Nickel and Iron. By H. N u : v oka, BigaJcushi. (Plate XXVII. 189 Oxyamidosulphonates and their Conversion into Hyponitrites. By Edwabd Divees, M.D., F.E.S., Professor, and Tahemasa Haga, F.C.S., Assistant Professor of Chemistry, Imperial University 211 On a Condensation Product of Acetone and Aldehydammonia. By MlTSUBTJ KUHARA, I'll. D Capillary Attraction in Relation to Chemical Composition, on the Basis of R. Schiffs Data. By K. [keda, Rigakushi, of the Imperial 241 University Determination of the Elements of the Sun's Spin. By S. Hikayama, Ricjakashi, of the Imperial University -',0 On the Fineness of the One Yen Silver Coin. By Yoshimasa Koga, Ri(,al.mhi, F.C.S., andOsAMU Yamagata, Rigahishi, Assayers in the Im- perial Mint at Osaka 289 On Cordierite as Contact Mineral. By Yasushi Kikuchi, Rigakmhi, As- sistant Professor of Geology. Imperial University. (Plate XXV ft I)... 812 Transient Electric Currents produced by twisting Magnetized Iron, Steel, and Nickel Wires. By H. Nagaoka, Bigakushi, of the Imperial University. (Pia i varieties occurs plentifully at Shimamura and Okamigö. It is very important, as it may be called almost cosmopolitan, being found in nearly all the countries where Jurassic flora flourished. Another species of Podoza- mites, viz., Podozamites tenuistriatiis, may be considered as a veay close ally of P. lanceolatus, being quite like the latter in form, only with finer veins. Very remarkable are the leaflets of P. Beinii which are as numerous as those of P. lanceolatus. On the one hand it exhibits nearly rounded leaflets, and on the other, very elongated ones, thus affording passages to that cosmopolitan species through its variety JURASSIC PLANTS FEOM KAGA, HIDA, AND ECHIZEN. 15 ovalis. The genus Nilssonia is represented by at least 3 species; one of which, N. oriental is, is found both in England and Siberia. It characterizes the flora of Hakogase. The two others are new, one of them AT. nipponensis being closely akin to AT. acuminata Schenk of the lîhaetic of Europe. Anomozamites, the brother genus of the preced- ing, is so far Aery doubtfully represented. Very interesting is the occurrence of a species of Diuonites closely related to D. Brongniarti of the Wealden. But the most interesting of all is Dictyozamites indiens Fstin., a genus and species hitherto restricted to the Rajmahal flora of India. It is found great abundance at Ozô. Allied ot the preced- ing, but with coarser veins and nets, is Dictyozamites grossinervis, a new species, found at Shimamura. The family of Cycadeae is known so far by a seed described by Dr. Geyler under the name of Gycadeospermum japonieum, which is the largest known from this order. 5. Coniferae. Represented by 6 genera and 10 species, of which 4 genera with 7 species are Taxaceae, and the rest, Abietaceae. Of the first family the new genus of Ginkgodium is the most abundant in individuals. It is closely allied to Ginkgo, but decidedly dis- tinguishable by its numerous, simple, parallel veins and its short petiole. So far only a single species has been discovered forming however, a characteristic feature in the flora of Shimamura. The next important genus is the Ginkgo itself. Though much less numerous in individuals, it has yielded 3 species, all belonging to forms already described. The most important of these is Ginkgo digitata which has been found in Yorkshire, Spitzbergen and Siberia. In the general form of the leaves this specie approximates very closely to our living G. biloba, however with rarer veins. G. sibirica and G. cfr. lepida, with the lamina separated into many narrow lobes, are closely allied to each other. So far, they are strictly Asiatic. Then 16 M. YOKOYAMA. comes Czekanowskia, which, though imperfect, seems to be referable to an already known form. The two species of Taxites, being present only in isolated leaflets, do not admit of strict specific determina- tion. Much less numerous are the Abietaceae, represented by 2 o-enera and 3 species. The most interesting as well as the most important of these is Plnus Nordenskjoldi which Prof. Schmalhausen brings under his new genus Ctjclopitys founded on similar leaves found in Siberia which are arranged in whorls around the stem as in our recent Sciadopitys. The species occurs in Japan only in isolated leaves, for which I reason have adopted the older generic denomination of Heer. It occurs also in Siberia, lîussia and Spitzbergen, and perhaps also in Andö in Norway and Nancy in France. Pinus cfr. prodromus, obtained so far only in fragments in Japan, has already been found in Spitzbergen and Siberia. The last is the very interesting genus Palissija. It is decidedly Rhaetic in Europe, and as a Jurassic plant has hitherto been confined to India — to the three groups of Kajmahal, Kach and Jabalpur. Its discovery in Japan, a country intermediate in climatic conditions between Siberia Avhere the genus is unknown and India where it is known, is of high interest, imperfect though the specimen itself is. Conclusion. Out of the 3G well determined species of plants which Japan has afforded, 20 species have been identified with those already known in other countries. Of these IG or 80 % are ft>uucl m tne ' ^rown Jura ' 20) of Siberia ; viz., 1. Thyrsopteris Murrayana Brgt. 2. ,, prisca Eichiv. •20) Dr. Oswald Heer : — Beiträge zur Juraflora Ostsibiriens und des Amurlandes, p. '10. (Memoirs de l'Acad. impér. des Sciences de St. Petersbourg, VIIe Série, Tome XXII, No. 12 et dernier). Flora Fossilis Arctica, vol. IV. JURASSIC PLANTS FEOM KAGA, HIDA, AND ECHIZEN. 1? 3. Dicksonia acutiloba Hr. var. 4. ,, gracilis Hr. 5. ,, cfr. Gleh'iiana Hr. 6. ,, nephrocarpa Bunb. 7. Asplenium whitbiense Brgt. 8. ,, argutulum Hr. 9. ,, distans Hr. 10. Nilssoiiia orientalis Hr. 11. Podozamites lanceolatus Lindl. 12. Ginkgo digitata Brgt. 13. ,, cfr. lepida Hr. 14. ,, sibirica Hr. 15. Pinns Nordenslijoldi Hr. 16. ,, cfr. prodromtis Hr. With the flora of the same epoch 21) in Spitzbergen we have 6 species in common that is, 30 % of the indentified species ; viz., 1. Vecopteris cxilia Phill. 2. ,, Saportana II r. 3. Podozamites lanceolatus Lindl. 4. Ginkgo digitata Brgt. 5. Pinus Nordenskjoldi Hr. 6. ,, cfr prodromus Hr. Our flora has a nearer relation to that of the Yorkshire coast,22) distant though it is, having as many as 9 species or 45 °/0 m com" mon. They are the following : — 1. Thyrsopteris Murrayana Brgt. 2. Dicksonia 'nephrocarpa Bunb. 21) Dr. Oswald Heer :— Beiträge zur fossilen Flora Spitzbergens, p. 27. (Kougl. Svenska Vetenskaps Akaderniens Hauclliügar, Baudet 14, No. 5). Flora Fossilis Arctica, vol. IV. 22) Bath Oolite. 18 M. YOKOYAMA. 3. Asplenium whitbiense Brrjb. 4. ,, argutulum Hr. 5. ,, distans Hr. 6. Peeopteris exilis Phill. 7. Nilssonia, orientalis Hr. 8. Podozamites lanceolatus hindi. [). Ginkgo digitata Brgt. With the Chinese and Mongolian Oolitic flora worked out by Newberry,23) Schenk24) and Schmalhausen,25) we have four iden- tifications, Asplenium'- whitbiense , A. argutulum, Macrotaeniopteris cfr. Bichthofeni and Podozamites lanceolatus ; while with the peculiar In- dian flora of Kach and Jabalpur26) we have only two, Asplenium whitbiense and Podozamites lanceolatus. The Jurassic flora of Russia — from the regions of Orenburg and Isjum in its southern part, and from Petschora-Land on the western flank of the northern part of the Ural Mountains — has several species in common with ours. Among the plants mentioned by Eichwald 27) and Schmalhausen28) we find many Japanese forms such as Asplenium whitbiense, A. argutulum, Thyrsopteris prisca, Podozamites lanceolatus and Ginkgo digitata, which last is an important Oolitic species. Our flora is also allied to that of Andö29) and Turkestan, 30) whence how- 23) J. S. Newberry : — Description of Fossil Fluids from the Chinese Coal-bearing Rocks in K. Pumpelly's Geological Researches in China, Mongolia, and Japan. (Smithsonian Contributions to Knowledge, Jan., 1866.) 24) A. Schenk: — Jurassische Pflanzen, in Richtlufen's China, vol. IV, part. X. 25) J. Schmalhausen : — Pflanzen aus der nordioestliclien Mongolei. (Melange biologique tirés du Bulletin de lAcad. Imper. d. Sc. d. St. Pefcersbourg, tome XI, March, 1883.) 26) O. Feistmantel considers both Kach and Jabalpur as of Lower Oolitic series. Cf. Juras- sic Flora of Kach, p. 71. and Jurassic flora of the Jabalpur Group, p. 20. 27) E. d'Eichwald : — Lethaea Russia, II. 28) J. Schmalhausen : — Beiträge zur Juraflora Russlands. (Bulletin de lAcad. Impér. d. Sciences de St. Petersbourg, Tum. XI, Jan. 1879.) 29) 0. Heer :— Ueber die Pflanzen-Versteinerungen von Andö in Norwegen. Flora Fossilis Arc- tica, vol. IV. 30) G. Romanowski: — Materialien zur Geologie von Turkestan. JURASSIC PLANTS FEOM KAGA, HIDA, AND ECHIZEN. 19 ever we know as yet a comparatively small number of species. We have still two interesting plants remaining, viz. Dictyozamites indiens Fstm. and a Sagenopteris ; the first hitherto confined to the Rajmahal or Liassic31) flora of Indin, and the second allied to Sage- nopteris rhoifolia Presl which is also partly Liassic, but principally Ehaetic in Europe. This latter has also been described from the Mesozoic formation of Queensland in Australia, which Feistmantel is inclined to believe to be Liassic.32) From what I have stated above, 19 species or 95 °/0 of those identified have been found in the ' Brown Jura ' of other parts of the world, and only 1 species or 5 °/0 in the lower deposits. There- fore, as Dr. Gey 1er had already done, 33) I do not hesitate to conclude that the Jurassic flora of Kaga, Hida and Echizcn belongs to the same geo- logical horizon as the floras of Siberia, Spitzbergen, and Yorkshire, namely, to the Bathonian Stage of the Inferior Oolite, with special relations to the flora of Siberia. This view is moreover justified by the occurrence of Czckanowsliia, Taxites and Palyssia, which have their nearest allies in forms already found in the Inferior Oolite. It may perhaps be urged, that although the plants from most of the localities are decidedly Oolitic, yet Ozö which has given Dictyoza- mites indiens and a Sagenopteris allied to Sagenopteris rhoifolia, might belong to a horizon somewhat lower than that of the Inferior Oolite. This question, as far as our [»resent investigation goes, must be an- swered in the negative, as the place has yielded, besides these two 31) Comp. Feistmantel's Jurassic Flora of the Rajmahal Group, p. 109. 32) 0. Feistmantel: — Palaeozoische und Mesozoische Flora im östlichen Australien, Palaeou- togr. 1879, Suppl. Ill, Lief. Ill, Heft 4, p. 17 A. In this work the author mentions 6 other species as occurring with Sagenopteris rhoifolia, viz., Sphenopteris elonaata Carr., Thinnfeldia odon- topteroides Fstm., Cijclopteris cuneata, Taeniopteris Daentreei Mc Coy, Otozamites cfr. Mandeslohi Kurr, unci Cardiocarpon australe Carr. These are quite foreign to our flora and possess an Indian aspect. 33) Ueber foss. Pflanz, a. d. Juraform. Japans, p. 223. 20 M. YOKOYAMA. species, forms which are also found in other localities ; viz., Omjchiop- sis elongata, Asplenium whitbiense, A. argutulum, A. distans, and Podoza- mites Pieinii, which are all very important Oolitic types. Among other interesting forms of the Japanese plants, I may mention Nilssonia nipponensis m. closely akin to N. acuminata Schenk of the Jihaetic, and Dioonites Kotoei m., and Equisctum ushimarense m., resembling respectively Dioonites Brong?iiarti Schenk, and Eqtiisetum Buchardti Schimp. of the Wealden. With the Infra- Liassic flora of Tongking described by Zeiller, 34) we have no species in common, save Podozamites distans , which Heer has shown35) to be one of the many forms of Podozamites lanceolatus. As seen from a foregoing table, 19 species or 89 °/0 of the whole plants are ferns, 15 species or 30% cycads, 10 species or 20 % coni- fers, and the remaining 11 °/0 are taken up by Rhizocarpeae, Equiseta- ceae and Dubia. Therefore the ferns are the most numerous in the Japanese flora, next to which come the cycads ; while conifers are only half as numerous as ferns. In the Indian flora of Kach and Jabalpur, whence indeed we at present know 48 species of plants, 40 % are cycads, 29 °/0 ferns and 29 % conifers, thus showing a great preponderance of cycads, over the two others. Again, a different order holds in the great flora of Siberia. There, the two able palaeophytologists Heer and Schmalhausen, have already discovered 127 species 36) of Jurassic plants. Of these, conifers take the lead, making up about 40 °/0 °f the whole flora ; while ferns and cycads range a little above 20 °/0, the former perhaps a little more abundant than the latter. Therefore our own flora may be considered as a 34) Zeiller: — Examen de la flore fossile des couches de charbon du Tongking iu Annales des Mines, 8e Serie, tome II, 5e livr., 1882. 35) 0. Heer: — Beiträge zur Juraflora Ostsib. u, d. AmurL, 1876, p. 107. 36)0. Heer: — Nachträge zur Juraflora Sibiriens p. 2 (Memoirs d. l'Acad. Imp. d. St. Peters, bourg, Vile série, Tome XXVII, no. 10). Flora Fossilis Arctica, vol. VI. Tabular View of the Jurassic Plants of Kaga, Hida and Echizen. Ocourrence of identical <-v No. Names. Kaga Eobizen, Hida allied Species in other Class 1. Cryptogahae. l i ai î p Order 1. Filicacbah. i 1 ä | i 's j Fain. 1. Polypodiaceae. I« O s 1 E-i o p 1. ThyrsopterisMitrraaana Tlrijt. ... + - Siberia, Sorkahire. 2 + Siboria, Itussia. a. + - — - + — - 4. + - " " + z Siberia. 6. .Silp.Tia. 6. eir. Qlelmùma Ur. ... + Siberia, Yorkshire (?) 7. „ nrphroeiirjio Bonla ... + — — - - — — Siberia, Yorkshire. 8. + + + + 0. Adiantites Keeriamts nov. sp. ... + 10. ,, Kochibeanus nov. up — + 11. „ lancçus nor sp + 12. Asptënium whitbiense Dreß + + Siberia, China, Mongolia, India, (Jabalpur and Kach), Yorkshire, Tur- kestan, Kajmahal. 13. + + Siberia, Mong.iiin., Russia, rorkBhire. 14. „ distant Hr + " + + ~ ~ + Siboria, Yorkshire. Farn. 2. Sphenoptuiideae. 15. Fain. 3. Pecopterideae. + Sphenopteris Williamsonit Brgt. of Oolite and S. Mantclli Dr)... + — + — — — — Siberia. 40. Cup ditlies ginkjoidet n >r pt..„ + JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 21 kind of connecting link between the northern or Siberian, and the southern or Indian, faciès of one great Oolitic flora. The occurrences of Indian elements like Dictyozamites and Palyssia like- wise bint at the same. It may be here added, that within a last few years Jurassic plants have been discovered in many other parts of our country. In the Mesozoic basins of Awa and Tosa in Shikoku, plants seem to be tolerably numerous, also animal remains, which are referable to the three main periods of the Mesozoic Era. In the province Kii, to the south of the city of Osaka, plants again occur ; and this is the case also in the little basin of Kagahara in the province Közuke, where however, as far as I know, they are very rare. Jurassic plants have also been brought from the northern part of Shinano, where they were observed in pebbles. These, I hope, will form a subject of another paper on the Jurassic plants of Japan. 22 M. YOKOYAMA. 2. Description of the Species. Class 1. Cryptogamae. Order 1. Filicaceae. Fam. 1. Polypodiaceae. 1. Thyrsopteris Kze. 1. Thyrsopteris Murrayana Brgt. sp. PL XII, fig. 5. Thyrsopteris Murrayana — Heer, Beitr. zur Jura -flora Ostsib. u. d Amurl., 1876, p. 30, pi. I, fig. 4, II, 1-4, VIII, 116. Beitr. 1878, p. 1, pi. I, fig. 6. Nachträge, p. 6, pi. I, fig. 1. Pecopteris Murrayana — Brongniart, Veget. Foss. I, p. 358, pi. CXXVI, fig. 1, 4. Sphenopteris Miirrayana — Zigno, Enum. Fi lie. Eoss. Oolith., p. 20. Hymenophyllites Murrayana — Zigno, Flora Foss. Form. Oolith., p. 92. Tympanophora racemosa — Lindley and Hutton, Fossil Flora of Great Britain, vol. Ill, pi. 170. Coniopteris Murrayana — Schimper, Palaeont. Veget., vol. Ill, p. 471. This species ocurs only in small fragments, one of which I have here figured. It is distinguished from its closely related species, Thyrsopteris prisca Eichw. in having the tertiary veins simple. The discovery of undoubted, though fragmentary, specimens of this plant in Japan is very important, as it not only occurs in Siberia, but has already been described by Brongniart and Lindley from the Oolitic flora of Yorkshire. Loc. — Okamigö. JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 23 2. Thyrsopteris prisea Eichiv. sp. PL I, fig. 3, 3a, 4. Thyrsopteris prisea — Heer, Beitr. zur Juraflora Ostsib. u. d. Amurl., p. 86, pi. XVIII, fig. 8. Schmalhausen, Nachtr. zur Jura- flora des Kohlenbassins von Kusnezk am Altai, p. 548, pi. I, fig. 2-4. Sphenopteris pmca-Eichwald, Lethaea Rössica, II, p. 14, pl. IV, fig. 2. Although our specimens are by no means complete, yet the elongated pinnae, the ovately-triangular and pinnatifid pinnules, and the obtuse lobes, together with the dichotomous tertiary veins, suffice l. ) show that the species, first found at Kamen ka and afterwards in ►Siberia, is also represented in the Japanese Jurassic system. From the preceding species, to which this is very closely akin, it is distin- guished in having the tertiary veins dichotomous, as shown in fio-, 3a. Loc. — Shimamura. 3. Thyrsopteris Kagensis m. PI. I, fig. 6, 6a. PI. XI, fig, 7. Frond bi-tripinnated ; pinnae elongated; pinnules coriaceous, alternate, acutely direeted forward, ovate-lanceolate, gradually tapering below and contracted at base, lobed or even pinnatipartite ; lobes or partitions narrow, acutehj directed forward, acute at apex ; veinlets dichotomous. This fern seems to have been twice to thrice pinnated. Some of the upper part of the upper pinnae show quite an entire margin, while others are furnished with lobes. A pinna in the lower part of the frond (PI. I, fig. 6, left) however exhibits lobes nearly separate from one another, so that each of them here takes the place of the pin- nules above. The substance of the pinnules seems to have been tolerably thick in consistence. A vein given off into each lobe or partition again dichotomizes as represented in fig. 6a. These veinlets 24 M. YOKOYAMA. are acutely directed forward like the lobes themselves. A fragment from Tanimura (PL XI, fig. 7) with acutely directed, in some cases irregularly lobed pinnules very probably belongs to the some species. The generic position of this fern is at present uncertain; but the general aspect of the pinnules, and their mode of lobing remind us of many of the Thyrsopteri which are widely spread in the Jurassic system. Rare. Loc. — Shimamura, Tanimura. 2. Dicksonia L'Herit. 4. Dicksonia gracilis Heer. PI. I, fig. 5, 5a. PL XII, fig. 13. Dicksonia gracilis — Heer, Beitr. zur Juraflora Ostsib. u.d. AmurL, 1876, p. 92, pi. XVII, fig. 3., Beitr. 1878, p. 13, pi. Ill, fig. 8-14. The specimens from both localities represent the upper part of a frond. Although they are rather imperfect, and the impressions more or less indistinct, yet the thickly-set narrow pinnae, with small, broadly lanceolate, oblique, entire-margined, rather acutely pointed pinnules, tend more to point to D. gracilis than to its close ally D. acutiloba Hr. Heer speaks of his gracilis as having obsolete and aca- tiloba distinct venation. Just so in our specimens ; the venation is in most cases very indistinct. By strong enlargement, however, we can observe faint undivided secondary veins rising acutely from the fine median vein, as shown in fig. 5a. This species seems to have been very rare. Loc. — Shimamura, Okamigö. 5. Dicksonia acutiloba Heer var. PL I, fig. 2a, 2, lb. Dicksonia acutiloba-Heev, Beitr. zur Juraflora Ostsib, u. d. AmurL, 1S76, p. 92, pi. XVIII, fig. 4. JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 25 Closely akin to the preceding, but distinguished in having more sharply pointed pinnules and dichotomous secondary veins. A speci- men here figured (fig. 2) shows slender pinnae with acutely directed, lanceolate pinnules, which are therefore narrower than the majority of those figured by Heer. They correspond more to those represented in the left-hand extremity of Heers figure. Fig. lb shows a fragment of a pinna with long acute pinnules whose inferior veinlets are dichotomous (fig. 2a) ; so it is probably one of this species. Yet on account of the narrower nature of the pinnules as compared to those described by Heer, I name the Japanese form a variety. Very rare. Loc. — Shimamura. 6. Dickson ia cf. Glehniaiia Heer. PL XIV, fig. 11, 11a, 12a. Dicksonia Gleliniana — Heer, Beitr. zur Juraflora Ostsib. u. d. Amur!., 1876, p. 91, pi. XVII, fig. 4, XVIII, 6, 7. Heer's diagnosis runs as follows : " Frond bipinnated, coriaceous ; pinnae alternate, acutely directed forward, narrow ; pinnules oval, strongly oblique, narrowed at base, decurrent, entire, apex obtuse ; veins very fine." In the alternate, obliquely-oval pinnules, obtuse at apex and with simple lateral veins, our specimens agree with the figures and descrip- tions of Heer, but they are too meagre and fragmentary for an exact determination. The substance of the frond, just as Heer says, seems to have been quite thick. Fig1. 11a is an enlarged view of fis:. 11. to show the veins more distinctly. According to the investigations of Prof. Nathorst, this species is doubtfully represented in the Oolite of Yorkshire. Loc. — Hakogase. 7 Dicksonia nephroearpa Bunb. sp. PI. I, fig. 1, la. S pli en opter is neplrrocarpa — Bunbury, Quart. Journ. of Geol. Soc. , 26 M. YOKOYAMA. 185Ï, vol. VII. p. 179, pi. XII, fig. la, 11». Dicksonia clavipes — Heer, Beitr. zur Juraflora Ostsib. u. d. Amurl., 1876, p. 33, pi. II, tig-. 7, 71). On a small piece of sandstone from Shirnamura, there occur fertile pinnae of a fern, which possess comparatively large, kidney- shaped sori 1.5 — 2 mm in breadth, each of which is borne at the apex of a shoit lobe1 more or less narrowed towards the base. Under a magnifier, a vein is seen u'oinu' into each sorus. sometimes giving off delicate lateral veinlets (fig. la). These pinnae no doubt belong to a Dicksonia which is at least very closely allied to D. clavipes Hr. from Kaja in Siberia (Beitr. 1876, p. 33). This latter species is considered by Prof. Nat h ors t as identical with Sphenopteris ncphrocarpa Bunb. from the Oolite of York- shire. To this opinion Heer assents in his Nachträge, p. 6. As I see no sufficient character to separate the Japanese from the Siberian fossil, I identify them, and following Heer (Nachtr., p. 6) in the denomination of the species, I call it Dicksonia ncphrocarpa Bunb. S[). To this group may be referred the fertile pinnules figured by Old- ham and Morris as Sphenopteris Bunburyana (The Flora of the BajmaJial scries, pt. XXX J I tig. 5, 6', /), which Feistmantel includes, together with our present species, under Hymenophyllites Gup. (Jurassic Flora of the Bajmahal Group, p. 26-27). Loc. Shirnamura. 3 Onychiopsis m. Fertile segments different from the sterile. Sori terminal, linear, on each side of the midrib, parallel with the margin, involucrate ; the involucrum of each side confluent over the midrib. This new genus, which I have founded on a plant first described by Dr. Geyler under the name of Thi/Vsopieris elougata and afterwards JURASSIC PLANTS PROM KAGA, TirDA. AND ECHIZEN. -< mentioned by myself as Dicksonia elongata in the Bulletin, shows a fructification apparently resembling that of Cryptogramme B. Br. and Onychium Kauf I. of the recent flora. The sori were probably linear, placed on each side of the midrib as in Onychium. The involucrum may possibly have been formed of the rolled-lip margin of the seg- ment or pinnule as in Cryptogramme. The general appearance of the fertile segments and the terminal nature of the sori remind us strongly of the latter genus. Onychiopsis may provisionally be treated under Polypodiaceae, in the neighbourhood of the tribe Pteridae, until the discovery of spo- rangia points out its true systematic position among the ferns. 8. Onychiopsis elongata Geyl. sp. PI. II. tig. 1—3. PI. Ill, fig. 6 d. PI. XII, fig. 9,10. Frond slender, bi-tripinnated ; sterile pinnae alternate or rarely op- posite, elongated, their length rapidly increasing towards the borer part of the frond : pinnules alternate, acutely directed forward, lanceolate or linearly -lanceolate, entire or lobed or even piiiuafelg parted : lobes or par- titions acute at apex and acutely directed forward just l ihr the pinnules themselves. Venation obsolete, secondary reins simple, each, going into a lobe. Fertile pinnules elongated, irith a linear terminal sums on both sides of the midrib. Thyrsoptens elongata — Geyler, Ueberfoss. Pflanzen a. d. Juraform. Japans, p. 224, pi. XXX. fig. 5, XXVT, 4.5. Schenk in Richtho- fens China, vol. IV, part X, p. 263, pi. LIV, fig. 1. Dicksonia elongata-Y okoyama, On the Jurassic Plants of Kaga, Hida and Echizen (Bull. Geol. Soc. Japan, part B, vol. I, No. 1) p. (5. This plant seems to have been very slender and graceful in general appearance, with twice to thrice pinnated fronds whose lightly bent pinnae are in most cases set alternately along a slender rhachis (fig. 2, -3 M. YOKOYAMA. PI. II). The pinnules are much elongated, being longest near the base of the pinna. They are quite entire in the upper pinnae, but become lobed as they get downward, and the incision between the lobes becomes deeper and deeper, so that at last the individual lobes look quite like the pinnules of the upper part of the frond. Fig. 3, PI. II, represents a pinna belonging to the lower part of the frond, whose extremely elongated pinnules give such an appearance as above alluded to; fig. 3a is an enlarged view of a part of one of them. The length of a pinnule here attains about 37 mm, the breadth being only 3 mm. Fig. 10, PL II also represents an inferior pinna. Along with these sterile pinnae are found fertile pinnules which are either borne quite on a separate pinna or, as is sometimes the case, mixed with the sterile pinnules (comp. fig. 1, 4 a, b, PI. II, fig. 6d, PI. Ill, fig. 9, PL XII). The sori -are placed two at the end of each pinnule which is considerably narrowed looking like a winged stalk. These double sori are opposite, elongated, linear, meeting each other along the midrib. This splendid fern is the chief and characteristic fossil of the Japanese flora, being found in all of the fossil localities. 4. Adiantites Gap. 9. Adiantites Heerianus m. PL XII, fig. 1, la. lb, 2 Pinnae elongated; pinnules alternate, acutely directed forward, rhomboidal, attenuated'^ below, acute at apex, acutely lobed ; veins equal, fine, repeatedly dichotomous The four pinnae shown in fig. 1 and 2, 1 believe, to belong together. The pinnules are, as a general rule, obliquely rhomboidal in shape, attenuated towards the base, and acutely pointed at the apex. They are mostly acutely lobed, but becoming entire and more lanceolate in JURASSIC PLANTS FEOM KAGA, HIDA, AND ECHIZEN. 29 shape towards the upper part of the pinnae. As to the venntion, no distinct median vein is observable, but the veins are fine, nearly equal in size, and several times forking (fig. la) Pinnules of a pinna lying in the middle of fig. 1 show dark spots around their margin, which are probably due to the presence of fruit- dots. Each of these dots or sori seems to have been borne on the tip of each lobe at the ends of veins (fig. lb), just as in the recent Adiantums of the group of A. capillus- veneris L. Therefore our plant, together with Adiantites Schmidtianus Hr. (Beitr. zur Jura flora Ostsib. u. d. Amurl., 1876, p. 36, pi. II, fig. 12, 13), in which Heer observed a similar kind of fructification, very probably belongs to the genus Adiantum. However as Dichsonia in a fossil state sometimes shows quite a similar looking fructification, it many be most prudent at present to refer our fossil to the provisional genus of Adiantites Gup. This species may be compared to A. nympharum Hr. (Beitr. zur Juraflora Ostsib. u. d. Amur!., 1876, p. 93. pi. XVII, fig. 5) which, however, has obtusely lobed pinnules. Rather rare. Loc. — Shimamura. I may here notice that the pinnae figured as Adiantites amurensis Hr. by Dr. Gey 1er (Foss. Pflanz, a. d. Juraform. Japans, p. 225, pi. XXXI, fig. 2. 3) seem to possess pinnules, most of which are acute at the apex and are quite different from those figured by Heer (see Beitr. 1876, pi. XXI, fig. 6 ad). I doubt whether they do not belong to Héerianus tn. although it cannot be positively decided, so imperfect are the Geyler s specimens. 10. Adiantites Kochibeanus m. PL I, fig. 7, 7a. Fr)nd pinnated; pilinae elongated; pinnules alternate, acutely directed forward, entire, broadly lanceolate, cuneate at base, acute at apex ; 30 M. YOKOYAMA. veins mann, equal, divergent, repeatedlg ddcliotomons. This fern seems to have been pretty slender in its general ap- pearance. Along a rhachis, which is by no means strong, are arrang- ed alternate or opposite, elongated pinnae more or less directed forward, but often bent apparently by the weight of ovately or obovately lanceolate, thick, entire pinnules, which are tolerably acutely di- rected forward. Veins are pretty distinct, and radiate towards the apex and margin of the pinnules, two to five times forking on their way (fig. 7a). This species is closely related to the preceding in the form of the pinnules which, however, are not in this case lobed. Very scarce. Lo.c. — Sh.ima.mnra. 11. Adiantites laneeus m. PI. XIV, fig. 3, 8a. Pinnules alternate, acutely directed forward, lanceolate, acute at apex, constricted at base, entire: reins numerous, equal, divergent, repcatedl;/ dichotomaus. Along a slender rhachis, are arranged lanceolate pinnules winch are tolerably acutely directed forward. At the base they are sensibly narrowed and apparently provided with a short petiole. They appear to have been thin in texture, with numerous distinct veins diverging towards the apex and margin, and repeatedly dichotomizing on their way (fig. 3a), reminding as of the venation of many of the recent Adiantums. None of the pinnules in our specimen are perfect ; but in two of them, in which the apex and base are preserved, the length measures 50—42 mm. and the breadth 6t — 7 mm. I know none among the Mesozoic plants which can aptly be com- pared to this species. The preceding species, to which it bears some r semblance in form of the pinnules and the mode of venation, has JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 31 much smaller pinnules and less numerous veins, and is in general more strongly built. Ver}J rare ; I possess only a single specimen, the one here figured. Loc. — Hakogase. 5. Asplenium L. 2. Asplenium whitbiense Brogt. sp. PL III, tig. 3. PI. X, fig. 1, 2 a. Asplenium whitbiensis — Heer, Beitr. zur Juraflora üstsib. u. d. Amur!., 1876, p. 38, pi. I. fig. 1 c, III, 1-6, p. 94, XIV, 8, XX, 1, 6, XXI, 3, 4, XXII, 4g. 9 c, Beitr. 1878, p. 3, 15, pi. II, fig. 14-17. Naehtr. p. 7. Schmalhausen, Beitr. zur Juraflora RussL, p. 17, pi. II, fi"-. 1-10, XIV, tiu'. 4, 5. Schenk in Richthofen's China, vol. IV, part X, p. 24(5, pi. XLVI, fig. 5, 6, 7, XLVIII, 3-5, XLVIII, 1-4, XLXIX, 4a, 6b, p. 247, pi. XLVIII, fig. 5a, p. 253, pi. LII, fig. 1-3. Pecopteris whitbie?isis — Brongniart, Veget. Loss. p. 231, pi. 109, fig. 2-4. Lindley and LIutton, Fossil Flora. Vol. II, p. 145, pi. 134. Xewberry in Pumpelly's Geological Researches in China, Mongolia and Japan, p. 122. pi. IX, fig. 6. Aletliopteris whitbyensis — Feistmantel, Fossil Flora of Jabalpur Group, p. 87, pi. II, pg. 2-7. Fossil Flora of Kach, p. 22, pi. Ill, fig. 1-4. Aletliopteris indica — Feistmantel, Fossil Flora of Rajmahal Hills, p. 37, pi. XXXVI, fig. 4a, pi. XLVI, fig. 3-4. Although our specimens are by no means well preserved, yet the short and more or less falcate pinnules show us that we are dealing with the wide-spread A. ivhitbiense Brgt. sp., and indeed with the variety tenue of Heer. a 9 M. YOKOYAMA. The specimens from Ozö have the pinnae a little more acutely directed forward than in the typical forms, with some of the falcate and blunt pinnulae more closely set together, thus reminding us also of the sterile pinnulae of Pecopteris Williamsonis Brgt. (e. g. Hist. Yégét. Foss., pi. CX, fig. 12) the original specimen of which I had an opportunity of examining in the museum of Stockholm. Loc. — Shimamura, Ozö. 13. Asplenium argutulum Heer. PI. Ill, fig. 1. PI. XII, fig. 8. PL XIII,fig. 9. PI. XIV. fig. 2. Asplenium argutulum — Geyler, Ueber foss. Pflanz, a. d. Juraform. Japans, p. 225, pi. XXXI, fig. 1. Hear, Beitr. zur Juraflora Ostsib. u. d. Amurl., 1876, p. 41, pi. Ill, fig. 7, p. 9(3, pi. XIX, fig. 1-4. Schenk in Kichthofen's China, vol. IV, part X, p. 246, pi. XLVI, fig. 2-4. Schmalhausen, Beiträge zur Juraflora Russlands, p. 23, pi. II, fig. 12. This species is a close ally of the preceding, hut with pinnules narrower, straighter, and more acute at apex. A specimen from Shimamura (PI. Ill, fig. 1) shows sonn1- of the pinnules crenulate, on which account it may be prudent for the present to consider it as a va- riety. That from Hakogase figured in Plate XIV has also a great resemblance to some forms of Pecopter.s Phillipsii Brgt.=P. cxilis Phitt. (comp. Brongniart, Hist. Végét. Foss., pi. CIX, fig. 2). This species like the preceding has a wide geographical distribu- tion, as it is also mentioned from Yorkshire by Prof. Nathorst. Not frequent. Loc. — Shimamura, Ozd, Hakogase, Okamigô. 14. Asplenium distans Heer. PI. Ill, fig. 2. PI. XI, fi. 4. PI. XIII, fig. 4. PI. XIV, fig. 1. Frond pinnated ; pinnae elongated ; pinnules either free or unit- ed at base, lanceolate, entire, more or less falciform, acute at apex ; JURASSIC PLANTS FROM KAGA, IT ID A, AND ECHIZEN. ÔÔ 10-20 mm. long, 4-6 mm. broad; secondary vein« fine, acutely direct- ed forward, clichotomous. (Heer). Asplenium distans — Heer, Beitr. z. Juraflora Ostaib. u. d. Amurl, 1876, p. 97, pi. XIX, fig. 5-6. Pecoptcris recentior — Phillips, Geology of Yorkshire, p. 119, pi. VIII, fig. 15? Zigno, Flora Foss. Form. Oolith. I, p. 195. Neuropteris recentior — -Lindley and Hutton, Fossil Flora, vol. I, p. 195, pi. 68. Gröppert, Syst. Filic. p. 205. linger, Gen. et Spec. Plant. Foss. p. 85. Sternberg, Flora der Vorwelt, II, p. 76. Cladophlebis recentior — Brongniart, Tabl. Gen. Yégét. Foss. p. 105. Alethopteris recentior — -Schiinper, Pal. Végét. vol. I, p. 566. I* ter in recentior — Ettingshausen, Farn. d. Jetztwelt, p. 113. This species is distinguished from the {.receding by having long- er and narrower pinnules, and more delicate secondary veins direct- ed more acutely forward. The pinnules are entire and attached to the rhachis by their whole base. This is seen in every one of the specimens here figured. In a specimen from Shimamura (PI. Ill, fig. 2) which may be con- sidered the best we possess, the pinnules are quite free from one another, the longest measuring 19 mm. in length and 5£ mm. in breadth. Veins are here very distinct. They arc delicate, all dicho- tomous and acutely directed forward. A specimen from Ozö (PI. XIII, fig. 4) is on!)7 a fragment from the upper part of a pinna; but it is readily recognised by the acute course of the veins. That from Ha- le ogase is much better as far as the number of pinnules preserved is con- cerned ; but all of them are more or less wanting1 along their margins, so that they look more slender than they really are. Their slightly con- tiguous nature at the base, and their falcate shape are, however, perhaps best shown in this specimen (PI. XIV, fig. 1). The Ushimaru speci- men (PI. XI, fig. 4) is also a fragment like that from Ozö. I possess 34 M. YOKOYAMA. another from the same locality with more pinnules preserved ; but they are so mutilated, and in part so indistinct, that I thought it not worth while to figure it along with decidedly better, though more fragmentary, specimens. This species described from Gristhorpe and the Amoor regions, though obtained from the four localities, seems to have been frequent in none of them. Loc. — Shimamura, Ozo, Hakogase, Ushimaru. Fam. 2. Sphenopterideae. 6. Sphenopteris Brgt. 15. Sphenopteris sp. PL XIV, fig. 13, 13 a. I obtained only a fragment of this fern, which I believe is to be placed in the group of Sphenopteris Darallioides Scliimper. My specimen belongs to the upper extremity of a frond, or perhaps of a primary pinna, and shows five long linear pinnules, two on each side and one at the apex. These pinnules are all close together, obtuse at the apex and acutely directed forward. They mea- sure about 7 mm. long and only 1 mm. broad, each pierced with a single delicate vein in the middle (fig. 13 a, enlarged). I know none among the Oolitic plants that can aptly be com- pared to this species, except Sphenopteris Williamsonis Brgt. from Yorkshire (Lindley, Fossil Flora, p. 131, pi. 131). the greater part of the group of Davallioides being hitherto known from the Palaeozoic flora. Our plant is also not unlike Sphenopteris Mantelli Brgt. of the WYalden (Schenk, Flora d. Norddeutschen Wealdenformp. 208, pi. XXV, fig. 6) in the formation of the pinnules (comp. 6 a, pt. XXV of Schenk). However our specimen is too imperfect for a strict specific determination. Loc. — Hahogase. JURASSIC PLANTS PROM KAGA, HTDA, AND ECHIZEN. 35 Fam. 3. Pecopterideae. 7. Pecopteris Brgt. 16. Pecopteris exilis Phitt. PI. I, fig. 8-10, 9a. Frond tripinnated ; pinnae elongated ; secondary pinnae also elongated, linear, opposite or alternate, a little directed forward; pin- nules entire, oblong, obtuse, close together, free, but contiguous at base in the upper pinnae ; median vein distinct, evanescent. Pecopteris exilis — Phillips, Geology of Yorkshire, p. 119, pi. VIII, fig. 16. Bunbury, Quart. Journ. Geol. Soc. 1851, VII, p. 188, pi. XIII, fig. 5 a. 56. Heer Beitr. zur Foss. Flora Spitzb. p. 29, pi. VI, fig. 1, 16. Zigno, Flora Foss. Form. Oolith I, p. 141. Schimper, Pal. ATégét. vol. I, p. 536. Pecopteris obtusifolia — Lindley and Hutton, Fossil Flora, vol. II T, pi. 158. Pecopteris exiliformis — Geyler, Ueber foss. Pflanz a. d. Juraform, Japans, p. 226, pi. XXX, fig. la. This species, first found in Yorkshire and afterwards in Spitzber- gen, is also represented in the Japanese flora. Fig. 9 shows a speci- men in which two primary pinnae are preserved. These pinnae pos- sess long linear secondary ones which are often curved and provided with alternate oblong obtuse pinnules, exactly agreeing with the figure of Lindley. Pinnae of the second order measure 15-20 mm. in length and 1-5 mm. in breadth; but they become much shorter above where they appear crenate, owing to the contiguous state of pinnules at their bases. The rhachises (fig. 9.) which are by no means strong are slightly geniculate, especially in a pinna represent- ed (3u the right-hand side of the figure : but I believe this to be no permanent character of the plant. 36 M. YOKOYAMA. Fig. 8 and 10, both exhibiting crènate pinnae, probably be- long to the same species. The latter figure represents a rather in- distinct specimen. It shows, however, a primary rhachis which, when compared with that of the second order, may be called pretty strono-. It is by no means weaker than that of an English specimen fio-üred li y Lindley, as the latter might represent apart of a frond lvinsr lower in position than our own. Venation is indistinct, save a median vein which is in most cases very clearly seen. It is comparatively delicate, disappearing near the apex of the pinnules (fig. 9a). What Dr. Gey 1er described as Pecopteris exiliformis from the Tetorigawa-valley is, I dare say, no other than the present species, lie considers his plant to be more slender in general appearance than the Scarborough species; but his figures as well as my own only show the upper parts of primary pinnae, while conversely the figure of Lindley represents only their lower parts. Be that as it may, after a careful comparison between my specimens and Lindley 's figures, I could find no character sufficient to separate the Japanese from the English species. Btinbury observed in a pinnule of this fern capsules arranged in a single row on each side of a midrib [Quart. Journ. Geol. Soc, 1851, p. 188, pi. XJ J L fig. -mi, b), a fructification which he compares with that of Aneimia and Mohria among Schizaeaceae. Rarely found at Shimamura. 17. Pecopteris Saportana Heer. Pecopteris Saportana—Geylev, lieber foss. Pflanz, a. d. Juraform. Japans, p. 22G. pi. XXX. fig. 1. Heer, Beitr. zur foss. Flora Spitzb., p. 29, pi. VI, Jig. 4-7a, YJL 46. JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 37 This species was described by Gey 1er, but not a single new specimen was obtained referable to it. Fam. 4. Taeniopterideae. 8. Taeniopteris Brgt. 18. Taeniopteris (?) PL X, fig. 2c. A very small fragment of a fern (?) from the above locality situated by the side of Nilssonia ozoana m. shows very fine curved veins which are often found to fork, when examined with a magnifier, just as in Taeniopteris densinervis Feistmantel (Jurassic Flora of Each, pi. J I, fig. 6a). The specimen, however, is too meagre even for a certain generic determination. Loc. — Ozo. 9. Macrotaeniopteris Schimp. 19. Macrotaeniopteris ef. Riehthofeni Schenk. PL III, fig. 4, 5. Frond elongated, entire, acuminate at apex; secondary veins numerous, dense, rising at acute angles at base and then becoming horizontal towards the margin, simple or dichotomous. The two figures above cited represent two faces of the same leaf. It is a fragment, belonging to the apical part which is long drawn out. The specimen measures about '11 mm. in breadth near its base, from which place it appears to taper downward. Therefore the frond is not a large one such as is possessed by many of the species des- cribed under this genus. The midrib is rather weak and gets weaker towards the apex. The secondary veins are dense, many of which furcating near the point of rising along the midrib, but many also remaining simple for their whole length. Very rarely, forking takes 38 M. YOKOYAMA. place after they have gone on for a considerable distance towards the margin. Our plant shows a great resemblance to Macrotaeniopteris Bich- thofeni Schenk (Richthof en' s China vol. TV, pi. LI, fig. 4, 6) in having acuminate fronds and numerous dense veinlets, with which I indeed compare it, notwithstanding the somewhat smaller size of frond. Loc. — Simamura. Order 2. Rhizocarpeae. Fam. 1. Salviniaceae. 1. Sagenopteris Presk 20. Sagenopteris sp. PI. X, tig. 3, 3a. On a piece of black sandy shale from Ozö there lie a few scattered fragments of longly-obovate leaflets with an evanescent median vein and numerous fine veinlets, radiating towards their margin and forming, as they go, much elongated nets or aréoles. The leaflets are more or less oblique in shape, and acute at apex. These characters at once show that we are dealing with a species of Sagenopteris that is at least very closely allied to S. rhoifolia Presl., under which name indeed I mentioned it in Bull. Geol. Soc. Japan, Part B, ATol. I, No. 1, p. 6. But now lam of opinion that it is better to consider these fragments under the simple name of Sagenop- teris sp. Loc. — Ozö. Order 3- Calamarieae. Fam. 1. Equisetaceae. 1. Equisetum L. JUBA.SSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 39 21. Equisetum ushimarense m. PL XL fig. 1—3. Rhizome ribbed ; iwfegrs roundly ovate, single or joined like beads.. The three specimens here figured I believe to belong to one and the same species. They all represent underground stems which are generally slender. In fig. 3, these stems measure H — 2 mm. in breadth with internodes 15 — 20 mm. in length, and are seen with 2 — 3 strong ribs. Tubers are of various shapes owing to distortions under pressure ; but in general they are roundish or roundly-ovate and single or rarely joined like beads. (Fig. 3, small ones on the left). In fig. 3, the diameter of these tubers is approximately 5 — 9 mm, while in fig. 2 (right) a slender rhizome only 1 mm. broad possesses a large tuber 15 mm. long and 13 mm. broad, in fact the largest we have. Fig. 1 shows comparatively a broad rhizome (5 mm.) with 3 — 4 strong ribs and a tuber 8 — 11 mm. in diameter. Sometimes these underground stems are still seen with root-hairs attached. It is much to be regretted that no overground stem was found with these remains. Equisetum Buchardti Sell imp. (Schenk, Flora der nordivestdcutseh . Wealdenform., p. 205, pi. XXII, fig. 1 — 5) from the Wealden exhibits quite a similar kind of spherical tubers which are said to have been tridentate at the apex. An Oolitic species with tubers has been des- cribed by Heer (Beigtr. z. Jurafiora Ostsib. u. d. Amurl., 1876, p. 99, pi. XXI I, Jig. 5 — 7) from Bureja, which however had them more elongated than in our own. Numerous in a greenish-grey arenaceous shale at Ushimaru. 22. Equisetum sp. PL XII, fig. 7. Only a small fragment of an overground stem was obtained, 40 M. YOKOYAMA. which is 11 mm. broad and provided with numerous fine crowded longitudinal striae. It is too imperfect for specific determination, but that it is one of the Equisetaceous plants is shown by the presence of a joint. Loc. — Okamigd. Class. 2. Phanérogame. Subclass 1. Gymnospermae. Order 1. Cycadeaceae. Fam. J. Zamieae. 1. Anomozamites Schimp. 23. Anomozamites sp. PI. VII, fig. Id. A rather imperfect specimen of a leaf of an Anomozamites, or perhaps, of a Nilssonia, with subopposite quadrangular segments rounded at angles and possessed of fine simple parallel veins, looks not unlike some of the figures of A. inconstans Gap. from the Infra- Liassic series of Tongking (Zeiller. Examen de la flore fossile des cou- ches de charbon du Tongking. pi. XT. fig. 7), and also those of Nihsonia comtula Hree (Beitr. .:. Juraflora Ostsib, u. d. Amurl, 1878, pi, IV, fig. 12, IS) from the Lena regions in Siberia. However, a positive deter- mination must be postponed until better specimens are obtained. Loc. — Simamura. 2. Nilssonia Brgt. 24. Nilssonia orientalis Heer. PI. XIV. fig. 4-9. Leaf entire or rarely segmented, truncated at apex, rounded at base; veins fine, dense, slightly curved and directed forwards. JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 41 Nilssonia orientalis-Heer, Beitr. z. Juraflora Ostsib, n. d. Amurl., 1878, p. 18, pi. IV, fig. 5-9. Specimens of a Nilssonia represented in the above cited figures are, I believe, to be identified with what was described by Heer as N. orientalis from Ajakit on the Lena in Siberia. Our specimens are all entire-margined, which though variable in form, always possess a truncated apex. Fig. 5, though broken, shows an entire leaf which is about 23 mm. broad and only 44 mm. long ; so it is tolerably short. Fig. 4 represents a large leaf, 25 mm. broad with 46 mm. of the length preserved ; therefore it is probably a much longer leaf than the above. In all of our specimens the veins, which are curved and directed forward, are densely crowded, occurring as many as four to the millimetre, and thus exactly agreeing in venation with the Siberian specimens. Every one of our specimens shows a strong convexity on the upper surface of the leaf, and a sharp edge on the rhachis where the blades meet from both sides. Sometimes a leaf is found quite doubled over. Heer compares this species to A", pohjmorplia Schenk of the lihaetic flora of the Franconia and Sweden, the entire forms of which indeed the Japanese leaves in some cases resemble. \ et a decidedly denser state of veins distinguishes the latter from the Rhaetic species. According to the investigations of Prof. Nathorst, this species also occurs in the Oolite of Yorkshire. Very numerous at Halcof/ase, though mostly fragmentary, and forming there the most abundant fossil. 25. Nilssonia ozoana m. PI. X, fig. 2b, 11-14. 42 M, YO KO Y AM A. Leaf elongated, entire or rarely segmented, parallel-sided, rounded at apex ; veins very fine, dense, perpendicular to the rhachis. This species at first sight reminds us of the genus Taenioptcris among ferns, but U decidedly different from that genus in having the blade of the segments inserted on, and not laterally to, the rhachis. The leaf seems to have been in most cases entire, with sides strictly parallel except near the apex, where it is sometimes seen slightly to taper (fig. 11). Fig. 12 shows such a parallel-sided leaf. It is 58 mm. long and about 12 1/2 mm. broad without any appreciable difference in breadth in its upper and lower extremities. Fig. 14 shows a nar- rower leaf which measures 9 mm. in breadth. But that the leaves were not always entire is shown by fig. 2 b, which represents one side of a leaf divided into rectangular segments about 8 mm. broad. The veins are very fine and dense, about four to the millimetre. They are very faint, and almost obliterated in fig. 11, but are distinct in fig. 13 and 14. Leaves are in most cases strongly convex on their upper surface along the rhachis, but concave on both sides of it, as is seen in fig. 13 and 14. This species is allied to the preceding in having very dense veins and generally entire leaves, but is distinguished in having the former strictly at right angles to the rhar-his and the latter narrower and more elongated. Fragments are frequent at Ozo. 26. Nilssonia nipponensis m. PI. VI, fig. 8d. H. VII, fig. 2-7, 8a. PI. XII, fig. 6. PI. XIII, fig. 1. Leaf petioled, segmented, incisions sharp ; segments opposite or alter- na 'e. acute, more or less concave in the upper margin, con rex in the lower, JUEASStC PLANTS FROM KAGA, HIDA, AXD ECHIZEX. 13 upper one* longer, with the uppermost shortened, lower ones shorter and triangular with the upper margin straight , ceins dense, simple, parallel, equal, rising at rigid angles to the rhachis. The leaves were petioled (fig. 1, PI. XIII). Segments are in most cases separate from one another but sometimes a slightly conti- guous at base. They may be opposite as in fig. G, PI. VII, or alter- nate as in fig. 1, PI. XIII. They arc always acutely pointed at the angle where the upper and lateral margins meet. They are generally concave in the upper margin and strongly convex in the lower, and are longer than broad ; but in the upper part of the leaf, they are much broader than long (fig. I, PI. VII), with the upper and lower margins more parallel-sided. The lower segments are more or less triangular in shape with the upper margin straight (fig. 6, PL VII, and fig. G, PI. XII). The uppermost segments are comparatively short, as may be seen from fig. 3, PI. VIF, which represents the apical part of a leaf. Fig. 5, PI. VII, is also such an example, while fig. 7 and 8a show segments lying intermediate between the apex and the base. This species is very closely allied to the European Rhaetic form, N. acuminata Schenk (Flora der Grenzschichten, p. 131, pi. XXXII, fig. 1 — 7, XXXIII, 1.), from which, however, it is distinguished by having less acuminate segments, pierced with much denser veins (ca. 3 falling in a millimetre). Not frequent. Loc. — Shimamura, Ohamigö. 27. Nilssonia(?) PI. XIII, fig. 3. A broken leaf, about 31 mm. in breadth, with the length preserv- ed for GO mm, seems to have been longly-elliptical in shape to judge from its rapidly tapering state both above and below. It also 44 M. YOKOHAMA. appears to have been entire-margined with a comparatively weak midrib. The leaf has been flatly pressed on to the stone, and the venation is almost wholly defaced. By a careful examination, how- ever, it exhibits faint traces of delicate parallel veins perpendicular to the rhachis ; and as the blade seems to lie upon the rhachis, our speci- men is probably referable to Nilssonia. A positive determination however must be postponed until better specimens are discovered. 3. Dioonites Bomem. 28. Dioonites Kotoei m. VI VII, fig. la, b, c, le. VI XIV, fig. 14. Leaf pinnated ; segments opposite or alternate, lightly curved and more or less directed forward, long, linear-lanceolate, acute, inserted on the rhachis with the whole base; veins fine, equal, parallel, 7-14 in number. Segments of this species vary much in length and breadth, but are commonly 7-9 times as long as broad, and broadest.uit base. The longest segment in our specimens measures 4^ mm. in breadth and 33 mm. in length, with apex a little broken off. The segments are in most cases quite separate from one another; but sometimes a little contiguous. Their directions are not always the same, some being directed more acutely forward than others, some being tolerably falcate in shape, while others are more straight. Veins are delicate, and rise perpendicular to the rhachis (fig. le enlarged). Very closely akin to Dioonites Bronguiarti Schenk (Flora der nordwe st deutsch. Wecddenform., p. 336, pi. XXXII, fig. 2, 2a) from the Wealden of Germany, from which however the Japanese form is distinguished in having longer segments traversed by a greater num- ber of veins (in I). Bronguiarti, 5-6). Our fossil may also be com- JUKASSIC PLANTS FEOM KAGA, UIDA, àND ECHIZEN. 45 pared to the Gristhorpe forms such as 2). medianus Beau and D. an- gustif alius Beau (Leckenby, Brae. Geol. Soc, vol. XX, 1S64, p. 77, pi. VIII, fig. 2 and 5), the first of which however has shorter, and the second, more straight and gradually sharpening segments. Rather rare. Loc. — Shimamura, Tanimura. 4. Zamites Brgt. 29. Zamites parvifolius Geiß. Zamites parvifolius- Geyler, Ueber foss. Pflanz, a. d. Juraform. Japans, p. 227, pi. XXXII, fig. 2a. Zigno, Flora Foss. Form. Oolith., II, p. 57. I have not been able to find any specimen that can be refera- ble to this species. 5. Podozamites Fr. Braun. 30. Podozamites laneeolatus Lindl. et Hutt. sp. Podozamites lanceolatus-Geyler, Ueber foss. Pflanz, a. d. Jura- form. Japans, p. 228, pi. XXXII, fig. 1, 4, XXXIII, 1-3, 4b, XXXIV, 3a, 5b. Heer, Beitr. z. Juraflora Ostsib. u. d. Amur]., 1876, p. 45, pi. I, fig. 3a, p. 106, XXIII, lc, 4a, b, c, XXVI, fig. 2-10, XXVII, 1-8. Beitr. 1878, p. 6, p. 20, pi. V, fig. 1-11. Beiträge zur foss. Flora Spitzbergens, p, 35, pi. VII, fig, 1-7 c, d. Schmalhausen, Beitr. zur Juraflora Russl., p. 29, pi. V, fig. 3, 4, 5c. Schenk, in Richthofen's China, vol. IV, p. 248, pi. XLIX, fig. 4, 5, p. 251, L, 1-6, p. 255, LT, 3, LH, 8, p. 258, LI, 7, p. 261, LIV, 2c. Feistmantel, Foss. Flora of Jabalpur Group, p. 11, pi. HI, fig. 4-7, IV, 1-10. This well known species is very widely spread in the Jurassic flora, being known from Yorkshire, Spitzbergen, Russia, Siberia, 4fi M. YOKOYAMA. Mongolia, China, India, Persia and Tongking. From the Rhaetic it lias been described under the name of Podozamites distans. In Japan it occurs in great abundance and can be distinguished into the follow- ing varieties : — a.) P. lanceolatus var. genuina Heer. PL IV, fig. 2. PI. VII, fig. 8b. PI. XIV, fig. 12b ? This variety is distinguished by having long narrow leaflets, each of which is drawn out into an acuminate apex. The leaflets of this variety are frequently met with at Shima- mura and Okamigo. Fig. 2, PL IV, has the upper part of the leaflet strongly falcate, is about 50 mm. long and 9 mm. broad, and is fur- nished with about 25 fine longitudinal veins. Fig. 8b, PI. VII, has the apical end broken off; but, to judge from its superiorly gradually narrowing" form, it is evident that it has terminated in an acuminate apex. It measures 5'5 mm. in breadth, and has about 27 veins. Fig. 121), PI. XIV, from Hakogase is a doubtful case, but its general form points more to this than to any other variety. It possesses 25 veins, and is 5*5 mm. broad near the base. b.) P. lanceolatus var. intermedia Heer. PI. IV, fig. 3a, 4a. PL V, fig. 3, 7, 9. Leaflets in this variety are gradually narrowed towards the apex which is bluntly pointed. Specimens are by no means rare at Shimamura and Ohnnigd, but those from the former locality are alone figured here. Fig. 3, PL IV, which is the longest of the specimens with both apex and base preserved, measures 83 mm. in length and 12*5 mm. in breadth; fig. 4a, which is shorter, is 55 mm. long and 14-15 mm. broad, whilst fig. JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 47 3, PI. V is 57 mm. by 105 mm. Fig. 7 and 9, PI. V, represent leaflets which seem to have been longer when compared with the above three, and are indeed the typical specimens of this variety, the shorter ones representing what mri}7 be considered as passage forms between this and the succeeding group. Veins vary between 20 and 25 in number. e.) P. lanceolatus var. Eiehwaldi Heer. PI. IV, tig. In, 4b, c. PL V, fig. 2, 4, 5a, 1). c, 6. PI. XI, fig. 6.? Leaflets are more or less parallel-sided, with apex obtuse or nearly rounded. As in Bureja, China and Spitzbergen, this is the most frequent of all the varieties of P. lanceolatus occurring in Japan. The speci- mens I have figured vary considerably in size and breadth, some being much smaller and some much longer than others. Smaller forms are represented in fig. 1, 4, PI. IV, and fig. 5, PI. V, of which the last is perhaps the smallest that has as yet been figured, though not completely preserved. Fig. 4, PI. V, is a broader form and corresponds to the. typical figures of Heer (comp. Beitr. 1876, pi. XXVII, fig. 1). So is fig. 6. Fig. 26, PI. V, seems to have been considerably longer than others; but as it has the sides nearly parallel and the apex blunt, it must be referred to this variety. A leaflet (PI. V, fig. 2a) which I formerly considered as Podozamites imlcliellus Hr. (Bull. Geol. Sec. Japan, B. Vol. I, p. 6) is probably an abnormal form of P. lanceolatus Eiehwaldi, as it seems to have been furnished with a short stalk. So is the leaflet from Ushimaru (PI. XI, fig. 6) with 25 veins and men- tioned in the Bulletin as a new species. Found at Shmamura, OJcamigo, Tanimura and Uhimaru. 48 M. YOKOYAMA. d.) P. lanceolatus var. minor Heer. PI. V, fig. 8. A single specimen with three leaflets attached to a rhachis was obtained at Shimamwa. These leaflets are narrowly lanceolate with acute apex and with about 18 veins. Heer says of his variety minor, that the leaflets are narrow, linear-lanceolate with sharply pointed apex and with 14-10 veins (Beitr. 1876, p. 110). To this our speci- men must be referred, so like is it to the figures of Heer (Beitr. 1876, pi. XXVII, fig. 7, 8), although in number of veins it slightly exceeds the Siberian. A leaflet on the right hand side of our drawing is very un- favourably situated. It looks narrower than if actually is. It measures about 6*5 mm. in breadth, and is not at all narrower than others which also possess a nearly equal breadth. e.) P. lanceolatus var. latifolia Heer. PI. IV, fig. lc. PI. V, fig. 1. Fl. VI, fig. 1. This variety is distinguished by having larger, longly-oval, obtuse leaflets with 20-30 veins. The specimens, though all incomplete, I. cannot but identify with this variety. Fig. lc, V\. IV, which is wanting both at apex and base, measures 20 mm. in breadth, and is pierced with about 30 veins. Fig. 1, PI. V, shows two leaflets, one with the apex and the other with the base. These leaflets agree in form with those figured by Heer from Spitzbergen (Beitr. 3. Foss. Flora Sjiitzh.. pi. VIII, fig. 3). Veins are very numerous — about 30 in number-and the breadth in the broadest part of the blade is 19 mm. Fig. 1, PI. VI has 28 veins and is 17 mm. broad. Specimens were obtained at Shimarnura where they seem to have been rather rare, JURASSIC PLANTS FROM KAGA, HID A, AND ECHIZEN. 49 f.) P. laneeolatus var. brevis Schenk. PL XII, fig. 18. Leaflets small, longly oval ; apex acute ; veins fine, about 20 in number. A specimen from Okamigö is longly-oval in shape, 30 mm. long and 11 mm. broad with the broadest part near the middle of the leaflet. It is pointed at apex, and possessed of about 20 delicate veins. It agrees very well in general form with one figured by Schenk ( Richthof en' s China, vol. IV, p. 251, pi. L,fig. 1) from Patat- shu, west of Peking in China, though ours is somewhat longer. g.) P. laneeolatus var. PL V, fig. 5d. Leaflets small, abruptly narrowed above and ending in acute apex. A leaflet which I have here separated from others is very small when compared with the leaflets of other varieties. It corresponds in size to the smallest form of P. laneeolatus Eichwaldi already alluded to (PI. V, fig. 5), from which it differs in having the blade quickly narrowing above and ending in pointed apex. It is 24 mm. long and 6 mm. broad, and possesses about 20 fine veins. It may be a young leaflet belonging to one of the many varieties of P. laneeolatus, which however is very difficult to determine. 31. Podozamites tenuistriatus Geyl. PI. XII, fig. 19. Leaflets lanceolate, gradually attenuated above, suddenly nar- rowed at base, acute; veins fine, 14-16 in number; interstitial veinlet s very fine. Podozamites tenuistriatus-Geyler, Ueber foss. Pflanz, a. d. Jura- form. Japans, p. 228, pi. XXXII, fig. 2 b. 50 M. YOKOYAMA. Dr. Geyler established this species on a leaflet from the Tetori- gawa-valley, which was 50 mm. long and 10 mm. broad, and pierced with 14-16 fine veins. The only specimen which I could obtain of this species is a leaflet from Okamigö, 32 mm. long and 10 mm. broad, therefore somewhat shorter than the one figured by Geyler. It is furnished with about 15 very faint delicate veins between which I have often observed one or two extremely fine veinlets. Its form is lan- ceolate with the broadest part near the base, and above gradually nar- rowing into a rather acute apex. Podoz. cnsiformis Hr., with which Geyler compares this species, has much slenderer leaflets pierced with a somewhat less number of coarser veins. I may here advance my opinion concerning the fragmentary speci- mens of a Podozamitcs identified with P. cnsiformis Hr. by Geyler (Foss. Vflanzcn, p. 227, pi. XXXII, fig. 1). This author gives for his specimens 20-22 veins, whilst Heer's species possesses only 10-13 (Bcitr. 1876, p. 46) ; and he rejects Zamites Sclimidtii Andr. from the Liassic flora of Steierdorf for the very reason that the latter has 14-16 veins. Here evidently Dr. Geyler is mistaken; and this mistake has been noticed by Heer (Naclitr. p. 3) who considers the Japanese specimens as referable to P. tcnuistriatus Gcyl. But, as already mentioned, they possess 20-22 veins, whilst leaflets of P. tcnuistriatus have only 14-16, only a little more than in P. cnsiformis. Therefore I think it most proper to refer the specimens described by Dr. Geyler as P. cnsifor- mis Hr. to P. lanccolatus var. gcnuina Hr., to which indeed his broken leaflets show the greatest resemblance. Loc. — Okamigd. 32. Podozamites Reinii Gcyl. PI. Ill, fig. 6a b c. PI. IV, lb, 3b. PL YI, fig. 2, 3b c d, 4-7, 8a bee. PI. IX, fig. 12a. PL XII, fig. 4. Leaflets alternate, distant, more or less directed forward, ovate, JURASSIC PLANTS FROM KAGA, HID A, AND ECHIZEN. 51 oblong, or nearly round, obtuse at apex, unequal at base, shortly petiolecl; veins numerous, simple, parallel, 34-50 in number; inter- stitial veins very fine. Podozamites-Remii Geyler, Ueber foss. Pflanz, a. d. Juraform. Japans, p. 229-230, pi. XXXIII, fig. 4a, XXXIY, 1, 2, 5a, XXXIY, 3 b, 4. Leaflets of this Japanese species of Podozamites are very plentiful- ly found at Shimamura and Okamigö, where they show great diver- sities in shape, which however are, I believe, to be considered merely as a variability so frequent among the leaves of cycads. On this ac- count I do not adopt the varietal names of latifolia and angustifolia proposed by Dr. Geyler. The shape of the leaflet is either ovate (e. g. fig. 2, 6, PI. VI), or oblong (e. g. fig. 3b, PI. IV, fig. 3c, 7a, PL VI) which often ap- proaches to elliptical (e, g. fig. 7b, PI. IV, fig. 8c, PI. VI, fig. 12a, PI. IX). Between these forms there are all sorts of gradations from one to the other. A small round leaflet represented in fig. 4, PI. XII, 11-12 mm. in diameter and with very fine veins a little over 30 in number is, I believe, an abnormal form of this species. The base of the leaflet is either subcordate (PI. Ill, fig. 6a, PI. VI, fig. 2, 4, 6), or it passes gradually into a petiole without leaving any indentation on either side of it (PI. Ill, fig. 6c, PI. IV, fig. lb). It is to be noticed here that in all of the leaflets of this species interstitial veins are in most cases distinctly observable, generally one (fig. 6e, PI. Ill), but sometimes two between the principal ones; and that in all of them the two sides of the base are more or less unsym- metrically formed, as may be seen most prominently in fig. 6c, PI. Ill, fig. 3b, PI. IV and fig. 3b, 3d, PI. IV. Loc. — Shimamura, Ozö, Yanagidani, Tanimura, Olcamifjo. 52 M. YOKOYAMA. 33. Podozamites sp. PL XII, fig. 12. This is an imperfect specimen which is wanting both at apex and base. It is attached to a stone in a curved state. It is 3 mm. broad, and may have been long drawn out at apex. Veins are distinct, about 8 in number with a finer one between. The leaflet is probably referable to Podozamites, and indeed I mentioned it in the Bulletin as a new species. But now I consider it better to be designated with a simple name of Podozamites sp., as the specimen is not complete enough for exact determination. Loc. — Ohamigo. 34. Podozamites sp. PL VII, fig. 9. A leaflet longly elliptical in form, obtuse at apex, nearly sessile at base, and furnished with 12 rather distant parallel veins. It meas- ures 30 mm. long and 12 mm. in breadth, and resembles much the narrower forms of P. pulchellus Heer (Beitr. z. foss. Flora Spitzb., p. 38, pi. IX, fig. 10—14), from which however our specimen seems to differ in possessing 5-10 very fine dense interstitial veins. Whether Heer' s species may have so many veinlets as in ours is a question which can only be settled after the discovery of a much greater num- ber of specimens. Loc. — Simamura. 6. Dictyozamites Oldham. u Leaflets many-veined, subauriculate at base; veins dicholomous, reti- culated." (Oldham.) Oldham and Morris had described in their Rajmahal Flora a very interesting plant which then accorded most with the genus Diclyopteris Gutbier of the Carboniferous System, and therefore called by them JURASSIC PLANTS FEOM KAGA, HIDA, AND ECHIZEN. 53 Dictyoptcris falcata. Oldham, however, who had at first followed the opinion of his colleague afterwards came to the conclusion, that the plant in question is not a fern, as at first appeared, but a cycad near to the genus Otozamites Br. His reasons were : Firstly, that the Indian leaflets possess not the slightest trace of a midrib which in Dictyopteris is present as a " quasi-midrib or bifid midrib," from which " all the veins of the fronds diverge in a flabel- late or radiated form," while in the former " the strong nerves pass out parallel to each other from the base of the leaf, and proceed to- wards the apex in nearly right lines interrupted only by the anasto- mozing or reticulating cross nerves, which pass from one to the other: the areolae thus become subquadrangular, not hexagonal." Secondly, that the texture of the leaflets is coriaceous. Thirdly, that the equal leaflets are regularly disposed along a rhachis which is provided with a terminal leaflet. And then he proposed for the Indian plant the new generic name of Dictyozamites. Dr. Stur of Vienna also came to the same conclusion by examin- ing the figures of this plant. The above view of these two eminent men has also been adopted by Dr. Feistmantel who fully treats of this subject in his "Indischen Cycadeengattungen Ptilophyllun and Dictyozamites " (p. 17). I possess a set of leaflets from Kaga and Hida, which I treat under the name of 36. Dictyozamites indicus Feistm. var. distans m. PI. X, fig. 4-10. PL XI, fig. 5. Leaf simple, elongated ; leaflets either short and blunt, or long, falciform, and acute at apex; attached to the rhachis with, the middle of the base which is either very shortly petioled or quite sessile and 54 M. YOKOYAMA. distant from one another ; corners of the base distinctly auriculate ; leaf terminated by a single leaflet ; veins numerous, fine, rising at base, and radiating and anastomozing ; thus forming nets which are long and sub- parallel in the middle of the leaflet, shorter and polygonal towards the apex and the margin. Dictyozamites indicus-F eistmante], Jurassic Flora of the Rajmahal Group, in the Rajmahal Hills, p. 70, pi. XLVI, fig. 7, 8. Jurassic Flora of the Rajmahl Group from Golapili, p. 180, pi. II, 5, 6. lieber die indischen Cycadeengattungen Ptilophyllum Morris und Dictyo- zamites Oldham, p. 18, pi. IV, fig. 7, 7a, 8, pi. V, fig. 1-4, pi. VI. Dictyopteris falcata and Dictyopteris falcata var. obtusifolia-Oldham and Morris, The Fossil Flora of the Rajmahal Series, Rajmahal Hills, Bengal, p. 38, pi. XXIV, fig. 1, la, pi. XXIV, fig. 2, 2a. This species is greatly variable in the form of the leaflets. Feist- mantel had already united Morris' species and variety. The distinguish- ing characters are the ending of the leaf at apex, the mode of attach- ment of the leaflets to the rhachis, the basal ears, and the reticulation of fine veins. The first two of these characters are unfortunately not to be seen in our specimens, as most of them occur in an isolated state. A single specimen was obtained with the rhachis preserved, but the preservation is such that the important character of the attachment of the leaflets is not distinctly observable. The basal ears are very distinct in fig. 4, 5a, 6, 8 and 9. The leaflets are more or less oblique with upper part often curved to one side. Fig. 5, PI. X, represents a leaflet which has the apical portion most strongly falcate. Fig. 7 shows one with obtuse apex, which corresponds to Feistman- tel's 'short and blunt' while fig. 5, 8, and 10 represent his 'long and falciform ' leaflets. These latter are all more or less sharp at apex, and fig. 10 may indeed be said to possess an acuminate apex. Most of our leaflets seem to have been sessile, as we see no in- JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 55 dicatioD of a stalk except in fig. 8, which shows a short but distinct one. The mode of venation is exactly as Feistmantel describes and figures. Fine, equal, numerous veins proceed from the base to the apex and margin of the leaflet, in such a way as to form long and narrow nets along its median line, and shorter and polygonal ones around its periphery. Agreeing in all of the above mentioned characters, our plant is distinguished from the Indian one in having the leaflets not so closely set along the rhachis as in the latter. On this account I think it better to raise it to a distinct variety (see fig. 4). This species is said to be highly characteristic of the Eajmahal Group in India. Its discovery therefore in the Jurassic system of Japan is especially noteworthy, as at Ozô it has been found associated on the one hand with a Sagenopteris closely akin to S. rhoifolia Presl. and on the other with the common Oolithic types such as Asplenium argutuhim, A. distans, and Onychiopsis elongata. This is the most numerous fossil at Ozo, where it takes the place of Podozamites lanceolatus of other localities, of which indeed not the least trace is here to be found. Rarer at Ushimaru. Loc. — Ozo, Ushimaru. 36. Dietyozamites grossinervis m. PL VII, fig. 10. Leaflets longly ovate, very shortly petioled at the middle of the base ; basal corners subauriculate ; veins equal, coarse, rising at the base of the leaflet and radiating, forming elongated nets which are longer in the middle and shorter towards the apex and margin of the leaflet. This species is closely akin to the preceding, in fact agreeing with it in the general form of the leaflets, their mode of attachment to the 56 M. YOKOYAMA. rhachis, the auriculation of the basal corners, and in the general mode of venation, but is decidedly distinguished in having a less number of coarser veins with a correspondingly less number of nets, and in having the leaflets set more closely together. Feistmantel indeed figures leaflets which show a small number of aréoles (c. g. Gycadcen- gatlungen, pi. IV, fig. 7, VI, 4); but all of these pertain to much smaller forms. Our specimen shows larger leaflets with few nets. Therefore it is quite unsafe even to create a variety out of our specimen, the number and size of veins being considered as very im- portant characters in the discrimination of fossil plants. However, there is no doubt that our plant is very nearly related to Dictyozamites indiens, and I regard it as belonging to the same genus. Our specimen is very meagre. The apices of the leaflets could not be well exposed, therefore it is very difficult to decide whether they ended acute or obtuse. Very rare. Loc. — Shimamura. Fam. 2. Cycadeae, 7. Cycadeospermum Sap. 37. Cycadeospermum japonieum Geyl. Cycadeospermum japonicum-Geyler, Ueber foss. Pflanz, a. d. Jura- form. Jap., p. 231, pi. XXXIII, fig. 5. I have not been able to find any specimen referable to this species. Order 2. Coniferae- Fam. 1. Taxaceae. 1. Ginkgodium m. Leaf coriaceous, entire or lobed, gradually narrowed towards ike base JUEASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 57 which is thickened at its margin and gradually passes into a short petiole. Veins numerous, simple, parallel: interstitial veins very f ne. The leaves which I am going to describe below under this genus, I brought under Baiera in the Bulletin, But now I have reason to believe that they must be treated as a distinct genus, relat- ed on one side to Ginkgo and on the other to Baiera. With the former it has in common the thickening of the lower margin of the leaf and with the latter the numerous, simple, parallel longitudinal veins. But these veins coming down directly on the thickened mar- gin without converging, decidedly distinguish oar plant from the two above named ones. This mode of venation may be compared to that of Whittleseya Netub. from the Coal- Flora of Pennsylvania (Les- quereux, Descript. of the Coal-Flora of Pennsylvania, 1880, vol. I, PL IV). The discovery of this genus standing between Ginkgo and Baiera shows more strongly the close relationship existing between these genera which was first pointed by Heer (Beitr. z. Juraflora Ustsib. u. d. Amurl., 1876, p. 51). 38. Ginkgodium Nathorsti m. PI. II, fig. 4e. PI. Ill, fig. 7. PI. VIII. PI. IX, fig. 1-10. PI. XII, fig. 14, 15. Leaf coriaceous, gradually attenuated below into a short petiole, entire or lobed ; apex obluse. Longitudinal veins dense, simple, parallel ; in- terstitial veins very tine, simple. The leaves which I mentioned in the Bulletin (page 8) as three new species of Baiera, 1 now unite into one, as I was convinced by the examination of many specimens that an apparent diversity in the shape of the leaves is merely a variability. Some leaves are broad and more or less narrowly fan-shaped, and are nearly or quite entire (PI. 00 M. YOKOYAMA. VIII, fig. 4, PI. IX, fig. 1, PJ. XII, fig. 15), or deeply lobed (PI. VIII, fig. 2a, 5, 6, PI. IX, fig. 2, PI. XII, fig. 14). Some are quite entire and lanceolate with subparallel sides (PI. All I, fig. 2c, PI. IX, fig. 9, 10). Between these forms there are all sorts of grada- tions, both in shape and in the depth of the central slit, just as in our recent Ginkgo biloba. The lower margin of the leaf is thickened as in Ginkgo, but a little more strongly than in the latter. The petiole, however, is very short in comparison to the very long one of Ginkgo. The lower end of the petiole which is the point of attachment to the stem, is slightly expanded as is best seen in fig. 2a, PI. VIII, and fig. 5, PI. IX. As to the number of veins, they vary from 20 to 40, according to the breadth of the leaf. In the lobed leaves 20-30 come in a lobe, and in the lanceolate entire ones we find 30-35. For example a leaf represented in fig. 9, PI. IX, which is 66 mm. long and 21 mm. broad, possesses 30 veins with an equal number of interstitial veins, and that represented in fig. 2c, PI. VIII, which is a little longer, has 35. Fig. 11, PI. VIII, representing a small leaf only 13 mm. in breadth, shows 23 veins. In leaves of an oval shape (PI. VIII, fig. 14, PI. XII, fig. 15), the veins vary between 30 and 40. In all of our specimens, interstitial veins are more or less distinct- ly observable and are always single (PI. VIII, fig. la, 11a, PI. IX, fig. 10a). It is here to be added, that when the leaves are lobed they seem to have been always two-lobed, the apparent anomalies seen in some of the figures being only accidental. Very numerous at Shimamura, but rarer in other localities. Loc. — Shimamura, Yanagidani, Okamigo. JUEASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 59 2. Ginkgo Thunb. 39. Ginkgo digitata Brgt. PI. XIII, fig. 2. Leaf semicircular, entire or with two to six lobes, petioled; pe- tiole long, slender, canaliculated above; veins numerous, repeatedly dichotomous, flabel lately divergent. Ginkgo digitata-Heer, Beitr. z. foss. Flora Spitzb., p. 40 pi. X, fig. 1-6. Heer in Regel's Gartenflora, 1874, pi. 807. As G. digitata integriuscula Heer, Beitr. z. foss. Flora Spitzb., p. 44, pi. X, fig. 7-9. Beitr. z. Juraflora Ostsib. u. d. Amur]., 1878. p. 25, pi. VI, fig. 5, 6. Nachträge, p. 5. Baiera digitata-Schim^er^ Pal. Vegét., vol. I, p. 423. Cyclopteris digitata-Brgt., Végét. foss. I, p. 239, pi. 61, fig. 2, 3. Zigno, Flora Foss. Form. Oolith., I, p. 102. A leaf of this species in an excellent state of preservation was procured from Okamigö. It measures 30 mm. in length and 52 mm. in breadth. It is fan-like in shape, and furnished with three compar- atively shallow clefts on the top, one of which is nearly in the middle of the leaf and attains the depth of one-third of its length. The other two are on each side of the central one and are much nearer to it than to the lateral margins of the leaf. They are about one-half as deep as the central one. Veins are very distinct, all of them diverging from the base of the leaf and ending in its upper extremity, after having repeatedly forked on their way. The Japanese specimen, according to the descriptions of Heer, must belong to his variety quadriloba (Foss. Flora Spitzb., p. Fl. X, ßg. 3a), although the central incision is much shallower than in the one figured by that author. Heer described in his contributions to the Spitzbergen and Sibe- 60 M. YOKOYAMA rian floras closely allied, nearly entire-margined leaves as Ginkgo in- tegriuscula Hr. But this species he afterwards placed under G. digitata as a variety only (Nachtr. p. 5), as Prof. Nathorst discovered in the Oolite of Scarborough specimens which show passage forms between these two species. Seems to have been very scarce in Japan. Loc. — Okamigô: 40. Ginkgo ef. lepida Heer. PL XIV, %. 10. Ginkgo lepida-Wccr, Beitr. z. Juraflora Ostsib. u. d. Amur]., 1876, p. 62, pi. XII, pi. VII, fig. 7. XTachtr., p. 17, pi. IV, fig. 7b, 9-12, pi. V, fig. In, 2, 3a, 4. A leaf which I figure here is not very complete. It is divided into four lobes, and therefore furnished with three slits, the central of which is the deepest, nearly reaching to the base of the leaf whence it goes off into a petiole. The petiole however is unfortunately not preserved. The other two slits are shallower. The lobes are narrow, parallel-sided and acute at apex, as is seen from one of them. Veins are parallel to the sides and about 6 in a lobe. Judging from these characters, our plant is probably referable to the above named species of Heer, which is distinguished from the very nearly related G. sibirica Hr. in possessing narrower and especially acute lobes. Very rare. Loc. — Hakogase. 41. Ginkgo sibiriea Heer. Ginkgo sibirica-G eyler, Ueber foss. Pflanz, a.d. Juraform. Japans, p. 231, pi. XXXII, fig. 6. Heer, Beitr. z. Jnraflora Ostsib. u. d. Amurl., 1876, p. 61, pi. VIT, fig. 6, IX, ob, XI, p. 116, pi. XX, fig. 3b,. 6c, XXII, 3, Beitr. 1878, p. 25, pi. VI, fig. 8a b. Nacht r. p. 16, pi. IV, fig. 13, V, 5-8. Schmalhausen, Beitr. z. Jnraflora Russh, p. 34. JUKASSIC PLANTS PEOM KAGA, HIDA, AND ECHIZEN. 61 I have not been able to find any specimen undoubtedly referable to this species. 3. Czekanowskia Heer. This genus is doubtfully represented in Japan, and only by fragments which most approximate to 42. Czekanowskia rigida Heer (?) PI. XII, fig. 11. PL XIII, fig. 10. An indistinct specimen from Ozö (fig. 10, PI. XIII) has linear leaves about 1 mm. broad, which dichotomize and are in some cases seen with a distinct median vein. Leaves on the right-hand side of the stone appear as if fasciculated. Fig. 11, PI. XII, shows a single leaf 1 mm. broad, and once forked with a distinct canal running throuo'h the lobes. From these characters our fragments seem to be referable to the above denominated species which is found in Siberia, China, Russia and England, and also in the Rhaetic of Sweden. (Comp. Heer's Beiträge z. Juraflora Ostsib. u. d. Amur]., 1876, p. 70, PI. V, fig. 8-10, PI. VI, 7, X, 2a, etc. Beitr. 1878, p. 7. PI. I, fig. 16, p. 76, PI. IV, fig. 3b c. Schenk in Pichthofen's China, vol. IV, p. 251, PI. L, fig. 7, p. 262, PI. LTV, fig. 2a. Schmalhau- sen's Beitr. z. Juraflora Russl. p. 36, PL V, fig. 2e, 6a, 7. etc.). Loc. — Ozo, Okamigo. 4. Taxites Brogt. 43. Taxites sp. PL X, fig. 15-19. PL XII, fig. 16, 17. The leaflets in the above cited figures all belong, I believe, to one and the same species. They are linear-lanceolate, 10-15 mm. long and 2.5-3 mm. broad, obtusely pointed at apex and constricted at base, with a weak midrib. In form they remind us of many of the so-called Cijeaditex, e. g. such as C. zcimioides Tueckenby (Proc. Geol. 62 M. YOKOYAMA Soc, 1864, vol. XX, p. 77, PI. VIII, fig. 1) and C. Saladini Zeiller (Examen de la Flore Fossile des Couches de Charh. du Tonking, p. 322, PI, XJ, fig. 8, 9, 10 A, PL XII, fig. 8, 8 A, 9, 9 A, 10). But the former seems to have had much more linear leaflets with a stronger midrib, and the latter is said to have had a cordate base. Our plant may be most closely allied to Taxites brevifolius Natli* from the Oolite of England whose original specimens I had an opportunity to examine at Stockholm. In the Bulletin I mentioned the leaflets now under consideration as a new species of Cycadilcs. But as Prof. Nathorst believes them to be undoubtedly the remains of a conifer, and as they occur only in an isolated state, I now satisfy myself by calling it simply Taxites sp. Loc— Ozo, Okamigd. 44. Taxites .sp. PI. A7 1, fig. 3a. Two fragments of linear leaves. 2-3.5 mm. broad, gradually nar- rowed above and furnished with a distinct evanescent midrib, on both sides of which are seen fine longitudinal striae. I mentioned them in the Bulletin as belonging to Cycadites gramineus Hr. (?). But now I take the opportunity of bringing them under Taxites like the preceding species, and indeed very near to T. longifolius XatJi. occurring in the Rhaetic formation of Schonen in Sweden. Loc. — Shimamura. Fam- 2. Abietaceae. 5. Pinus L. 45. Pinus cf. prodromus Heer. PL XII, fig. 3. Pinus prodromus-Yieer, Beitr. z. loss. Flora Spitzb., p. 44, pi. VII, * Nathorst, Berät tehe, afgifven till Kongl. Vetenskaps- Akademien, om en vied under stöd nf allmänna medel utförd vetenskapliga resa till England, p. 73 JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 63 fig. 7a, X, 11-14. Nachträge z. Juraflora Sibir., p. 27, pi. VII,' fig. 12c. Many linear leaves mostly 1 mm. in breadth, but sometimes a little broader, are found with a strong midrib, on both sides of which we can often observe fine longitudinal striae. Though they are always found scattered and only in fragments, I believe they are to be identified with the species discovered in Spitzbergen and Siberia. A fragment of a leaf of this species is also found by the side of Adiantites Heerianns. (Fig. 2, PI. XII). Loc. — Shimamura, Tanimura. 46. Pinus Nordenskjoldi Heer. PL IX, fig. 12b. Cyclopitys Nordenshjoldi-Y okoyama, Bull. Greol. Soc. Japan, Part 13, vol. I, No. 1, p. 8. ? (?) Cyclopitys Nordenskjoldi— Schmalhausen, Beitr. z. Juraflora Russlands, p. 40. Nachträge zur Juraflora d. Kohlenbassins von Kuznesk am Altai, pl. I, flg. 1. Firnis Nordenskjoldi-lAeeY, Beitr. z. foss. Flora Spitzb., p. 45, pl. IX, fig. 1-6. Beitr. z. Juraflora Ostsib. u. d. AmurL, 1876, p. 76, pl. IV, flg. 8e, p. 117, pl. XXII, fig. 4a, b, XXVII, 9a, XXVIII, 4. Beitr. 1878, p. 26, pl. II, fig. 7-10. Nachtr.. p. 2S, pl. I, 8b, 6b, IX, 36. Long linear acuminate leaves, 2-2.5 mm. in breadth, thick in texture and scattered on a stone, agree very well with those described under the above name from Spitzbergen and Siberia. The central vein is tolerably thick, and through a magnifier we can observe in some leaves 3-4 finer longitudinal parallel veins, a fact which was also noticed by Heer (Beitr. 1876, p. 117, VI. XXVIII, fig. 4c). Dr. Schmalhausen identifies the isolated leaves described by Heer as belonging to a pine with his Cyclopitys Nordeushjoldi in which the 64 M. YOKOYAMA needle-shaped leaves are arranged in whorls as in our recent Sciadopi- tys. In the Bulletin I followed the above author in the generic deno- mination of our species. But in reverence to Prof. Nathorst of Stockholm who thinks it still unsettled whether the plants described by Heer and Schmalhausen really belong to one and the same species, I resume the older denomination of Heer. Loc. — Shimamura. 6. Palis sy a Endl. Species of this genus are mainly known from the Rhaetic and Lower-Liassic beds of Europe, but also from the Rajmahal, Jabalpur and Kach series in India. 47. Palissya sp. PI. IX, fig. 11. A single specimen of a branch of a coniferous tree belonging to the family of Abietaceae shows linear-lanceolate leaves, 10-11 mm. Ions' and 1^ mm. broad, constricted at base and decurrent to the stem. Each of these leaves is pierced with an indistinct midrib. This species is undoubtedly very closely akin to P. jabalpurensis Fcislm. (Flora of the Jabalpur Group, p. 16, VI. IX, fig. 1) from the Oolite of India. It is also not unlike a Palissya figured by Feistman- tel, and described as closely related to P. inclica Feistm. (Jurassic Flora of the Rajmahal Group, PL XLV,fig. 9) from the Liassic series of the same country. To our great regret the Japanese specimen has the apex of all the leaves broken, and so it does not allow us to institute a stricter comparison with the above fossils. Loc. — SJiimamura. Incertae Sedis. 48. Vallisneriites jurassieus Heer. (?) PL III, fig. 8. PI. XIII, fig. 5-8. Fragments of very long parallel-sided leaves, 2t-5 mm. in JURASSIC PLANTS FROM KAGA, HIDA, AND ECHIZEN. 65 breadth, resemble very much those described by Heer under' the above name from Ust-Balei in Eastern Siberia (Beitr. z. Juraflora Ostsib. u. d. Amurl., 1878, p. 8, PL I, fig. 22-27). It is very much to be regretted that in nearly all of our specimens, the surface of the leaf has become entirely smooth, and even in some in which we can trace very faint delicate longitudinal striae, these striae are by no means s<; distinct as in the Siberian specimens. Fig. 8, Fl. Ill, represents the longest leaf among our specimens. It is complete at one end, but broken at the other. Its length is 150 mm. and breadth 3.5-4.5 mm. I obtained only one specimen from Shimamura, but a good many from Ozo. 49. Carpolithes ginkgoides m. PI. X, fig. 20-23. Fruit ovate, sharply edged, roundel 'at base, sharply beaked at apex. There occur many small ovate bodies in the black shale of Ozo, which are furnished with a sharp beak at the apex. This beak in one specimen (fig. 22) is very long drawn out. The length of these bodies is 7-1 1 mm. and the breadth 3.5-1.5 mm. They are all pro- vided with a sharp longitudinal edge which is very distinct in fig. 20 and 23. Their surface, when examined through a magnifier, reveals delicate longitudinal striae. These are perhaps nuts of Ginkgo, like those of Eastern Siberia, considered by Heer as belonging to G. Sibirica Heer { Beitr. '2. Jura- flora Ostsib. u. d. Amurl. . p. 58, PL XI, fig. 13-16). Our specimens look most like fio\ 13 of Heer; however they are all much narrower in form than those of Siberia. . Loc. — Ozo. 66 Index of the Species described. A. Adiantites Heerianus ... „ Kochibeanus „ lanceus Anomozamites sp. Aspleniuui argutulum „ dis tan s „ whitbiense c. Carpolithes ginkgoides Cycadeospertnum japonicum. Czekanowskia rigida D. Dicksonia acutiloba „ cf. Glehniana „ gracilis ,, nephrocarpa Dictyozamites grossinervis. „ indicus var. dis- tans Dioonites Kotoei E. Equisetum sp „ ushimarense G. Ginkgo digitata „ cf. lepida „ sibirica Ginkgodiiun Nathorsfci M. Macrotseniopteris cf. Rich- thof eni Page. 13, 28 29 30 40 13, 32 13, 32 13, 31 65 15, 56 61 13, 24 13, 25 13, 24 13, 25 15, 55 15, 53 15, 44 14, 39 39 15, 59 16, 60 15, 60 15, 57 14, 37 N. • • • Page- JNilssonia mpponensis 15, 42 „ orientalis 15, 40 „ ozoana 41 » (?) 43 o. Onychiopsis elongata 13, 27 P. Palissya sp 16, 64 Pecopteris exilis 14, 35 „ Saportana 14, 36 Pinus Nordenskjoldi 16, 63 „ cf. prodromus 16, 62 Podozamites lanceolatus ... 14, 45 „ Reinii 14, 50 „ tenuistriatus 14, 49 „ sp 52 s. Sagenopteris sp). 38 Sphenopteris sp 34 T. Tasniopteris (?) 37 Taxites sp 61 sp 62 Thyrsopteris kagensis ... 23 „ Murrayana ... 13, 22 „ prisca 13, 23 V. Vallisneriites jurassicus ... 64 z. Zamites parvifolius 14, 45 On Pyroxenic Components in certain Volcanic Rocks from Bonin Island by Yasushi Kikuchi. With Plate XIV bis. The following account of the Pyroxenic components in the rocks of Bonin Island is based upon the investigation of materials collected by myself, during a short visit to that island in November, 1887. Ogasawarajima or Bonin Island is the name applied to an island- group of volcanic origin, apparently forming part of the chain of of volcanoes which extends from Central Japan through the Shichitd Group of the Prov. Izu. Themost important of the Bonin Island- Group is that known as Chichijima (Peel Island, Lat. 27° 5' N., Long. 142° 15' E. Green.) It has a spacious harbour known as Futamikö (Port Lloyd) which without very good foundation has been regarded as the relic of a volcanic crater. The geological col- lections were principally made in this island, and also a fewT in Ha- hajima (Coffin Island) which is situated about 20 geogr. miles S.S.W. of Chichijima. The volcanic activities seem to have subsided Ions: since in these islands, and consequently their general features differ from those of such islands of the Shichitd or Volcano Group which have active volcanic cones. The interior is in some places covered with dense tropical forest, among which rise numerous pointed 68 Y. KIKUCHI pinnacles and ridges, due to the extrusion of the harder rock- masses by erosion. The most prominent of these ridges rises about 1100 ft. above sea-level. The volcanic rocks are chiefly Andésite ; sometimes in a mas- sive state, occasionally showing prismatic joints ; sometimes interstra- tified with tuffs. These tuffs usually contain fragments of volcanic rock, of various sizes, forming agglomerates. The volcanic rocks are generally much altered, and as the product of alteration, we find secretions of Chalcedony, Calcite, and Zeolite among the fissures. This is especially the case among the more basic glassy varieties, the characteristics of which will be given briefly in this place. They are also much decomposed at the surface into a peculiar reddish soil resembling Latérite. In the first account relating to the geoloo'v of the island, found in the narrative of the famous American Expedition under Commodore Perry,* it is stated (Vol. I, p. 202) ''The geological formation of the island is trappean, with its various configurations and mineralogical peculiarities ; columnar basalt appears, and hornblende and chalcedony are found." The 'hornblende' mentioned in the above citation most probably refers to the crystallized Rhombic Pyroxene here to be described. This mineral therefore seems to have attracted the atten- tion of early observers. Also in some Japanese book, treating of the products of the island, we find its crystal-form figured and mentioned as Olivine, and stated as also occurring in Kitajima, a small island to the north of Ototojima. In the summer of 1884, Mr. T. Suzuki of the Geological Survey visited Chichijima, and made a collection of rocks, an account * Narrative of the Expedition of an American Squadron to the China Seas and Japan performed in the years 1852, 1853, and 1854, under the command of Commodore Perry, com- piled by F, L. Hawks. 3 vol. 1 Atlas. Washington 1856. OX PYROXENIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 6^ of which was given in the Bulletin* of the Geological Society of Japan. His collection besides contains specimens from the neighbouring island of Ototojima (Stapleton Island). A part of his collection he kindly placed at my disposal to supplement my own. The kinds of rocks from Chichijima and Ototojima, as enumerated by Mr. Suzuki are as follows : — 1) Augite-andesite, 2) Augite-andesite glass, 3) Quartz Augite-andesite, and 4) Basalt. It may be noticed however, that the ' Augite-andesite glass ' represents basic glasses, of the nature of Basalt-glass like Tachylyte. The absence of Olivine in these glasses is quite peculiar and noteworthy, its place being taken by the Rhombic Pyroxene and Augite. In Chichijima, where I had the best opportunity for observing general geological features, volcanic rocks occur either in a mas- sive state probably consolidated from lava-flows, or as agglom- erates of volcanic pebbles, associated or almost cemented with tuffs. It is usual to find among the former class, a lighter coloured, porphyritic rock, the ground-mass of which is microcrystalline, usu- ally with small quantities of glass ; the essential mineral constituents being Plagioclase, Augite, Rhombic Pyroxene and Magnetite. In fact, it resembles the Augite-andesite which is found in Honshiu. It seems to be especially well developed at Hahajima. Among the pebbles that compose the agglomerates however, we find various remarkable modific- ations. The peculiarity of these rocks is their basic character, with a dark coloured glass basis, in which the Pyroxenes are usually developed as Rhombic and Monocliuic forms. Their mode of occurrence suo-- gests that they were formed as the accumulations of volcanic ejections, which were hurled away from a fluid lava, rapidly cooling on its way, and thus inducing an incipient state of crystallization. In some varieties however, the structure becomes porphyritic ; the crystals of Pyroxenes, * Part. B. vol. I (Japanese). 70 Y. KIKUOHI and a comparatively few crystals of Plagioclase, being developed within a partly rnicrocrystalline groundmass. The latter consists of a brown coloured basis, containing microcrystals of Plagioclase, filiform Au- gite, and Magnetite. These darker rocks are probably the rocks des- cribed above as Basalt. But they differ from the usual type of Basalt in the absence of Olivine. h\ the more glassy varieties, Plagioclase is absent except as ' Rhombic lamellae.' These rucks are often dark pitch- coloured glass, with numerous round vesicular cavities filled with Zeolites. Under the microscope, they are always found to be of light greyish or brownish coloured glass, in which a light green coloured Augile is developed in slender filiform crystals in innumerable thick clusters. They are often found alternating with sheets of a peculiar greenish rock, resulting from the infiltration of Chalcedony and a green fibrous chloritic mineral through its mass, which thus has the appear- ance of a Jasper, and exhibiting microscopically a diabasic structure. This was especially well observed near Omura in Chichijima. The glassy rock mentioned above often shows a most interest- ing modification, in which the basis becomes more glassy and contains very sharply defined crystals of Rhombic Pyroxene. In these varie- ties, the glass often exhibits pearlitic structure under the microscope, while the specimens from Ototojima are in spheroidal masses which have the tendency to separate in concentric layers like an onion. The microscopic porphyritic components are Rhombic Pyroxene, and sometimes monoclinic Augite of a light green colour, together with a microlithic form of the same mineral, Plagioclase and some Picotite. Olivine has not been hitherto recognised. Very characteristic microscopic objects in these glasses are the 'Rhombic lamella?' already referred to. They have been found in basic glasses of other localities. A short account of these lamella?, may be given here. They are found making gradual transition to the porphyritic crystals ON PYROXEXIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 71 of Plao-ioclase. Thus there seems to be much probability that they are Plao-ioclase, as first sugro-ested by Penck.* Kreatz f considers them as Anorthite. Doss \ also observed the gradual transition to typical Plaçnoclase in the Palagonitic tuffs of Haurân. The 'Rhombic lamelke7 are especially well developed in the pearlitie glass of Susaki in Chichijima, and in the spheroidal glass of Ototojima. In the latter, the edges of these plates are sometimes so thickly covered with very delicate dark filaments of Augite, as to conceal the crys- tal almost entirely. They are extremely thin, and there is usually no action upon polarised light. Very sharply defined faces may ofren be recognised on the sides of the rhomb. The angles of the rhomb measures 52° and 128°, and may be considered to be formed of the faces P (001) and x (10Ï), to which the face //(20Ï)is sometimes added (fig. 22). In more perfect crystals, it has been observed that the (jne side of the rhomb is formed of two faces in place of the face #(101); in such cases the crystal would probably consist of faces P (001), ilf(010), y (201), o(llï), p (111); the last two faces being in the same zone with x. Fine fissures parallel to P. probably the cleavage- fissures, are especially well developed in the specimen from Susaki. The lamellae become broader in some cases, and the cross section exhibits twin-lamellae of the Albite-type. The twin of the Carls- bad-type has also been observed, the most typical of which is copied in fig. 21. ; the zone [P : M] running practically parallel to the zone [M : x~\, so that a rhombic modification sometimes results. This characteristic is peculiarly analogous to the rhombic modification of Anorthite-twin, e.g. that of Miyakejima described by the author. § * Studien üb^r lockere vulkanische Auswürfllinge. — Zeitschft. d.d. geol. Gesellft. 1878 p. 99. fUeber Vesuvlaven von 1SSI und 1883.— Tscherniak's Min. u. Petro. Mitth. 1884. p. 139. X Die basaltischen Laven und Tuffen der Provinz Haurân und vom I)î ret et-Tulûl in Sirien— Tscherniak's Min. u. Petro. Mitth. Bd. 7. 1886 p. 5^7 § On Anorthite from Miyakejima. This Journal, vol. II, p. 31. 72 Y. KIKUCHI The direction of extinction can be measured in thicker plate«, in which there is a distinct effect upon the polarised ray. On the face M, the direction of extinction makes -40° with the edge P : M. Thus there seems to be much probability that the ' Rhombic lamelke' are Anorthite. All of these glassy rocks contain more or less well-defined crystals of Rhombic Pyroxene, but it is in the more glassy modifications found in patches among these rocks that the well-detined crystals of this mineral are developed at the expense of the light-green coloured filiform Augite. Such modifications have been found so far on the coasts of Tatsumiura and Miyanoura in Chichijima. The specimen obtained from Tatsumiura presents a dirty brownish green colour, resulting from the alteration of the dark glass- into a product resemb- ling ' Palagonite,' in which green translucent crystals of Rhombic Pyroxene may be picked out. The specimen from Miyanoura also presents a similar character, but the degree of alteration is much less, the vesicular cavities being chiefly lined with Zeolites.* Microscop- ically examined, the glass transmits ;i faint greyish green colour. The glass would correspond to the ' Sideromelane ' in Tachylyte. The fissures and cavities are filled up with a palagonitic matter, which transmit a sulphur-yellow colour in thin sections. The palagonitic alteration- product above mentioned is of a dirty green colour, and is made up of fibrous microcrystalline aggregates, showing a weak double refraction. When it is found filling up round pores, the periphery is often radially fibrous, so that an aersrreo'ate polarization exhibiting a dark cross is observed between crossed Niçois. Before the blowpipe, the glassy portion swells up to a great bulk and easily fuses, due to the large amount of water it * One of these Zeolites is Stilbite, in combination of the faces, oP (001), a>Pdb (0101, oo P 5c (100), P* (101), 2P (221) ; another is Cltabasite, with characteristic striatums on. the rhombohedral face. There are also some fibrous crystals of an uncertain Zeolite. ON PYROXENIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 73 contains : but the altered portion almost resists fusion. It is found also that the latter is softer, and the specific gravity is less, than the 2"lass. The glass and the alteration-product are scarcely attacked by hydrochloric acid. The mineralogical composition of the rocks of Tatsumiura and of Miyanoura thus far indicated, would seem to be quite singular. It is probably to be classed as special rank among the glassy form of basaltic rocks. I am indebted to Mr. T. Suzuki, for the chemical analyses given below, of the glassy rocks of Ototojima and Miyano- ura. The analyses were undertaken by Mr. R. Fukuda of the Geological Survey. I II Si02 53-18% 54-44% Ah 03 16-18 12-90 Fe20, 10-30 7-08 Ca 0 10-12 5-12 MgO 6-72 12-75 K2 0 -35 -35 Na2 O 1-85 2-06 loss by ignition .... 1'65 5*54 100-35 700-24 I. is the dark spheroidal glass of Sp. G. 2*725, already mentioned, found at Kurose in Ototojima. II. is the glassy portion of the rock from Miyanoura, comparatively free from the alteration-product. Sp. G. = 2*75. The higher amount of lime in analysis I. probably points to the existence of basic Plagioclase, viz. in the form of the ' Rhombic lamella*.' The high percentage of magnesia in II. is accounted for by the rich de- velopment of the Rhombic Pyroxene, the analysis of which is given in the sequel, while Augite is very scarce. The existence of a large quan- tity of water is also indicated by the large quantity of loss by ignition. 74 T. KIKUCHI It has been pointed out by Judd and iCole * that the Basalt- glasses are of rather rare occurrence in Europe, and that when they occur, they are always found as selvage (Saalband) of dykes, or as small fragments thrown out of volcanic vents and cooled rapidly while passing through the air, whilst in some Pacific Islands (notably in the Sandwich Islands) Basalt-glasses are found as lava-streams. It has also been observed that the glasses from these islands are more transparent in thin sections than those found in Europe. Although it is not likely that the Basalt-glasses in the Bonin Islands have flowed as lava-streams, yet they seem to be nearer to those of the Sandwich Isl- ands, &c, since they are always lightly coloured, transmitting a greenish grey or light brownish colour. Fragments of similar rocks probably derived from lapilli, have also been dreged from the bottom of the Pacific by the Challenger. f Another peculiarity of the glasses of Miyanoura and Tatsumiura. is the almost total absence of Magnetite, which seems not to have separated from the magma. Olivine is also entirelv wanting ; the recognisable mineral constituents being Rhombic Pyroxene and Augite, with some Picotite, which is however more generally found as enclos- ures within the Rhombic Pyroxene. We shall now proceed to give the characteristics and the relation of the two Pyroxenes which are developed in these glassy rocks. Rhombic Pyroxene. Owing to the incoherent character of the glassy rocks, in conse- quence of alteration into a palagonitic substance, the crystals of the Rhombic Pyroxene which are unaffected by alteration, can be rather * Ou the Basalt-glass (Tachylyte) of the Western Isles of Scotland — Quart. Jour. Geol. Soc. vol. XXXIX, 1883. p. 457. f Report ou the Scientific results of the Exploring- Voyage of H. M. S. Challenger — Narra- ti-e of the Cruise— Vol. I, pt 2., 1885 — p. 813. This occurrence is the more interesting as it seems to contain IJhonibie Pvroxene. ON PYROXEXIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 75 easily obtained in a fresh state from the matrix. These crystals have a stout prismatic habitus, "5 — 1 cm. in the direction of the vertical axis, usually flattened along the macropinacoidal face. They are somewhat brittle. The larger crystals are dark green in colour, the smaller ones transmitting a pistachio-green colour and in still smaller crystals a light olive-green colour. The observed faces,* when referred to the axial system of vom Rathf are as follows : a = ocPco (100) b = ooP£ (010) m = ooP (110) e = Pt (212) i = 2P? (211) The predominating faces are a, b, and e ; the prism m truncating the edge of a : b as narrow bands. The face i is usually small (fig. 1), Thus the crystal is rather simple, being a stout prism with a very flat Pyramid e, and tabular along the Macropinacoid a. In this respect it re- sembles the crystal of Hypersthene (Amblystegite) from Laach volcanic bombs, described by v. Path. (1. c). The faces of the Pyramids e, i, are very dull and sometimes depressed a little, while the pinacoids and the prism-faces are brighter, but usually owing to surface irregularities, they do not give a very satisfactory reflection. The following goniometric measurements were performed with a small reflecting goniometer (Füss' Wollaston model). It may here suffice to show an agreement with the corresponding angles of Hypersthene given by v. Rath (1. c. p. 530):— * In the position adopted by Tscheruiak (e. g. Lehrbuch der Mineralogie, 2te. Aufl. p. 436) the crystal is so placed that the acute prism-angle is turned forward, in oi'der to bring it into analogy with monoclinic Pyroxene. In this position, the signatures and the symbols of the faces here mentioned are respectively, b (010), a (100), m (100), e (122), i (121). Fig. 9, repre- sents a crystal formed of the faces a, b, m, and i figured in this position. f Mineralogische Mittheiluugen — Ueber ein neues Mineral aus Laach. — Poggendorf. Ann. d. Physk. u. Chem. Bd. CXXXVIII, 1869. p. 529. 76 Y. KIKUCHI v. Rath (Amblystegite) e : e (212 : 2Ï2) = 153° 11' 152° 22 %' e : e (212 : 212) = 12.1° 12' 121° 81/,' i : in (211 : 110) = 138° 35' 138° 55' a: m (100 : 110)=- 136° 5' 135° 50' From the last value the prism-angle of the Pyroxene would become 92° 10'. The characteristic cleavage of the prism is distinctly developed, the cleaved face exhibiting a pearly lustre. Cleavage parallel to Brachypinacoid is scarcely observable, but it can sometimes be recog- nised as distinct fissures in microscopic slides cut at right angles to the Vertical axis. The crystal has the tendency to separate along the Macropinacoid a, the plane of separation presenting an uneven surface. This is however not due to any interposition, as is usual in the massive variety of this mineral. The phenomenon of pleochroism is very marked. The axial colours examined along the three pinacoidal faces gave with a section of nearly *5 mm. ; a = reddish-brown, ß = greenish-yellow, y = yellowish-green, with a weak absorption a > ß >> y. These characters agree with the observations of Hatch* on the Rhombic Pyroxene of Charchani. In still thinner sections the pleochroism becomes very faint. The section parallel to the Base, when examined under a conver- gent polarised ray, shows the emergence of the positive bisectrix y the plane of the optic axes lying parallel to Brachypinacoid b, while in the section parallel to the Macropinacoid a, the negative bisectrix a emerges perpendicularly. The characteristics given above chiefly refer to the larger crystals which are found scattered within the glassy rocks, but the microscopic * Vulkangruppe von Arequipa— Tscherrnak's Min. u. Petro. Mitth. Bel. 7. 1887. p. 339. ON PYROXENIC COMPONENTS IN CERTATN VOLCANIC ROCKS. 77 examination has revealed the fact that innumerable crystals of the Rhombic Pyroxene are also developed in the glass in smaller but sharply defined forms which, though on the whole closely allied to those already described, exhibit certain differences in habitus. They are more slender in form, being elongated along the vertical axis. The pleochroism is very feeble ; in still smaller crystals being hardly noticeable, transmitting only a very faint greenish colour. Most of the crystals are formed of the faces a = coPôc (100), b = ooPao (010), m = coP (110), and i = 2Pt (211) as repre- sented in fig, 2. The flatter pyramid e = P2 (212) appears rather seldom so that the crystal appears sharper at the terminations. Sometimes a pseudo-monoclinic form results by the shifting of the pyramidal faces i as in fig. 7. The plane-angle formed by the edges of the pyramid i over the vertical axis, when measured along the brachypinacoidal face b, is 80 1/°, while the corresponding angle when measured along the macropinacoidal face a, is always near 120 v2°* These crystals are found in all sizes usually '05 — 1 mm. in length, gradually passing to the larger forms. Twin. — Numerous stellate aggregates of crystals are often observed under the microscope. This is probably due to the union of the crystals according to several twin-laws. One of these is a cross-shaped penetration-twin (fig. 6), in which the two individuals have the faces of the Brachypinacoid parallel to or coinciding with each other, and the directions of the vertical axes make an angle of 601/2°. This twin is therefore analogous to that described by F. Becke,f the twinning-plane being Macrodome P 5c (101). * Rosenbusch, gives the two values as 80° 52', and 120° 38' respectively— Mikroskop. Phy- siogr. 2te Aufl. Bd. I, p. 392. f Uebar Zvnllingsverwachsungen gesteinbildender Pyroxene u. Amphibole — Tschermak's Min. u. Petro. Mitth. Bd. 7. 1887. p. 95. 78 Y. KIKUCHI Parallel-intergrowth of the Rhombic Pijroxene with a green Augite. — On examining- the microscopic crystals of the Rhombic Pyroxene under crossed Niçois, it is found, as might have been expected, that they extinguish the light parallel and at right angles to the direction of the vertical axis. But the crystal is sometimes bordered on both sides by very narrow bands, in which the direction of extinction is oblique. This phenomenon is observed only when the crystal is found lying horizontally upon the macropinacoidal face a. The nar- row bands show a simultaneous extinction of 42°-45° with the prismatic direction as represented in fig. 3. It may be observed that in the crystal here figured, the terminal faces are formed by the flat Pyramid e = Py, a case rarely seen in the smaller crystals. The breadths of these bands are not constant ; sometimes they become so narrow as to be almost inappreciable except by a most careful scrutiny. This is a phenomenon arising from the regular intergrowth of the Rhombic Pyroxene with a Monoclinic Pyroxene. We must consider that the Brachypinacoid b= coPoo of the Rhombic Pyroxene is attached to the Orthopinacoid a = ooPôô of the monoclinic Augite, the vertical axes of both crystals being parallel. Thus the macropinacoidal face a = ooPöö of the former would become parallel to or coincide with the clinopinacoidal face b = ccPSc of the latter, which would ex- tinguish the light obliquely. In this position, the characteristic prismatic angles of the two Pyroxenes would come to similar position. Such parallel-intergrowth of the Rhombic and Monoclinic Pvroxenes was first observed by Trippke* in a variety of Enstatite. It has also been recently observed in the Pyroxenes in some younger effusive rocks. Frederick H. Hatchf describes it as occurring in * Ueber denEnstatit aus den Olivenknollen des Gröditzberges — Xeues Jahrb. f. Min. Geol. u. Pal. 1878. p. 673. 1 1. c p. 327. OX PYROXENIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 79 Andésite from Arequipa in Peru ; and Judd,* as occurring in the lavas of Krakatoa. The narrow band of Augite just described may be also recognised by the fact that it is traversed by fine fissures running parallel to the direction of the vertical axis, and at right angles to it. Under polaris- ed light Augite shows a decidedly stronger interference-colour than the Rhombic Pyroxene, so that the marginal bands are in vivid con- trast with the median portion. With regard to the microscopic characters of these crystals, the longitudinal section shows traces of prismatic cleavage-fissures running parallel to the longer sides. The basal section is quadrate, formed by the traces of the faces, ooP^b , ocPàb , and c©P at the corners, the cleavage-traces parallel to the last face being very distinct. The cleavage parallel to ooPco is also occasionally observed as fine fissures. The basal section sometimes shows a peculiar zonal structure resulting from isomorphous layers. The central portion has, as shown in figure 12, a nucleus in its centre and is divided into two parts. It is distinctly pleochroic, being brownish Ij1 a axis, and greenish-brown H1 ß axis. The outer zone is of a faint brownish colour, and pleochroism is almost inappreciable. The central portion, which is evidently more ferrifer- ous than the outer, must have crystallized before the less ferriferous portion. The crystals in general do not show any sign of alteration whatever. Enclosures. — The most frequent enclosures are gas-pores, and glass- enclosures with fixed bubbles. The symmetrical arrangement of these enclosures is very characteristic. Figures 4, 5, 7, 9 illustrate the typical forms of this arrangement. The most frequent form is that shown in fig. 7, 9, the enclosures being arranged at both ends of the crystal conforming to the faces of the pyramid i = 2?i, i. e., * The lavas of Krakatoa,— Geo]. Mag., Dec. Ill, vol. V, 1888. p. 1. 80 Y. KIKUCHI four at each end. Sometimes they are more numerous at one end, being arranged one upon the other (fig. 5) or, rarely, in rows diverg- ing from the centre (fig. 4). These phenomena indicate the inclu- sion of extraneous matter during the successive growths of crys- tals. A similar kind of symmetrical arrangement of enclosures has been described by E. Cohen* in the Olivine entering into the composition of the lavas of Sandwich Island. Sometimes we find very numerous glass enclosures of an elong- ated form running parallel to the direction of the vertical axis, such crystals presenting a hazy appearance in microscopic sections. They are shown magnified on the left hand side of fig. 4. Very frequent mineral enclosures are certain small octahedra which are scattered within the crystal with no definite arrangement. They are often in very sharply defined forms, sometimes with their faces depressed in steps, and sometimes in twins, the twinning-plane being an octahedral face. The smaller crystals transmit a dark brownish colour, and are not affected by acids. They are most pro- bably Picotite, an inference also supported by the presence of chromic oxide in the analysis given below. The sample for the chemical analysis was obtained from specimens from Tatsumiura by separating with Thoulet's solution. The micros- copic examination of the separated powder showed that there was much glass attached to the crystals. To remove this glass, the sample was lightly treated with hydrofluoric acid. A light green coloured residue was obtained, and the microscope showed that the glass was completely removed, and the smaller crystals came out in a ver)7 sharply defined form. The crystals were by this treatment scarcely attacked, but when the experiment was tried for a longer time, some of the crystals showed * Ueber Laven von Hawaii und einigen andern Inseln des Grossen Oceans, nebsfc einigen Bemerkungen über glasige Gesteine im Allgemeinen. — Neues Jahrb. f. Min. Geol. u. Pal. 1880 IT. p. 32 ON PYROXENIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 81 signs that they were beginning to be attacked. Peculiar transversal canal-like fissures were seen to traverse the crystal from both sides, on the pinaeoidal faces, at right angles to the prismatic direction, as shown on the lower half of the crystal in fig. 5, probably corres- ponding to the line of chemical activity. The minute enclosures of Picotite could not be, in any way, removed. The sample thus obtained, was analyzed by Mr. T. Shimizu, to whom I here express my sincere thanks. The result of the analysis is as follows : — Si 02 55-04% A1203 -88 Cr203 -49 Fe 0 9-40 MnO -18 CaO 1-55 MgO 32-65 H20 -45 100-64 Rejecting the small quantities of alumina, lime &c. in the above result, and calculating the rest up to 100, we have : — Ox. ratio. Si 02 56-69% 30-234 14-05 FeO 9-68 2-151 1 MgO 33-63 13-452 6-25 100-00 From this we see that the composition m:iy be represented by FeO. 6 MgO. 7Si02, which is equivalent to Fe Si 03 + 6 MgSi03. This when calculated in percentage gives as follows : — Si 02 57-38% FeO 9-83 MgO 32-79 100-00 82 Y. KIKUCHI Thus the Rhombic Pyroxene under consideration may be called Bronzile. The Rhombic Pyroxenes which have been recognised in Andésite and other effusive rocks, are so far usually of the more ferriferous kind known as Hypersthene. A well crystallized variety corresponding to the composition of Bronzite has hitherto been un- known to the writer. It ought to be remarked however, that the the ratio of the two silicates Fe Si 03 and Mg Si 03 seems to vary to a certain extent in the different varieties of rocks in which they occur. Thus it is found that the specimens found in the dark porphyritic rocks are probably more ferriferous, as shown by their stronger pleochroism. It has also been indicated on page 79 that in the rock of Miyanoura, the crystal sometimes occurs built up of isomorphous layers, (fig. 12) also indicating this variation in chemical composition. The mean specific gravity was found to be 3* 3 05 With regard to hardness, it was found that the sharp edge of the Pyroxene can scratch the Adularia Felspar with ease ; thus we may consider H = 6. The pyramidal face e, which is across the direc tions of the well-defined cleavage, is decidedly harder than faces lying in the prismatic zone ; the latter can be scratched with difficulty with a sharp edge of the Felspar, but we have never succeeded in scratching the former. Thus on this face H — 6*5. Thin pieces subjected to the tip of blow -pipe flame fuse at the edge, into a dark globule. A Green Augite. The occurrence of a light greyish green coloured Augite in paraliel-intergrowth with the Rhombic Pyroxene has already been considered. In some specimens of the glassy rock, we find the former more abundantly than the latter. At first sight, these two minerals when found together seem to be almost indistinguishable from each OX PYROXENE COMPONENTS TN" CERTAIN VOLCANIC ROCKS. SM other. On closer examination however, it is found that the Auo-ite shows a stronger interference-colour. Also n certain section shows an oblique angle of extinction making 40° with the direction of the vertical axis, such being a clinopinacoidal section generally formed of the traces of the faces ocPcô , -P, o P. The section at right angles to the vertical axis shows a trace of cleavage parallel to ooP. Sometimes it is fourni that this section formed of ooPöö , ce Poo, and zc\\ is separated into two halves by a trace of a twin-boundary parallel to ccPdc . As we cannot in o-eneral have the section filling exactlv at riirht angles to the vertical axis, such a section generally shows a difference in the angular values of extinction, with respect to the line of boundary. This is in fact a common form of the Augite-twin, with ooPco as a twinning face, which cannot of course be expected in the Rhombic system. In the ball-like masses of the glassy rock obtained from Ototojima this kind of Augite is specially well observed in association with t he Rhombic Pyroxene, with Plagioclase both as well-defined crystals and as 'Pthombic lamellae,' and with some Pieotite. More usually, however, the Augite is developed in the glassy rocks of Chichijima as very slender crystals, sometimes almost tilling up the entire glass-mass in confused aggregates so as to make the glass resinous in lustre. The development of these crystals, as indicated on pag. 70. it always rudimentary, having the tendency to assume fern-like shapes somewhat simulating the Pitch- si one of Arran in Scotland. The crystal always appears in elong- ated form along the vertical axis, the angle of extinction in clinopinacoidal section measuring nearly 40°. The cross section is always six sided formed by the prism-faces, and a pinacoid ; the former presenting a peculiarly concave outline* (fig. 13c). The *On account of the small size, and the curved character of the faces, the exact orientation of this section cannot be determined. 84 Y. KIKTJCHI terminations of the crystal frequently end in fine filaments, which are often curved and branching into filiform threads (fig. 13a) shown magnified in fig. 14. Numerous longitudinal fissures as well as transversal clefts are also characteristically developed. The crystals are often found in irregular stellate aggregates, which are probably twin-groups. In some cases, the directions of the vertical axes differ by 60°. Sometimes a narrow median band is inserted, in which the extinction is straight (fig. 13b), showing that the Rhombic Pyroxene is in parallel-intergrowth with the Augite in the manner already described. These rudimentary forms of the Augite-crystals are very characteristic in the glassy rocks of Bonin Island, especially of the rocks found near Omura and Kiyose in Chichijima. In the glasses of Miyanoura, and of Tatsumiura already mentioned as contain- ing the crystals of the Rhombic Pyroxene, the monoclinic Pyroxene appears also in skeleton-forms, which, as may be seen from the accounts given below, must have had a close relation to the crystals of the Rhom- bic Pyroxene during their growth from the magma. A series of these micro-crystals are illustrated in fig. 15 — 20.* The most, simple form is a rod-like body bifurcated at both ends, or sometimes X shaped with numerous delicate spiny proeesses proceeding from the branches (i\g. 20). These are very delicate bodies measuring nearly *025 mm. in length. The fine processes usually become more elongated in larger forms, which measure '07 — '00 mm. in length, and arc so arranged that one set attached to one branch is always parallel to the other branches (fig. 15 — 18), these two sets being disposed nearly at right angle to * These skeleton-crystals are strikingly analogous to those described by H. Vogelsang from a certain slag. (Die Krystalliten. p. 40. Tar. VI). The form assumed by these microlites also suggests an analogy with the so called " hour-glass shaped" structure found in some crystals of Augite. This structure is probably due to the unequal accretion of crystal-molecules in different directions as indicated by these rudimentary or so to say retarded crystals, combined with a certain chemical change during the growth of the crystal. ON PYROXENIC COMPONENTS IN CERTAIN VOLCANIC ROCKS. 85 each other. These processes are not only peculiarly curved but some- what rolled up along a certain axis, so that the entire shape often does not come out at once at one focal distance. However these bodies normally have their branches and processes extended along one plane. Their form usually assumes a rhomboidal outline as represented in fig. 15, 17, 18, or rarely disymmetric as in fig. 16. The form represented in fig. 18 reminds us of the clinopinacoidal section of Augite. Indeed it has been found that some of these well formed microlites extinguishes the light obliquely, the direction of extinction making1 40° with the longer sides of the rhomboidal figure. It is to be noted that the smaller forms hardly exhibits an interference-colour. A very interesting phenomenon which has been observed with regard to these skeleton-crystals is, that they are very frequently at- tached to the crystals of the Rhombic Pyroxene in a very regular man- ner. It is found that the crystals of the Rhombic Pyroxene, when looked at upon the macropinacoidal face a = oo P 56 , have the angles and edges of the terminal pyramidal faces (i or e) set with several curved spines arranged in a very symmetrical maimer. Fig. 3, 7 — 9, show such crystals. It will be found that at the solid angles of the pyramidal faces these spines are much larger than elsewhere and sometimes branching. They are always disposed in such a way that they all lie parallel to the macropinacoidal face a. Thus when these crystals are looked at upon the brachypinacoidal face b = cc P cc , these spiny processes appear all running in one direction as in fig. 6, 9. 11, i. e.. parallel to the face a. There is no doubt as to the identity of these spines with skeleton-crys- tals of Augite alreadv described, so that the regular attachment of the spines to the crystals of the Rhombic Pyroxene can be explained by the tendency which the Augite has to form parallel-intergrowth with the latter. The narrow bands of Augite attached to the crystale of the 86 Y. KIKUCHI Rhombic Pyroxene (described on pag. 7 }>k(>\ at// a , Jurassic Flor« . Jour. Sc Co/I. Vol. 111. PI I. i - .fv ! -f^\ . . , . ■ t 1 b^r IK£i 9a .hi i Irr- in lajiicZe/n rgt. sp. 4, 5. — Macrotœniopteris Riehthofeni Schenk 6abc, 6e. — Podozamites Reinii Geyl. Od. — Onychiopsis elongata Geyl. sj,. 7. — Ginkgodium Nathorsti m. 8.— Vallisneriites jurassîcus Hr. (?). Ybkoyamcc, Jurassic TTvra Jour. Sc. Co//. Vol.111. PI. Ill Auxtor I" la.pi.dt m < it / PLATE IV. Plate IV. Shimamura. Fig. la. — Podozamites lanceolatus L. et H. car. Eichwaldi Hr. lb. — Podozamites Reinii Geyl. le. — Podozamites lanceolatus L. et IL car. latifolia Hr. 2. — Podozamites lanceolatus L. et H. car. genuina Hr. 3a. — Podozamites lanceolatus L. et H. car. intermedia Hr. 3b. — Podozamites Reinii Geyl. 4a. — Podozamites lanceolatus L. et H. car. intermedia Hr. 4b, e. — Podozamites lanceolatus L. et H. car. Eichwaldi Hr ïoJco i rcancL , Jurassic / '/era . Jour. Sc Co//. Vol l/l. PI IV. • Allctor l" Zauoiolem del . PLATE V. Plate V. Shimamura. Fig. 1. — -Podozamites lanceolatus L. et H. var. latifolia Hr. 2a (?) ^>. — Podozainites lanceolatus L. et H. var. Eichwaldi Hr 3. — Podozamites lanceolatus L. et H. c r. intermedia Hr. i, Öa h c. — Podozamites» lanceolatus L. et H. var. Eichwaldi Hr öd. — Podozamites lanceolatus L. te H. var. 6. — Podozamites lanceolatus L. et H. var. Eichwaldi Hr. 7. — Podozamites lanceolatus L. et H. var. intermedia Hr. 8. — Podozamites lanceolatus L. et H. var. minor Hr. 0. — Podozamites lanceolatus L. et H. var. intermedia Hr. ydkoyajnœ, Jarassic Flora Jour. Sc. Co//. Vol. ///. PI. V JK,.\f, Jnr/ci- ùl l 'ti/nr/r-/// ciel PLATE VI. Plate VI. Shimamura. Fig. 1. — Podozamites lanceolatus h. et IL var. latifolia Jlr 2. — Podozamites Reinii Geyl. 3a. — Taxites */>. 3b de, 4-7, Sa h ce. — Podozamites Reinii Geyl. Sd.— Xilssonia nipponensis m» Yokovcu/t« , Jiartss/c Flora Jour. Se. Co//. Vol. HI PI. VI. JUf Aizttcr- in îapùZan del. PLATE VIL Plate VII. Shimamura. Fig. la b c, le. — Dioonites Kotœi m. là. — Anomozamites sp. 2-7, 8a. — Nilssonia nipponensis m. 8b. — Podozamites hmceolatus L. et H. var. genuina Hr. 9. — Podozamites sp. 10. — Dictyozâmites ffrossinervis m. ^lokorV(trtia, Jurassic flora Jour. Sc. Coll. Vol. III. PI VII .hir/(,i- /ri /(t/iic/cm '!rl PLATE VIII. Plate VIII. Shimamura. Gink^odium Nathorsti m. Yokohama , Jurassic Floin . Jour. Sc. Coll. Vol. ///. P! VIII. i PLATE IX. Plate IX. Shimamura. Fig. 1-10, 10a. — Ginkgodium Nathorsti m. IL— Palissya sp. 12a. — Podozaiiiites lieimt Gcyl. 12b. — Pinus Nordehskjoldi Mr. Yckoyama, Jurassic Floret. Jour. Sc. Co//. 14?/. I I LP/. IX. 1 1 i h '■■ i i . /■■< toi in . PLATE X. Plate X. Ozö. Fig. 1. 2a. — Asplenium whitbiense Brgt, sp. 2b. — Nilsgmiia ozorma m. 2c, — Taeniopteris (?) 3, 3a. — Sagenopteris sp. 4-10. 8a. — Dictyozamites indicus Fstm. rar. distansm. 11-14. — Xilssonia ozoana m. 15-19. — Tnxites up. 20—23. — Carpolithes ginkgoides m. YckoyotyncL, Jiwctssic lier*/ Jour. Sc. Co/I. Vol.111. PIX. Avutor m lapidem del . PLATE XL Plate XI. Ushimaru. Fig. 1—3. — Eqnisetnm nshimarense m. 4. — Aspleiiinm distans Hr. i), — Dictyozamites indiens Fstm. rar. distans m. 0. — Podozamites lanceolatns L. et IT. rar. (?). 7. — Thyrsopteris kagensis m. )}'À,'' i -a/ru . Juras&ic Flora Jour. Sc. Co/I. Vol. ///'. Pi .liii/fT ni Zotpidem c/r/ PLATE XII. Plate Xïï. Shimamura. Fig. 2, lab, 2. — Adiantites Heerianus m. 3. — Pinus cfr. prodromus Hr. Okamigô. 4. — Podozamites Reinii Geyl. Qt — Thyr.sopteris Mumiyana Br all win» thus helped us. We also tender our liest thanks to the colleagues and friends who have since assisted us, by suggestions and otherwise, es- pecially to Major-Genera] Palmer. P. E.. for kindly helping us with our English and for many valuable hints. The questions we asked in our letters were mainly as follows : The names and addresses of the observers. The times at which they first noticed the eruption. Did they hear the sound of the explo- sion? Its nature, duration, loudness, etc. Did they see steam (com- THE ERUPTION OF RANDAI-SAX, 93 inonly culled smoke) rising up in the air? Its colour, height, form, etc. Did they see lightning in the steam, or lire on the volcano? How did the lightning or fire look? Did volcanic dust fall? Its thickness, colour, consistency, structure, etc. Were there any earth- quakes either before or after the eruption? Their times of occurrence, intensity, duration, nature, etc. The state of lakes, rivers, and springs before and after the explosion. The meteorological conditions, es- pecially the force and direction of the wind. Any other information bearing on, or which might seem to bear on, the eruption was also asked for. The answers to these question- are given in a tabular form at the end of this paper. Bandai-san, considered topographically and geologically. In Northern Japan, there run along the Pacific seaboard two principal masses of mountains, chiefly composed of crystalline and older rocks. The more northerly of the two. on the eastern side of the Kitakami River, has been named by Dr. E. Naumann the Kita- kami Mountain-land, and the other, situated to the east of the Abu- kuma River, the Abukuma Mountain-land. These two mountain- masses are remarkably similar in their geological structure, and the principal direction of strike is north and south. Thev are very old formations, consisting of granite, granite-gneiss, gneiss, and other crystalline schists, together with thick accumulations of Pakeozoic strata, much folded and faulted, and some patches of Mesozoic strata not less disturbed. We have, in fact, the relics of old land, the principal features of which must have been determined at the end of the Mesozoic Era, and much of which has, no doubt, been sub- sequently denuded away. On their western sides, the two mountain-lands face broad val- leys, in which the rivers already mentioned run in a meridional 94 S. SEKIYA AND Y. KIKUCHI direction — the Kitakaini from north to south, and the Abukuma from south to north — and along which passes the chief highway of Japan. The valleys separate the Abukuma and Kitakami ranges from a high ridge which, traversing the middle of Northern Japan, forms struc- turally the backbone of the region, and constitutes the main water- shed between the Pacific Ocean and the Japan Sea. This central ridge owes its origin to volcanic effusions of comparatively younger date, and consequently the natural barrier thus created is made up mostly of prominent volcanic peaks, and differs in its features from the neighbour- ing old land. The more important of the volcanic peaks are, Osore- yama and Yake-yama on the extreme north of the main island, and thereafter Gaiiju-san, Komaga-take (Province of Rikuchü), Zoô-zan, AzLima-san, Bandai-san, Nasu-dake, Shirane-san (Province of Shimo- tsuke), Akagi-san, &c. These mark the course of the line of weakness along which terrestrial disturbances of varying degree have manifested themselves, in times past, attaining their climax during the Tertiary Era, and thereafter declining into their present state of comparative quiescence. Bandai-san (Lat. 37° 30' N., Long. 140° 6' E.) is situated in the Yama District (kört) in the Province of Iwashiro, immediately adja- cent to the Abukuma mountain region, a part of which, formed of granite and gneiss, borders the east bank of the river Nagase that runs immediately past the foot of the volcano. Besides Bandai-san, there are in this part of the country several other volcanoes, both active and extinct, as shown in PI. XXIA"; the line of principal volcanoes belonging to the central ridge of Northern Japan, already referred to, is shown in the same Plate by a broken line. On the north-east of Bandai-san are Dake-yama and the Azuma-san group, the latter con- sisting of three principal peaks called the Eastern, Western and Mid- dle Azuma: while on the south are Nasu-dake, Takahara-yama, &c. THE ERUPTION OF BANDAI-SAN. 95 Immediately north-west of Bandai-san there is a small lake called Okuni-numa, at an elevation of 1065 m. above the sea-level, and surrounded on all sides by ridges, the highest of which is called Nekoma-yama, and rises 305 m. above the lake. Judging from its features, Okuni-numa is unquestionably an old crater. Between it and Bandai-san there stands a round -topped hill, till lately over- grown with forest, and known as Marumori-yama, which is apparently a small volcanic cone. The few naked tree-trunks, stripped of bran- ches and leaves, that are now to be seen on this hill-top vividly attest the severity of the recent eruption. From the fact that the older rocks of the Abukuma mountain- land underlie the volcanic groups in the vicinity of Bandai, it seems reasonable to infer that the volcanoes originated on the fractured edges of the old formation. Though all of the volcanoes that have been in eruption in recent times are shown as active on Plate XXI V, it is to be understood that such activity never exceeded intensified solfataric explosions, disturb- ing the upper crust alone. The late explosions of Nasu-dake and Azu- ma-san were of this class only, neither lava nor pumice having been ejected. The extreme volcanic energy which once raged in the district of Bandai seems to have gradually waned down to the present time. Denuding action has evidently played a more prominent part than plutonic agency in changing the forms of the mountains, and the decomposition of the rocks has produced a thick layer of soil, sup- porting a dense forest-growth, and concealing the old lava-flows and scoriaceous ejections which attest the volcanic origin of the hill-mass- es. Peasants worked daily among the green forests of Bandai, to collect fuel and to fell trees, wholly unsuspicious of the calamity that hung over them. The district about Bandai-san is made up principally of tufa- 96 S. SEKIYA AND Y. KIKUCHL ceous deposit« and sheets of volcanic rock, forming the basis of an elevated area known as the Aizu Plateau, which includes the districts of Yama, Aizu, Kawanuma and Onuma, in Iwashiro, its average height exceeding 500 in. above the sea-level, This plateau is surrounded on all sides by mountains of volcanic origin. On its southwest border stand the extinct volcanic peaks of Hakase-yama, Mikagura-dake &c, and on the south the conspicuosly Hat-topped Nunobiki-yama, formed of volcanic sheets. Among these mountains are found numerous hotsprings, more than 30 of which have been counted. The streams which rise in the surrounding mountains discharge into a depression on the south side of Bandai, there forming the Lake Inawashiro, which is one of the largest in Japan. This lake, the surface of which is 496 m. above the sea-level, is not a true crater-lake as is sometimes supposed. Its principal feeder was the river Nagase, flowing from the northern part of Bandai. The upper course of this river was. however, entirely stopped by the falling débris during the recent eruption, and the lake is now supplied mainly by its tributary, Sukawa, flowing from Azuma-san. The lake discharges northwest- ward, at the village of Tonokuchi, by a stream which flows through the Aizu Plateau under the name of Xippashi-gawa for about 11) kilometres, then joining the Aka-gawa. The latter stream collects all the waters of the Aizu Plateau, and finally runs into the Japan Sea near the port of Niigata. Recently another outlet was made on the eastern side of the lake, by means of a canal for irrigation. It seems probable that the Inawashiro Lake fills up a depres- sion formed by evisceration of the ground, resulting from the copious outpourings of volcanic products in its vicinity, notably those of Bandai. The origin of the lake, according to current tradition, is ascribed to a great terrestrial disturbance which took place in the THE ERUPTION OF BAXDAÏ-SAX. 97 ninth century. The districts known as Tsukinowa and Sarnshina, consisting of 40 villages, are said to have been submerged on that occasion. The name Bandai-san is usually given to a group of peaks, con- sisting of Obandai. Kobandai (lately destroyed), Kushiga-mine, and Akahani-yama, surrounding an elevated plain called Kumano-taira. (PL XV). This group, standing on the northern side of Lake Ina- washiro, forms a very conspicuous object in the landscape, and dis- plays the characteristic outlines of a volcanic mountain. When seen from the southwest side, from the town of Wakamatsu, it appears as a single pointed peak. It has sometimes been called the "Fuji of Aizu," from its resemblance to the well known Fuji-yama. Obandai, or Great Bandai, is the most prominent of the peaks, its summit being 1840 m. above the sea-level. It presents a highly rugged and precipitous escarp toward the Numano-taira, exposing volcanic strata which are the results of accumulations of augite-an- desitic lava and scoriae during its period of activity. Viewed at a distance from the east, Obandai has a highly characteristic appearance, descending by a very steep slope toward the central plain and by a gentle one in the opposite direction. Kobandai, or Little Bandai, was less known, on account of its being situated far away from the inhabited portions of the Aizu Plateau, and being also partly screened by its more prominent sister peak. From the latter fact it appeared to be lower than Obandai, and was therefore so regarded; and hence its name. But careful examination lia. shown that they were pro- bably of almost identical height, as will appear farther on. It is probable that the plain Numano-taira is the remains of the original crater — Atrio — and that the several peaks above men- tioned are [«arts of the Somma-wall which encircled it. But gradual denudation during long periods of quiescence, together with occa- 9$ S. SEKITA AND Y. KIKUCHI sionnl rendings of the crater-wall when explosions took place, have brought about the present form, namely, that of separate masses presenting more or less conical shapes. In the ISTumano-taira, or "plains with ponds," there were several small lakes or pools, as is usual in craters of this nature. Nearly in its centre, there existed before the eruption a solfatara on a small hillock called Iwö-yama, or "sulphur mount." from which sulphur was collected by the neigh- bouring villagers. The plain was also covered with dense forests, which were destroyed on the loth of July. The flanks of Bandai are cut into numerous channels called "sawa." The largest of them is that known as Biwa-sawa. which opens eastward from the Numano-taira. It was down this ravine that the smaller stream of mud and rook descended in the late eruption. Seen from the east, it presents a very conspicuous appearance. Fig. 2, PI. XYII, is a sketch made of this part of the mountain immedia- tely after the eruption. From our point of view we had a magnificent prospect of Obandai, with its rugged and precipitous wall on the nor- thern side. The plain of Numano-taira is seen to terminate in a very steep cliff, known as Futatsu-iwa, at which place the water of the lakes in the above plain made a sudden leap, forming a high water- fall. Immediately below this is a small depression called Hikage, which has been regarded by some as a secondary crater of the late eruption. Another large ravine is that lying between Obandai and Akahani-yama, opening southward, and named Katsura-sawa. There is also a bare glen on the southern flank of Obandai, known as Kara- sawa. These ravines or valleys may be considered to have been chiefly modelled by the paroxysmal explosions which, as the history of the mountain tells us, took place at intervals in past times. Denudation, however, has doubtless modified their original forms. The same remarks may be held to apply to the topographical features of the THE ERUPTION OF BAXDAI-SAX. 99 ■whole mass. For, not only must the original form of the mountain have Buffered by the sucessive eruptions, but the débris thus produced, obstructing the water-courses, must have gradually brought the sur- face to its present form. Some of the outbursts would seem from the history of Bandai-san to have been very similar in character to that of the loth of July last. The hot springs on the north- western side of Bandai, known as Bandai-no-yu, were latterly the principal remnants of the volcanic forces which once raged with so much vigour. There were three of these springs, all celebrated for their curing effects upon various diseases. They were known as Kami-no-yu, Xaka-no-yu, and Shimo- no-yu, respectively meaning the upper, middle, and lower bath, where small huts had been constructed for the accommodation of bathers, who flocked thither in summer from various parts of the neighbouring district. They were sulphur springs originating in solfatara formed by the issuing of steam and sulphuretted hydrogen from numerous rock-fissures. Several years before the eruption Prof. J. Milne,* of the im- perial University, ascended Bandai-san, when he took a sketch and described it. classifying it as an active volcano. Traditions and History. According to tradition, Bandai-san was originally a single mas- sive peak, the summit of which was burst open in olden times by a volcanic eruption, and split into several peaks, the event being pro- ductive of a terrible catastrophe. The débris of the explosion des- cended on all sides of the mountain, and the two districts Tsukinowa and Sarashina, containing some fifty villages, were engulfed beneath * The Volcanoes of Japan — Transactions of the Seismologieal Society of Japan, Vol. IX. Part II. 100 'S. SEKIYA AND Y. KIKUCHI the area thereafter occupied by Lake Inawashiro. This account of the splitting of one large mountain into several minor peaks is in- teresting, as it agrees with the suppositions suggested Ly the struc- ture of the volcano. For, as has been previously explained, the several ridges together constituting the Bandai group surround an elevated plateau which has all the appearance of an old crater. Another tradition, apparently referring to the same event, says that in the first year of Daidö (806 A. D.) Lake Inawashiro was suddenly formed, and in it a small island, now called Okina-shima, appeared. Religious traditions, on the other hand, not unnaturally connect the catastrophe with demoniacal agencies. A Buddhist temple in a village near Bandai-san contains a document, said to chronicle its founding;, which runs thus: "In olden times there used to dwell about Bandai devils who did much harm to the inhabitants; in the year of Daidö numbers of people living at the foot of the mountain were swallowed beneath the earth, and there was left a great lake (Inawa- shiro). The reigning Emperor, in order to subjugate the evil spirits which were probably the cause of all these terrestrial mysteries, despatched the famous priest Kükai, who, on arrival at the spot, per- formed ten days' secret prayer to Buddha, when the devils were com- pelled to vacate Bandai and to flee to the neighouring mountains. In commemoration whereof, Kükai caused this temple of Dainichi-ji to be built." Several ancient records describe the mountain as having at various times smoked and ejected lire and poisonous vapours. Such accounts are highly interesting, especially when we consider the volcano's apparent quiescence for ten centuries. We give below translations of some of the records published by the Geographical Bureau, with remarks on them. THE ERUPTION OF EAXDAT-SAN. 101 The Koden ("é"^, Old Tradition) says: ''This mountain vomited fire; and sulphur was spread over the country for 10 ri around, its vapour being injurious to health. After Lake Inawashiro was form- ed, the fire ceased and the vapour was dispersed." The TdgoJcu Ryolcddan (^ |U M tT M > Travelling Tales in Eastern Provinces) says : " To the east of Lake Inawashiro there stands a steep peak called Bandai- san ; from its high summit ascend blaz- ing fire and smoke, as if to burn the firmament." This account is followed by an illustrative sketch, as in the accompanying wood-cut, of the volcano belching forth fire and smoke. In the Ou Benranshi (J| $3 Ht H i*>, Handbook on Ou Provinces) the following account is given: " Aizu-yama, commonly called Eandai- san, lies to the east of Lake Inawashiro ; from its top ascends burn- ing1 smoke." It sounds strange to hear of flame and fire : but we ouo'ht not to put much confidence in the tales of travellers, which are too often exaggerated and grossly inaccurate. Though the phenomena caused by streams of molten lava in volcanic eruptions are commonly spoken of as presenting the appearance of flames and fire, we do not find in Bandai-san any indication of lava-flows that can have taken place within historical times. It may be added, however, that there are cases on record in which flames caused by the combustion of gases have been a feature of volcanic outbursts. The Shinpen Aizu Füdoki ($f H # t$- M ± IE, Accounts of Aizu) 102 S. SEKIYA AND Y. KIKTJCHI says : " In olclen times the destruction of a part of Bandai -san gave rise to Akahani-yama. The effects were very violent and extensive; earth and stones falling down dammed the stream of Sutrawa and inundated Ilibara." To the north-west of Oda village there are places respectively called Onamiyose (large wave beach) and Konami- yose (small wave beach). These localities, it appears, were formerly washed by the waters of lakes formed on the occasion above refer- red to, which, however, subsequently disappeared. As to the cause of this destruction of Banekii-san, it is not clear from the description whether it was volcanic or otherwise, but the phenomena exhibited seem to have been similar in character to those of the late eruption. Besides, there are authentic records of other terrestrial disturbances of much younger date, though perhaps of less magnitude, which have occasionally troubled this part of the country. These accounts seem to attest successive outbreaks of the same store of energy that wrought such havoc on the 15th of last July. In the 8th month of the 16th year of Keichö (IG 11), a violent earthquake occurred at Bandai, and the fall of earth and rocks that was produced by this convulsion dammed up the river Nippashi. the outlet of Lake Inawashiro, and resulted in the formation of three new lakes. Water issued in great quantities from fissures opened in the ground. Accumulations of water, caused by the stoppage of streams, formed several other lakes, and in one place a waterfall of considera- ble height. In the villages of Matsuno and Terauchi some temples were overthrown ; and there were innumerable damages of other kinds. In spite of the efforts of Gramö, the ruling Daimyö, who employed large numbers of men to cut an outlet for the accumulated waters, inundation spread over the districts Yama and Kawanuma, producing a lake at Yamazaki; and it was not until after several engineering attempts that a passage was effected, by the aid of which THE ERTTPTTOX OF BANDAI-SAX. 103 about one-half of the inundated area was at length reclaimed. During the period Höreid (1751-1783), another convulsion took place in Mount Hanzawa, and created the present lake of that name. The last recorded disturbance, although its particulars are not known, is said to have taken place about 80 years ago, when several lakes in Numano-taira were filled up, and great quantities of débris descended by the Biwa-sawa. Traditions and tales relating to this event were cherished with superstitious fear by the peasantry of the region, and listened to with wondering awe by the children, until there fell upon them the yet more terrible catastrophe of last Jul}T, which we now proceed to describe. Eruption and Attendant Phenomena. On the morning of July 15th, 1888, the weather in the Bandai district was fine, there being scarcely a cloud ; and a gentle breeze was blowing from the W.N.W. Soon after 7 o'clock, curious rumbling noises were heard, which the people thought to be the sound of distant thunder, often heard among the mountain-tops. At about half-past 7, there occurred a tolerably severe earthquake, which lasted more than 20 seconds. This was followed soon after by a most violent shaking of the ground. At 7.45, while the ground was still heaving, the eruption of Kobandai-san took place. A dense column of steam and dust shot into the air, making a tremendous noise. Ex- plosions followed one after another, in all to the number of 15 or 20, the steam on each occasion except the last being described as having attained a height above the peaks about equivalent to that of Obandai as seen from Inawashiro, that is to say, some 1,280 m, or 4,200 feet. The last explosion, however, is said to have projected its dis- charge almost horizontally, towards the valley on the north. And, con- sidering the topography of the mountain and the form of the crater, 104 S. SEKIYA AND Y. KIKUCHI it is probable that previous discharges were also more or less inclined to the vertical, in a northerly direction. The main eruptions lasted for a minute or more, and were accompanied by thundering sounds which, though rapidly lessening in intensity, continued for nearly two hours. Meanwhile the dust and steam rapidly ascended, and spread into a great cloud like an open umbrella in shape, at a height equal to at least three or four times that of Obandai. This cloud was gradually wafted away by the wind in a southeasterly di- rection. At the immediate foot of the mountain there was a rain of hot scalding ashes, accompanied by pitchy darkness. A little later, darkness was still great, a smart shower of rain fell, lasting for about five minutes. The rain was quite warm. These phenomena, as well as the terror and bewilderment which they caused among the peasantry, were described in thrilling terms by the newspapers of the clay. While darkness as aforesaid still shrouded the region, a mighty avalanche of earth and rock rushed at terrific speed down the mountain slopes, buried the Nagase valley with its villages and people, and de- vastated an area of more than 70 square kilometres, or 27 square miles. Mr. Tsurumaki, a priest of Echigo, who was staying at the ISaka- Account of an noyu spa on the edge of the crater at the time of the eruption, and eye-witness. w]10 escapeJ death almost miraculously, sent us soon afterwards the following interesting and minute account of his terrifying ex- periences: ''I started from my native village on the 8th of July, in company with fair of my friends, for Bandai-san, and arrived there on the 12th, i. e. three days before the catastrophe. I had been there before, in July, 1885, when I stayed three weeks. On the day of my recent arrival (the 8th) the fog was unusually dense, and the volume of steam at Kaminoyu seemed to have lessened. On the 13th the fog was denser still, and remained so till the evening. The 14th was a bright day, the fogs of the previous days having cleared up. From THE ERUPTION OF BANDAI-SAX. 105 about 10 o'clock in the morning of this date the flow of the spring beo-an to diminish. But the fact that the amount of discharge is smaller in tine weather and larger in cloudy days is well-known anion a- bathers, so that we gave no heed to it. The morning of the 15th, which was the fatal day, dawned with a bright and pleasant sky, and the flow of the spring was as usual. At about 8 o'clock, however, there was a fierce convulsion of the ground, and we all rushed out of the house. In about 10 minutes (seconds?), while we were fearfully wondering what was the matter, a terrible explosion suddenly burst out from the slope of Kobandai, about one did * above a place at which steam has been issuing from time unknown. This was followed by a dense mass of black smoke, which ascended into the air and immediately covered the sky. At this time, showers of large and small stones were falling all about us. To these horrors were added thundering sounds, and the tearing of mountains and forests presented a most unearthly sight, which I shall never forget while I live. We fled in all directions, but before we had gone many metres we were all thrown prostrate on the ground. It was pitchy dark; the earth was still heaving beneath us; our mouths, noses, eyes and ears were all stuffed with mud and ashes. We could neither cry out nor move. I hardly knew whether I was dead or in a dream. Presently a stone fell on my hand, and I knew I was wounded. Ima- gining, however, that death was at hand, I prayed to Buddha. Later, I received wounds on my loin, right foot, and back. After the lapse of an hour the stones ceased to rain and the atmosphere had cleared from darkness to a light like moonlight. Thinking this a fine op- portunity to escape, I got up and cried, 'Friends, follow me ! '; but nobody was there. When I had descended about two did, there was a second, and after another did, a third explosion. In these sand * One ehö is nearly equal to 109 metres. 106 S. SEKIYA AND Y. KIKUCHI and ashes were ejected, but no stones. I reached Odera at noon, and there I received surgical treatment, etc." Deluge of rock The most striking feature in the whole of this eruption was the and earth. delude of rock and earth. Notwithstanding the violence of the phenomena, and the completeness with which the mountain was des- troyed, the nature of the eruption was comparatively simple. The destructive agency was merely the sudden expansion of imprisoned steam, unaccompanied by lava flows or pumice ejection. When the explosion took place, a considerable amount of rod; s and earth was projected into the air, and a part diffused in the form of dust, but by flu- the greater part of the bulk of Kobandai was just split into mighty fragments, which were thrown down much after the manner of a land-slip. Descending the mountain sides with ever accelerating velocity, the components of these avalanches were dashed against obstacles in their way and against each other, and were thus rapidly reduced to confused masses of earth and rocks. The loose and friable débris thus produced ultimately lost its adhesive power, and may have been compared with a little exaggeration to sand. If we suppose a mass of some 1.21 cubic kilometres, or 1,587 millions of cubic yards (which was the actual volume of the mountain destroyed), of sand to be suddenly precipitated from a lofty summit, it would flow down the sides in a torrent not very unlike that of water. That the earth and rock débris did flow down in this way we were convinced by examin- ing the actual state of things on the spot, and more particularly by witnessing afterwards with our own eyes a very similar phenomenon, though on a vastly smaller scale. One day, while we were at work in the crater, a huge slice of the precipitous wall of rock that had been bared by the explosion fell suddenly and crashed with a tremendous uproar down the steep in- cline beneath. This slab fell from a place about 300 metres high. The THE ERUPTION" OP BAXDAI-SAX. 107 great masses of earth and rocks were .shattered as they fell, and broken up into pieces, ever growing smaller as they descended. The beha- viour of this pulverized mass resembled the rush of a headlong torrent. Although boulders measuring 10 metres or more in diameter were mixed up with finer matter, as a whole the movement approximated to that of a fluid. No words can describe the fierceness and force of that impetuous downpour — its mad surgings this way and that, and the bold leaps with which it would now and then bound over low ridges that hindered its progress, and shoot onward down the neigh- bouring depression. It was a magnificent but somewhat awful sight to witness during an afternoon's ramble. In a like manner probably, but on a vastly more gigantic scale, the stream of materials on the 15th of July ran down the slopes of ]]andai-san, dividing as it went into two principal branches. The main branch flowed northward. Kobandai, it must be ex- plained, sloped on the north towards the Nagase valley, in an unbroken descent; and, as the mountain burst on this side, the debris dashed with great violence down this northern slope in the direction of Hibara, 9 kilometres away. One part of the torrent actually ran up the valley, toward the source of the River Nagase, burying on its way the three hamlets of Akimoto, Ilosono, and Osuzawa. A part, however, of the pulverized earth ran down the valley; reaching Kawa- kami spa and submerging it to a depth of probably more than 40 metres, it proceeded southward to Hinokuchi, 3 kilometres farther down. The other and much smaller branch took quite a different route, making an angle of nearly 120° with the main stream. It came down by way of Numano-taira, through Biwa-sawa, rapidly spreading as it descended, and dividing into three minor ramifications. The southernmost of these just reached the village of Miné, overwhelming 108 S. SEKIYA AND Y. KIKUCHI nearly one-half of the houses, with their inmates. Fi"'. 1 of PL XVII. is a view taken from the outskirt of Inawashiro, at a distance of nearly one kilometre from the mud field, showing the village of Miné in front, and the avalanche (d) of rock and earth descending upon its prey. Fig. 1 of PL XYIII. is another and nearer view of this mud- field. The combined volume of these two great streams entirely covered an area of 27 square miles, or 70 square kilometres, with a solid sea of mud and rock, beneath which were buried all the features of the landscape, together with people, cattle, and other living things. The grey tint. on PL XV. marks the area thus devastated. Velocity of the The descending matter must have moved with great velocity. descending mate- ., . . . ,,,.... j . by some survivors it was described as having reached their vicinity almost instantly after the eruption. From several calculations, made by comparing the time of the explosion with the times at which the streams of debris arrived at different points, we roughly estimated the average speed to have been 77 kilometres or 48 miles per hour. On its course the mud-stream must have swelled into «Teat waves, as in a surging current. This is attested by eye-witnesses. The Avave- like traces left on the sides of the hill (PL XX) show how the torrent surged upwards when it met any obstacle either obliquely or at right angles. In one case near Kawakami, the earth reached a height of at least 40 metres above the general level on a hill facing1 the direction of flow and at other 40 metres places a spur of the hill which the current struck obliquely caused an uprush of from 30 to 60 metres. 'The general appearance of the present surface is one of extraordinary havoc and confusion, irregular lumps of earth being mixed up with torn-off trunks and branches of trees, fragments of timber, and stray boulders of huge size. In some places the matter has been largely admixed with water, and is treacherous to walk on. THE ERUPTION OF BAXDAI-SAX. 109 In describing the phenomena of the earth and rock debris, the word mud has been frequently used by several observers, who speak of 'mud-stream,' 'mud-field' etc. \Ye also have used the term above, but it must be explained, to avoid misconception, that we have done so for convenience only. Some commentators, indeed, have er- roneously classed the phenomena with those of the 'mud volcanoes' of which we read in geological text-books that, while some have been known to throw up mud to a great height, in others liquid earth only oozes out quietly, and gradually forms an earth-ring round the crater. Such outbursts, however, are no more than moderate manifestations of subterranean energy, and are almost insignificant in comparison with the tremendous forces that destroyed Kobandai- san. Moreover, as far as our prolonged examinations went, there was no evidence of any discharge of mud from beneath. It is true that in the Nagase valley and other places there are now immense quan- tities of mud, but these became mud only after the eruption. During its descent, for example, a part of the débris, mingling with the waters of ponds and lakes in its course, doubtless acquired a muddy cha- racter and was thus assisted in its flow; and, again, that which reach- ed the stream of the ^aii'ase-o'awa became admixed with sufficient water to thin it to the consistency of a paste. But by far the greater volume was comparatively in a dry state, being moistened only by condensing steam, and must have derived its fluid or semi-fluid pro- perties from a rapid process of pulverization after the manner already described. With regard to the secondary mud-stream that ran down to Miné, there has been a diversity of opinion. Some visitors imagined that there must have been more than one crater — that, in fact, the materials which destroyed Miné had a separate origin from the rest; and a cleft or depression at Hikage in Biwasawa, which, as viewed 110 S. SEKITA AND Y. KTRUCHI from below, bore some resemblance to a broken crater-wall, was not unnaturally regarded as a proof of this assumption. But the spot, when examined by us, was found to be wanting as well in the cha- racteristic features of a crater as in any appearance of its having been the origin of a violent volcanic outburst. If there had been such an outburst as to produce the vast quantity of matter that descended towards Mine we must have seen the crater or cavity from which the matter issued, unless indeed it were supposed to have oozed forth gradually after the manner of some of the mud- volcanoes already spoken of — a supposition, however, which is absolutely at variance with the observed facts. The matter which descended toward Miné was really found, upon close examination, to be the loose red loamy soil that had formed the superficial covering of the flanks of Bandai, largely admixed with ash or dust, and boulders. The red colour of the soil was however concealed from view by the coating of grey-coloured "ashes," 10-30 cm. in thickness, that fell especially abundantly down Biwa-sawa, the wind having been directed straight into that ravine during the erup- tion. At Xinnano-taira the accumulation of ashes was especially thick, and from 1 hence it gradually lessened toward the lower part of Biwa-sawa. The mud field of Miné examined some time after was found to have been cut by the action of running water into nu- merous deep chasms often forming perpendicular walls and exposing the red loam underneath. Conical Hills. Among the various phenomena that constantly bewilder the eyes of visitors to the scene of the eruption, not the least striking are the numerous bh>" boulders, some of them measuring from 5 metres to 10 O J ^ metres each way, that are to be seen resting on the surface of the débris far away from the crater. These have evidently been carried along as part of the mud current, and not hurled through the air. Kot less curious THE ERUPTION OP BAXDAI-SAX. Ill are the quantities of small cones, varying from a few metres np to 15 metres in height, which are scattered here and there over the surface, standing out of the débris like so many miniature Fujiyama». Fig. 1, PL XXI. and PI. XXII will give some idea of these objects, as seen respectively from the northern and eastern side of Bandai. When closely examined, they are found to be disintegrated crumbling rocks, so aftected by the agency of heated steam and corroding gases as to have lost their compactness. They are similar in character to the disintegrated rock commonly found near the crater- walls of active volcanoes. During and after their swift descent down the mountain sides, these rock masses have crumbled away, and the debris, falling around their bases, has assumed a conical shape by forming ialuses around them. Fig. 2, PI. XVIII, is a representation of one of the smaller of these hills found near the former site of the Kawakami spa. On reaching the outposts of the mud field, no one could help beino- struck by the singular way in which the advancing" stream of rock and earth seemed to have suddenly stopped, showing a vertical or nearly vertical face, a few metres high. It is apparent that the debris of rock and earth in their swift descent behaved like a fluid, but on nearing to the plain below they gradually lost speed and were ul- timately brought to rest; the materials that followed, on account of their great friction and adhesion could not pass the limit set by their predecessors, and were piled layer on layer, thus forming a steep edge. As is usual in all volcanic outbursts, large quantities of greyish- v0]canic dust blue dust, or so-called ashes, fell during the eruption in the form of showers. Evidently much of this dust was produced by the mechanical trituration, during their flight through the air, of the rocks ejected by the explosions, which rocks, as already explained, had been rendered highly friable by the action of steam and gases. We found, in fact, that the dust was allied in character to the pulverulent matters com- 112 S. SEKIYA AND Y. KIKUCHI posing the conical heaps of débris above described, and that both were derived from the andesitic rock which composed the mass of Bandai-san. Hence it is apparent that this dust or ash is quite different in character from the ashes that are usually ejected from craters in other volcanic eruptions, e. g., that of Krakatoa in 1883. In such cases, the ashes are chiefly derived from molten magma, expelled by steam and mixed with fragments of the pre-existing rocks. They contain, therefore, more or less glassy matter, and are in fact pumiceous. On the morning of the 15th, the wind blew from the W.N.W., so that the dust was carried towards the E.S.E., gradually spreading as it receded from the mountain. On the coast of the Pacific Ocean, which is 100 kilometres or 62 miles from the volcano, the width of the dust fall was 50 kilometres or 31 miles. In shape the dust-strewn area resembled that of partly opened fan, as shown by the dotted space on PL XXIV. On the land it covered a space of about 2,050 square kilometres or 790 square miles. How much farther it spread over the ocean we had no means of ascertaining. At the immediate foot of the mountain, especially toward the S.E., the dense cloud of dust produced pitchy darkness, which, how- ever, did not last long; in the course of an hour the gloom had diminished to about that of the twilight of a rainy evening. But it was nearly 4 o'clock in the afternoon, or 8 hours afcer the eruption, be- fore the dust wholly ceased to fall. The thickness of the deposite was about 0*3 metre on the south and east flanks of Biwasawa. In part the fall was in the form of a sticky, scalding mud-rain, produced by com- n Lino-liner of the dust with condensing steam. It inflicted terrible burns upon people exposed to it, and was the cause of many deaths. The ground also became so hot from this rain and the later dust that people had great diffculty in walking upon it. Every object was covered with a thick grey coating. Foliage, especially, that of the THE ERUPTION OF BAXDAI-SAX. 113 sucji (Cryptomeria japonica), presented a very curious appearance, the leaves, branches, &e, being clothed with a thick, pasty and highly tenacious de- posit" resulting from a mix- ture of dust with conden- sing vapour. A bunch of sugi(a), and a branch of Pine (Piims massoniana),(b), thus covered are shown in the annexed illustration. In Shibutani and its neigfh- bourhoud, which experi- enced the full effect of the dust-fall, not a green thing could afterwards be seen. Houses, paths, trees, fields, in fact all visible ob- jects— wore a greyish hue. At Miharu a town 38 kilo- metres (24 miles) east of the volcano, the dust began to fall at about 9 a.m. and lasted till 2 p.m. During this period the sky had a dim and cloudy appearance, as on a misty day. The ashes hardly covered the ground, but all the leaves of trees and vegetables were tinted erey. At the coast a very slight film only was perceivable on the house-roofs and foliage. Damage to plants and crops by the fall of dust extended as far as 10 kilometres from Bandai sa li. 114 S. SEKIYA AND Y. KIKÜCHI The explosions were accompanied by terrible wind blasts, or coups de veut. In the parts most exposed to the fury of these blasts, houses were levelled to the ground and trees torn up by their roots. Everywhere, however, as might be expected, the fall was in a direction radially away from the forces of explosion, which was also the origin of these destructive and fearful gusts. In PL XV the area swept by the windblasts is shown by arrows, their heads pointing in the direc- tions in which the trees and other objects fell. It was curious to see the manner in which one particular field of growing rice, on the southeast of the volcano, had been thus levelled by the wind. The slender stalks were laid flat upon the ground as evenly and regularly as if they had been combed down in parallel lines. Not a stalk lay across its neighbours. The heads of rice in one furrow covered the roots in the next furrow, and the entire field looked like the warp of some huge loom ready for the weaver's hands. It would appear that the tremendous explosions of steam at quick intervals, lasting for about a minute, produced violent distur- bances of the air, consequent upon the sudden radial expansion of the liberated volumes of steam. When a large piece of ordnance is tired grasses, shrubs and objects in the vicinity are overthrown by the sudden expansion of the gaseous products escaping at the muzzle, which, displacing the air, imparts to it a forward impulse and violent vibratory motion. The eruption of Bandai-san may be aptly com- pared to the tiring of a tremendous gun — such an one, however, as can only be forged by Nature. Places screened by hills and mountain sides escaped. Maru- mori-yama, situated near the mouth of the crater, and fully exposed, received the severest damage. This hill, which was formerly covered with a thick forest, now presents a most melancholy appearance, the few trees left standing being as naked as telegraphic poles. The THE ERUPTION OF BAXDAI-SAX. 115 levelling of houses and shattering of forests are of common occur- rence in great storms. Bat on this occasion the destroying tempests especially near the crater were something more than atmospheric, consisting besides of heated blasts of steam and air, thickly mix- ed with dust and rock-fragments fierce enough to crush the trees and to strip them not only of branches but even of their bark, and wither- ing, scoring, and scorching everything in their course. Some of the most terrible effets of these tornadoes were wrought in the Biwa-sawa and its vicinity. Originating at the old crater, Xumano-taira, this glen, the deepest and widest in Bandai-san, des- cends directly and in an unbroken line to the villages Shibutani and Shirakijö. Notwithstanding the comparatively great distance of these two villages from the crater, the wind-blasts were impelled towards them, down the Biwa-sawa, with prodigious force, and wrought havoc from which places a little out of the direct course of the wind were happily exempt. In the woods on the S.E. slope of Akahani-yama and on the west side of Biwa-sawa, the effects of the storm were es- pecially striking; trees with a diameter of more than a metre had been laid prostrate on the ground in thousands; and a forest was thickly encumbered with fallen trees. Estimating the probable ve- locity of the wind from the effects produced in this localtiy, Mr. Y. Wada, of the Imperial Meteorological Observatory, thinks it can hardly have been less than 40 metres per second, or about 90 miles per hour. Here as everywhere else, the trees fell with their heads point- ing away from the crater, showing clearly that the wind radiated in straight lines from its origin. There is no evidence of whirlwinds or eddy movements of any kind. The cause and effect of the wind in this locality deserve special attention, for which a brief topogra- phical examination is first needed. South-south-west of the former site of Kobandai stands the massive peak of Obandai rising to a 116 S. SEKIYA AND Y. KIKUCHI height of 1,840 metres (6,035 ft.) above the sea, and enst of it, Kushiga- mine, the third peak of the Bandai group, with an altitude of 1,622 metres above the sea. Between them is Numano-taira, the large flat basin of the old crater, with 1,311 metres above the sea-level. At the base of the huge U shaped aperture formed by the slopes of the two mountains the Biwa-sawa begins to descend. Fig. 1 on the wood-cut shows roughly the relative positions of the peaks and the old crater in plan, and Fig. 2 shows a vertical section through A B. Yugeta^x Fig. 1. Öbandai Numano-taira Kushiga-mine. Fig. 2. Section thi^ough A B. Xow, when the steam emancipated by the eruption s iddenly ex- panded and drove before it the surrounding air, a portion of this air THE ERUPTION OF BAJSDAI-SAN. 117 and steam must have been forced through the U shaped aperture aforesaid. Emerging therefrom at a very high velocity, it swept the whole Biwa ravine which preserved more or less the U shape down- ward, and ravaged the eastern half of the inferior peak Akahani. The evidences of this were clearly apparent on the ground itself, the divi- ding lines between the wind-swept area and the comparative calm which prevailed on either hand having been so sharply defined that the belt of levelled forests and ruined houses was separated by a few metres only from the unharmed areas to the right and left of it. Nothing new O CD can be claimed for this phenomnon, which is a well-known charac- teristic of certain storms, especially of whirlwinds having their apex near the earth's surface. But its definition on this occasion was in- teresting and well-marked. Attempts have been made to associate these particular coups de vent with the moderate northwesterly breeze that prevailed at the time. Dut it is obvious that, though this wind determined the direc- tion of the dust-cloud in the higher regions of the atmosphere, it must have been powerless to contribute sensibly to those intense blasts which descended the Biwa-sawa and swept over the region at the base of the mountain. Next let us examine how the wind behaved on the other sides of the volcano. On the east, as before described, Kushiga-mine formed a screening wall which deflected the course of the hurricane; on the west, Yugeta-yama, the part of Kobandai that remained undestroyed, and other ranges of hills arrested the expanding steam and the rush of air, and saved the forests behind. But on the north, there being no obstruction in that quarter, to which moreover the discharges were directed at an inclination to the vertical, the effects were probably tremendous, as was evidenced by the condition of Marumori-yama. Indeed, every spot of the ground to the north of the crater, for several 118 S. SEKIYA AND Y. KIKUCHl kilometres, was utterly turned 'topsy-turvy' and every land-mark ob- literated. For this reason it was impossible to tell the real state of things in that quarter. It is well known that at volcanic outbursts the immense volu- mes of steam, suddenly expanding occupy a much larger space than that of the original bulk. This sudden expansion cools the tem- perature of the surrounding atmosphere and lowers its pressure. More- over, the steam in part condenses. To fill the partial vacuum thus produced, and to equilibrate the reduced pressure, there follows an inward rush of air towards the crater. The strong1 winds commonly described as a feature of volcanic eruptions, are probably due to this cause, and the same thing doubtless happened to a certain extent in the case under discussion. But the fearful blasts that wrought such havoc in the forests and villages on loth of July certainly were not counter currents of this class, however strong these may have been. It was the gusts from the volcano that in this instance wrought the real havoc. A whirlwind is described as having occurred during the volcanic eruption* in the Island of Sumbawa, in Java, on the 5th of April, 1815, when, soon after the ashes began to fall, a violent whirlwind ensued, which blew down nearly every house of Sangir; it tore up by the roots the largest trees, carrying them into the air, together with men, horses, cattle, and whatsoever came within its influence. The whirlwind lasted about an hour. It is not stated in this account, however, how the whirlwind was caused. Conical holes. Several visitors to Bandai, ascending from the south sides and approaching the summit, had their attention attracted by numbers of curious conical basin-like hole.-, evidently the fruits of the late eruption. Their size varied from "1 to 3 metres in diameter and from * Lyell's Principles of Geology, 12th Ed. Vol. II, p. 104. THE ERUPTION OF BANDAI-SAX. 119 a few decimetres to more than a metre in depth ; and they were gener- ally wider at the month than at the bottom. They were found in thousands in the neighbourhood of the crater, as well as on the ex- tensive slopes of 0 bandai and Akahani, a few kilometres away. As to the origin of these holes, though they were not perhaps a very important phenomenon of the eruption, it calls for a brief discus- sion, because it has been the subject which has caused a good deal of diversity of opinion and some lively controversy. First of all, they were regarded by some witnesses as miniature craters, each formed by a small explosion of steam. This supposition arose from the facts that several of the holes had the appearance of having been formed by ejections from below the surface, and that in some cases steam had been seen issuing from the holes. But the latter phenomenon was only to be seen near the solfatara adjoining the new crater, from which steam had been issuing for ages; and the appearance of fresh steam-jets in that locality would not be surprising after the convul- sions caused by the catastrophe of the 15th of July. Even, however, if it be allowed that there were a few slight steam explosions in this immediate vicinity, it is, in our opinion, impossible to extend any such hypothesis to the innumerable holes of like appearance that were scattered over the extensive and distant slopes of Obanclai, Akahani &c. Prof. J. Milne on the other hand, regarded the cause of the holes as seismic, but not volcanic, that is to say, he concluded that earth- quake-waves produced at the time of the eruptions, and passing through the soil, caused sub-surface compression and distortion and thereby ejected earthy matter from below, by the spouting action of water. We shall show farther on that the above theory can hardly be applied to any bur a few marshy spots in Numano-taira, to be presently described. Others, again, contend that the holes were caused by the fall 120 S. SEKIYA ANJD Y. KIKUOHI of stones projected into the nil* from the crater. This we take to be the true explanation. Puzzled at first as to the origin of the holes, hut determined to investigate it fully, we dug a number of them open. Thereby our doubts were soon cleared up. AVe found im- bedded in the ground brittle and freshly fractured stones that had apparently been shot forth by the volcanic outburst. AVhere the soil was rocky, fragments of similar stones were scattered all about. In one case we obtained very decisive evidence. A stone was discovered in the soft loamy soil form- ed by the decomposition of andesitic rocks, at a depth of nearly 2'5 metres below the surface. The stone was angular in shape, measur- ing nearly 05 metres each way, and showed a freshly broken surface. Leaves of bamboo, dwarf pine, and creeping surface niants had been carried with it into the soil. These were much packed and crushed, but were still fresh and green when dug out. On closely examining the hole we found that, although more or less funnel shaped near the surface, it had been sharply cut through in a tubular form, by the passage of the stone which lay at its bottom. The original tubular passage was, however, filled up by loose detritus mixed up with ashes, and here and there with shrubs and bamboo twigfs, while the surrounding: soil was a compact native red loam. The main features of this hole are shown in the accompanying diagram. It was stated before that the steam ascended perhaps 1,280 THE ERUPTION OF BANDAI-SAX. 121 metres (4,200 ft.) above the crater. If stones were thrown up to the same height as the steam column, their initial velocity must have been 158 metres per second, and this may be roughly taken as the final velocity with which, on falling down, they would reach the ground. But if we suppose that the steam and stones reached to double or treble the above height, which is not improbable, the initial velocity be- comes 224 and 274 metres per second respectively. The velocity need- ed for penetrating a soft loamy soil to a depth of 2*5 metres would be between 300 metres and 600 metres per second, according to the values of coefficients we take, but in these calculations, as we are using several more or less arbitrary assumptions we cannot take them as a sound basis for discussion. It seems not strange, however, to find that the basin-like holes wearing those appearances of having been blown out from below have led to the theory of their formation by internal explosion. Stones striking the ground with great force would make holes of larger diameters than their own by throwing; the surrounding earth outward. Again, if the fall of anv stone take place at a spot where there are rocks on or near the surface of the ground, the concussion will ."hatter the falling stone and at the same time blow up the ad- joiningd soil, thus producing the appearance of an eruption. We have witnessed great number of those cases on Obandai and Marumori-vama. The inhabitants at the base of the mountain noticed among the rising steam small and large white objects ascending and descending like shooting stars. At one time, they were so numerous that they almost looked like white rain. It seems very probable that these ob- jects were stones ejected from the crater. That, large fragments of rocks are hurled into the air during volcanic eruptions is a matter of common experience. But that they have left their traces upon mountain sides in the form of conical holes 122 S. SEKIYA AND Y. KIKUCHI has not, as far as we know, been recorded; perhaps it hns never been observed. It must not be forgotten, however, that there would have been no such holes on Bandai-san if there had not been the thick layer of soft loamy soil to receive the falling rocks. Mr. E. Odium, of the Toyö Eiwa Gakko (Oriental English school), Tokyo, made a thorough investigation of the origin of the conical holes. He went tw:ice to Bandai, the. second time for the sole purpose of examining the holes, and his observations on the spot were very complete. Indeed, we consider that the facts and proofs brought forward by this observer must be held to settle the ques- tion. In a paper read last autumn before the Seismological Society of Japan, Mr. Odium showed that hundreds of thousand of stones have been hurled into the air from the crater. People were wound- ed and forests were shattered by them. Sometimes fragments of rocks of considerable size were imprisoned on broken trunks of trees. Native mountain rocks had on their upper surface marks and scars made by the stone projectiles. On excavating some of the holes, Mr. Odium found in them imbedded stones, some of which weighed 4,000 lbs, or 1,814 kilogrammes. Under these stones grasses, wreeds, leaves, branches, and other kinds of vegetable materials wrere some- times discovered, often bruised and shattered till they had the appear- ance of having passed between rollers. Many of the stones fell in a slanting direction ; they were not lodged in the centres of holes, but almost always to one side, that is, the side away from the crater. The earth round the imbedded stones was solid and native — no sign of their having been disturbed by explosive action or the like was found. The whole mountain, the top as well as valleys, was covered with these pits as if it had had a heavy attack of smallpox. To suppose, con- tinues Mr. Odium, that the holes were formed by the spouting action of subterranean water, as is held by some authorities, we must assume THE ERUPTION OF BANDAI-SAN. 123 that the whole mountain was literally made up of water, and inunda- tion must certainly have resulted from the creation of such immense number of water jets in a short space of time. Mr. Odium made numerous measurements of the pits; they vary from a few feet in diameter to over thirty feet, and from 2 to 10 feet in depth. Lieut. Y. Nakashiina, of the Army Department, who surveyed the volcano after the eruption, and who, from the nature of his work, acquired an intimate knowledge of the whole area is in entire accord with our opinion as to the origin of the pits. Notwithstanding these evidences, however, the conclusions ar- rived at by us and other workers have been freely criticized, and doubts have been thrown upon them. Prof. J. Milne of the Imperial University, dissenting from our views and those of Mr. Odium, be- lieves that the cause producing the holes was seismic in character — to wit, the severe earthquake that accompanied the eruption. He quotes Robert Mallet in support of the hypothesis that they were pro- duced by the spouting action of water from beneath, resulting from seismic compression of the substance of the ground. Similar pits, he says, were made in the great Calabrian earthquake of 1783, and they were specially investigated by a committee sent from the Royal Academy of Naples. These gentlemen also dug into holes, Prof. Milne con- tinued to say, but we do not hear of their having found any boulders. We (the authors of the paper) think that the Neapolitan scientists did not strike into boulders, simply because the pits in the Calabrian plain were not formed by falling stones, which was the case on the slopes of Bandai. It was also argued that like phenomena were observed in the Charleston earthquake of 1886, when sand and muddy water were ejected, making more or less conical holes. But, as far as we under- stand the matter, the formation of holes during destructive earth- quakes by the spouting of water or sand is limited to plains and 124 S. SEKIYA AND Y. KIKUCHI watery places. In the case of Bandai, sucii formation might have oc- curred in the marshy plateau of Numano-taira and perhaps in other limited areas among the valleys. But we regard it as impossible that ejections of water or soil, produced by seismic action, could have occurred over the great area covered by the pits under discussion on the rocky summits and steep slopes of Bandai and Akahani. It is true that in Numano-taira large round or elliptical holes were made, some of them having diameters of over 7 metres and being left partly filled with water. But, though it is not impossible that these par- ticular holes were produced by seismic action, the more probable ex- planation is that the filling débris from Kobandai covered the ground to a great thickness, burying under if forests and ponds; and it is natural to expect that in this great field of loose debris depressions would be formed here and there by the falling in of the superincum- bent layers. We ourselves observed some of the holes gradually in- creasing in size by the filling in of their margins. In their paper read before the Seismological Society of Japan, Professors C. G. Knott (Tokyo) and C. Michie Smith (Madras) con- tend that by far the great majority of holes are due simply to the uprooting of trees caused by the hurricane. This they regard as es- pecially true of the region to the south-east of Obandai. They argue that if the countless numbers of holes were due to falling stones a great many other stones must surely have fallen on the same ground with velocities too small to make holes. But of these other stones there is no evidence. Besides they doubt if it is dynamically pos- sible for a filling stone to make conical holes of the size and form described so fully by Mr. Odium. For a stone to bury itself several feet in the ground and at the same time make a violent " splutter " is not, they think, at all credible. After the undoubted effects of uprooting of trees, and of landslips in a volcanic soil violently shaken THE ERUPTION OF BAXDAI-SAX. 125 by an earthquake, are given their full weight, they consider that a very limited number of holes will be left to be explained by falling stones ; and that these holes will be either flat basin-shaped bruises or tubular cavities of comparatively small diameter. We shall not stop to answer all the points criticized by Pro- fessors Knott and Smith as we think we have already spoken enough on the subject. It is quite true, we admit, that conical holes were produced by uprooting of trees in forests; indeed we saw numbers of such holes near the upturned roots, especially on Miné-yama. As the origin of these there could be no doubt. But by far the Greatest numbers of holes were found on bushy hills, on grassy slopes, in rocky glens, or in regions that were not occupied by forests, and on ground where the trees still stand uninjured. Such are the holes which we believe to have been formed by falling stones, and to which we have been referring thron o-hout. The fact that the bamboo leaves which we discovered under the stones, and which we brought back to Tokyo, were comparatively uninjured was much criticized by saying that, if they had been forced into the ground by the bonders, they must have been more severely damaged. Those which we discovered were, however, originally much folded and packed, while the specimens collected by Mr. Odium were crushed to a much greater extent. Mr. Otsuka, a school-master in Hibara, who kept a diary of Earthquakes, the weather and other matters, told us that there were slight shocks of earthquake on the 8th, 9th, and 10th of July. At about 3 o'clock on the afternoon of the 13th there were occasional shakings of the ground which were also felt at Inawashiro. Between 3 and 4 o'clock on the afternoon of the 1-1, i. e., the day preceeding the eruption, quite an extensive earthquake occurred in the neighbouring provinces, but although it was felt in the Bandai-san district its origin was far away 126 S. SEKIYA AND Y. KIKUCHT to the west near the coast of the Japan Sea. On the morning of the 15th, at a little after 7, a feeble earthquake occurred, but it was so slight that many failed to notice it. After half-past seven a severer shock ensued, lasting nearly 20 seconds. This was followed soon after by very violent convulsions of the ground; houses rocked and swayed, furniture fell down, and the frightened people felt the ground heaving beneath their feet. It was reported that the nature of this earthquake differed from that usually experienced, in that vertical motions greatly predominated. The shock was a long one, lasting, according to some accounts, for fully a minute. While it was still in progress the eruption took place. Considering the fierceness with which this earthquake shook the im- mediate vicinity of the mountain, it is remarkable that the intensity was very rapidly decreased as the seismic waves were propaged into the surrounding region, and the shaken area was limited to a radius of, roughly speaking, about 48 kilometres, or 30 miles. This may be accounted for by the fact that the origin of the shock was rather near the surface. On PI. XXIV the boundary of the area shaken by the earthquake is shown by a thick elliptical line. Volcanic eruptions are generally accompanied by earthquakes, and the shocks on this occasion prove that sudden expansion of steam and breaking up of the earth's crust may produce seismic vibrations. The efforts of the pent-up steam, struggling to force its way through the superincumbent masses, at last succeeded in bursting through a weak point, the explosion being accompanied by violent convulsions of the ground, that were propagated as seismic waves. On the 20th of July, at 11.50 a.m., a feeble shock was ex- perienced. Other minor shocks are said to have subsequently occurred. To see whether the ground in the crater was perfectly quiet after the eruption, we took with us a delicate though somewhat THE ERUPTIOX OF BANDAI-SÂ.N. 127 roughly made pendulum tromometer, which had a magnifying power of '27. We first set it near a fissure from which powerful jets of steam were issuing with hissing sounds. Id that position the instru- ment indicated very feehle vibrations of the ground, which were doubtless caused by the issuing jets. We next set it in ISTakanoyu, near our cam]), and made daily observations. As far, however, as the magnifying power of the instrument enabled us to judge, there was no evidence of lingering earth-tremors. And it would seem from this that the forces which produced the explosion had been completely expended in blowing away the mountain, and that the region had already become seismically quiescent. Near the volcano the detonations in the earlier part of the erup- sound, tion were described as deafening. Though rapidly lessening in inten- sity, the thundering noises lasted for nearly an hour, and did not en- tirely cease for several hours. The area over which the sound ex- tended was very much smaller than might have been expected From the tabulated reports at the end of this paper, and from other sources, it appears that the sounds of the explosions were not heard distinctly at a o-reater distance than 48 kilometres or 30 miles to windward of Bandai-san, though to leeward they were audible at the Pacific coast, a distance of 100 kilometres or 62 miles. How much farther they reached, over the sea, we had no means of ascertaining. Mr. K. Xakashima, of the Geological Survey, told us that, while he was as- cending Kinboku-san in the island of S:ido, on the morning of July 15th, he heard did! rumblings which were supposed at the time to proceed from the firing of heavy guns in the neighbouring harbour (Report 21). Sado is in the Japan Sea, nearly 161 kilo- metres or 100 miles west from Bandai-san. and therefore to windward of it. Possibly, however, the sounds heard by Mr. JSTakashima came from the volcano. It was also reported that peculiar detonating 128 S. SEKIYA AND Y. KIKUCHI sounds were heard on the same morning in Takai-kori, in the Province of Shinano, distant about 164 kilometres or 102 miles southwest from Bandai-san. The barograph in the Imperial Meteorological Obser- vatory in Tokyo, which is 212 kilometres or 132 miles south of Ban- dai, was not affected. The magnetometer in the same Observatory also o-ave no record that could be regarded as an effect of the eruption. While we were staying on the mountain we often witnessed the fall of large masses of the perpendicular crater wall, which, coming down from great heights with stupendous force, were smashed into thousands of fragments and descended with whirlpool-like move- ments to the lower levels. These slips produced terrible rumblings, which resounded throughout the crater, and were also heard far away. By the already panic-stricken inhabitants of the neighbouring villages these repeated noises were regarded with consternation as tokens of a further volcanic outbreak, and it was weeks before they became pa- cified and assured as to the real cause. Lightning'. From the ascending columns of steam and ashes vivid zig-zag- flashes of lightning were seen to dart forth, and were accompanied by loud roars of thunder. These phenomena, observed from several points around the mountain, may be regarded as resulting from the discharges of friction.il electricity which, as is well known, are liable to be brought about in volcanic explosions when steam at high ten- sion escapes through a narrow orifice, and collides with the surround- ing air and the more solid ejectamenta. Sparks. While the main eruptions were going on, the people in Inawa- shiro and the neighbouring villages saw through the densely falling ashes innumerable vivid sparks of fire on the slopes of Obandai and Akahani, at considerable distances from the crater. These sparks were quite different in nature from lightning, presenting rather an appearance as of the firing of a vast number of guns. The probable THE ERUPTION OE BANDAI-SAX. 129 explanation of the phenomenon is that .sparks of fire were produced by stones and rocks striking against each other in the air or falling on a rocky bed. Fragments of rocks are scattered in abundance on the slopes of Obandai. but we could discover nothing to lead us to believe that there had been combustion or any other heat mani- festations. Sensational newspapers in their accounts of the eruption spoke of lurid flames, of a blazing crater and other terrors all probably founded on the peasants' reports of the sparks above mentioned. It very rarely occurs, however, in volcanic eruptions that flames are pro- duced by the burning of gases issuing from craters. Sir W. Hamilton, in describing the Vesuvian eruption of 1779, noted that large vitrified masses (bombs), falling upon the ground, broke into many pieces, and set fire to combustible objects. In this case, however, the fire was produced mainly by the heat of the fused masses, whereas at Bandai the sparks were caused by impact. On the other hand, at the time of the great volcanic eruption of Tarawera, ]\Te\v Zealand, in 1886, the falling sand was said to have been hot and to have set the trees on fire.* As regards unusual optical phenomena, we have heard of one or Optical two only that seem to have been connected with the eruption. Any- p enomena- thing like the twilight-glows, haze, etc, which were such important features after the Krakatoa explosion, could hardly be looked for in this case, which, though exceedingly remarkable in many respects and interesting in the highest degree, was very much inferior in magnitude to the gigantic eruption of 1883 in the Sunda Strait. However, an observer at a place about 87 kilometres or 54 miles E. S. E. from Bandai noticed towards evening on the day of the eruption sparkling rays of red light issuing from the clouds (Report * Nature, Vol. 34, 1888, p. 392. sisms. 130 S. SEKIYA AND Y. KIKUCHI 15). Another, living in a village S. S. E of the mountain, saw with mixed fear and delight how the rising columns of steam from Bandai- san, refracting and decomposing the light that fell upon them, pro- duced a most beautiful display of variegated colours (Report 8). Premonitory It has been often asked whether there were any premonitory signs of the explosion. It is certain that slight shocks were felt on previous days, as well as half an hour before the outburst. But beyond this the evidence is vague. Some persons vouch to having heard mysterious rumbling sounds in the mountain prior to the eruption. Again, some animals are said to have shown alarm. No doubt before an explosion of such magnitude as that of Bandai-san, earth must have been in a seismically sensitive condition, and certain animals which are known to be highly susceptible to even minute earth-tremors may very well have been frightened on this occasion. Some well-waters are also said to have diminished in flow. None of these alleged facts, however, have been clearly established on the evi- dence of trustworthy persons. On the other hand, Mr. Tsurumaki says in his letter (page 105) that the bathers at Nakanoyu did not observe any abnormal changes in that spring, though it is situated on the very edge of the new crater. It would be interesting to know exactly how Lake Inawashiro behaved before the eruption. The water level of that lake is systematically recorded at two places on the shore, for purposes of irrigation. But we were unable to learn, either from conversation with Mr. Akiyama, who keeps the record, or from the entries in his books, that there had been any sudden fluc- tuations of level during the two months preceding the catastrophe. In reports lodged at the Fukushima Prefecture it is stated that the river Nagase, the Lake's chief feeder, decreased its flow from April or May, and that the water level of the Lake on the 1st of July was one foot lower than on the corresponding day of the previous year. A captain THE ERUPTION OF BANDAI-SAX. 131 of a steamboat running on the lake told us that his vessel, anchored close to the shore, was moved outwards nearly 0-6 metres by the distur- bance of the water. Though this disturbance occurred almost simul- taneously with the eruption, we are disposed to agree with the cap- tain in believing that it was a result of the earthquake immediately preceding it. On the whole, the only premonitory signs that were really trust- worthy were the slight shocks which occurred on previous days and half an hour before the eruption, and no implicit faith can be placed on the slight testimony in favour of other warning symptoms. Volcanoes have often been described as one of the principal change of topo- restorative agents in counteracting the denuding action of water that fiap ° o o tures. tends to bring the surface of the earth to a level. In the late eruption of Bandai, however, the effect was destructive and not constructive. The materials which had accumulated in past ages gave way, and were thrown down from a higher to a lower level in less than an hour ; the effect of this being analogous to a gigantic land-slip. In this way, considerable changes have been wrought in the topography and contour of the adjoining districts. How the torrent of earth and rocks devastated an area of some 70 square kilometres (27 square miles) has already been described in the preceeding pages. The general effect of this spreading out of débris was to effect the level- ling of the general contour ; all the surface ravines and gorges, being entirely filled up. The northern side of Bandai was before the erup- tion, an undulating grassy plain — the 'Hara' so characteristic of volcanic districts in Japan — drained by the river Nagase, and dotted here and there with some straggling hamlets. Professor J. Milne who visited this part of the country several years ago described it as consisting of grassy slopes without any exposure of rocks. The des- cending deluge of debris pouring across Obudaira, as the northern 132 S. SEKIYA AND Y. KIKUCHI slopes of Kobandai are called, engulfed all the familiar land-marks, and converted the district into a desert waste. Thousands of conical mounds, large and small, have been formed on this vast sea of mud, — giving a quite unique appearance. Not only have the depressions been filled up, but the higher ridges have been reduced in height. The deluge of earth débris in its quick descent, impinging on the pro- minences that were lying in its way, have actually leaped over them, scraping off the outer-crust of the soil, and exposing the native rocks beneath, as a glacier might have done. In other places the torrent of rocks dashed against the hill-sides and scoured them away. Large quantities of mud carried along with rock, débris, and boulders, have penetrated deep into every recess of the valley. The largest and longest of the mud streams is that which flowed down the slope of the river Nagase to the Kawakami spa. Near the village of Hibara, and the former site of the little hamlet of Hosono, the peaty deposits which had accumulated in the marshy ground have been ploughed up. Large clumps of red loamy soil mixed up with half carbonized wood and grasses are found turned up or standing in an irregular state of contortion.* Nowhere could the scouring action of the mud torrent be better realized than in these parts. The official reports relating to the area of land buried under mud are given in the following figures. The loss of property involved is said to have been immensely great. There is absolutely no hope of recovering or reclaiming the buried land. * Among this turned up peaty deposit, an interesting occurrence of vivianite may here be noticed. The formation of this mineral seems to have been brought about through the agency of organic matter. It is found in the form of a beautiful azure-blue coloured, fine powder, filling up the spaces especially of the rind and core of the fragmentary branches of half carbonized wood belonging to some Coniferous tree. The spot, however, where we have witnessed this phenomenon, being situated just on the sourthern front of Hibara lake, must have now been submerged under water. THE ERUPTTOX OF BANDAT-SAX. 133 Square kilometres. Square miles. Cultivated land 0*82 032 Plains 22-60 8-73 Mountains and forests 41*93 16*19 Rocky slopes and glens 5-36 2*07 Total 70-71 27-31 It is highly probable that, though at present the new land ap- pears like a desert waste, the growth of vegetation will take place in a comparatively short time favoured by the admixture of the fertile volcanic products. One of the most striking secondary effects of the eruption is the formation of new lakes, due to the damming up of the river Nagase and its tributaries by the débris of the shattered mountain. These lakes are four in number, and are shown on Plate XV. They may b3 con- veniently called Osuzawa, Hibara, Onogawa, and Xakatsu (or Akimoto) lakes, after the tributaries of the Xagase to which they are respectively due. The largest of them is Hibara lake, measuring nearly 4 kilometres or 2.5 miles from north to south, the breadth being nearly 1/3 of the length. These lakes continued for many months to increase in size, through the gradual accumulation of water within the newly formed barriers of debris. Thus it was not till fifteen days after the eruption that the village of Onogawa became coverd with water. The inhabi- tants then fled to Hibara; but were subsequently driven out from that village also as the waters gradually rose. Beside these four large lakes, there are scattered among the mud- field smaller patches of water caused by the accumulation of rain water, or formed by the smaller streams on the mountain-sides. These lakes will continue to increase in size until the water comes to the level of the lowest possible outlet from the hemmed-in basin. The 134 S. SEKIYA AND T. KIKUCHI issuing stream will soon cut deep passages through the loosely cohering debris; and it is to be expected that sudden yielding« of some of the barriers will take place. Hence will result a rush of es- caping water accompanied by violent floodings in the lower courses of the stream. It was thus that certain villages and cultivated field were flooded when the Nagase-gawa burst its lowest barrier, which immediately after the eruption quite stopped the flow of that river into Lake Inawashiro. Because of the loosely compact character of the debris, the configuration, size, and number of these lakes will alter greatly as time goes on.* On the 13th of April of this year (1889) about 6 p.m., a large portion of Onogawa lake was suddenly drained, and the torrent of water rushed through the mud-field, carrying mud, pebbles, and boulders to the lower levels. The embankment newly erected at Nagasaka and other places to protect from inundation, was destroyed, and the water spread out into the cultivated fields adjoining the dis- trict of Inawashiro. Considerable damage was done to bridges, and roads, but fortunately the houses and inhabitants escaped. New lakes or ponds are also being formed within the crater, by the condensing steam and the rain water. The waters of these lakes contain much soluble matters and some of them are hot. But as the crater-bottom is set in a sloping position, an accumulation of water to any considerable extent cannot take place. * As this is passing through the press, we are able through the kindness of Professors C. G. Knott and C. Miehie Smith to call attention to the remarkable changes in the region under consideration. These gentlemen visited Bandai towards the end of May, 1889. They report that Hibara and Ösawa lakes have united into one huge lake which forms the most promi- nent feature in the landscape. The other lakes are comparatively small. Of particular in- terest, also, are their observations on the erosion of the new earth by the action of running water. Thus the comparatively small stream that ran down Biwa-sawa has cut out of the new earth a deep V-shaped gorge, in many places attaining a de2)th of 40 or 60 meters. At these places the stream has not yet cut down to the original surface. Their observations are to be published in full in the Transactions of the Seismological Society of Japan. THE ERUPTION OF BANDAI-SAN. 135 The distant view of Bandai as seen from certain directions has been altered by the destruction of Kobandai. When viewed from its south side, however, e.g., from the town of Wakamatsu, the mountain apparently does not show any sign of great change ; the prominent peak of Obandai entirely screening from view the place of devastation. Before the time of the explosion, the top of Kobandai, when seen from this side, was presented as a small prominence on the left side of Öbandai. Fis:. 3, PI. XXI, is a sketch taken soon after the catastrophe. The dotted outline in this figure is the original form of Kobandai, which has now entirely vanished, and in its place columns of steam are seen rising. The most magnificent sight is presented to view when the moun- tain is seen from the northern side, where the full force of the ex- plosion may be best realized. The newly opened explosion -crater (Fig. 1, PL XVI, Fig. 1, PL XXI) is fully disclosed; the wreath of steam rising from its central linear fissures like a cumulus cloud. The sound of the evolving steam is distinctly heard from the village of Hibara fully 9 kilometres distant, evidently due to the fact that the crater opens unobstructed on this side, and is backed by perpen- dicular walls. On the right side we see a jagged rocky precipice, the remains of Yugeta-yama which formed a small prominence on the flank of Kobandai. The extensive waste of mud and rocks, gradual- ly sloping from the crater toward the plain below with its curious conical rock-hills and bared mountain sides, gives a most vivid and awful impression to the mind of the vastness of the devasta- tion. It would have been very interesting to compare the original form of the mountain as seen from the north with its present form. But as we have already stated this part of Bandai was very scantily peopled, and hardly ever visited by scientists so that there seems to 136 S. SEKIYA AND Y. KIKUCHI exist no sketch or photograph showing; an accurate outline of Ko- bandai before the explosion. The accompanying figure shows the probable appearance of the mountain previous to the eruption ; this reconstruction of the original scenery being made from the sayings of the people who knew the mountain well, and from general topographical considerations. The dotted line in the figure is the present outline of the mountain, viz., the reproduction of the outline as shown in Fig. 1., PI. XXI., reduced to one half. The prominent peak (k) is intended to be the restored form of Kobandai or Little Bandai, which when seen from this side must have looked more massive and prominent than Obandai, or Great Bandai (ci). To the west of Kobandai immediately over Kamino-yu (e), is seen a small prominence known as Yugeta-yama (?/), which was half destroyed, forming now a very rugged precipice as already mentioned. It is to be observed that the fumarole of Kamino-yu is situated within a small ravine having a very steep side- wall, probably itself a small explosion-crater formed at some former period. Four hamlets — Osuzawa, Hosono, Akimoto and Kawakami — have been completely buried beneath the rock and mud along with their THE ERUPTION OF BANDAI-SAX. 137 inhabitants and cattle, leaving no visible trace of their former existence. Even the few survivors who escaped death by their timely absence from home could not tell where their villages had been, as every land mark was entirely obliterated. Seven villages — Miné, Xagasaka, Shibutani, Shirakijö, Hinokuchi, Myöke and Ojigakura — were partial- ly destroyed either by the avalanche of earth or by the storm of wind and rubbish, and were thickly covered by ashes. The loss of life and cattle was also considerable. The destruction of the three spas in Bandai-san, viz., Kaminoyu, Nakanoyu and Shimonoyu, has already been mentioned. In all, 166 houses were either totally or partially destroyed. The total number of lives lost amounted to 461. The principal cause of the death was the deluge of rock and mud debris. In some instances people were buried under their roofs, having no time to es- cape; but in the majority of cases they were caught in the swift tor- rent of mud while endeavouring to reach some safe place. Many were also battered by falling stones. A most remarkable incident occurred in the village of Xagasaka which had 168 inhabitants, and was attended with a serious loss of life. This village being' situated behind Kushigamine was effectually screened from the direct attack of the mud current. On the morning of the eruption when stones and earth began to fall upon their roofs, accompanied by appalling noises and earthquakes, men and women rushed out of their houses leaving the old and young behind, and attempted to cross the valley of Xagase in order to reach the opposite hill which they thought to be a safer situation. They had only to travel not more than 500 metres across, but of ninety-two who thus fled, not even one reached the other side safely. Out of the total of 461, only 117 corpses were recovered, all the remaining bodies being entirelv buried under the mud. The 138 S. SEKIYA AND Y. KIKÜCHT number of wounded was 70 in all. They were mostly burnt and scarred by the hurricane of hot ashes and falling stones. The wounds in several cases were very peculiar. Fragments of rock projected violently from the mountain, impinged upon the unfor- tunate people like grape-shot. The sufferers were tossed about and often felt as if they were lifted up bodily into the air; at the same time their clothing was torn off, even to the under-garments. In many cases bits of stones were found sticking in the skin and flesh of the victims, and were with difficulty extracted. In one case the skin was completely peeled off from a woman's skull, probably torn off through entanglement with trees or other objects which were being violently hurled along. The mere contemplation of such experiences is hor- rifying. Bandai-san The winter climate of this part of the country is very severe. Snow after eruption. fe . &|j . the mid(j]e of autumn, so that in winter travellers are rarely seen, and the peasantry hardly ever venture far abroad. The result is that very little knowledge can be obtained about the behavior of the volcano during the cold months. Mr. S.Kobayashi, a school- master at Inawashiro, who helped us in various ways during our stay in Bandai-san, has, at our request, very kindly given us the following information : Letter dated Nov. 7th, 1888. — Since the eruption, even in bright and calm days clouds have been almost always seen round the summit. This has not been so in former years. This is very probable. The steam that issues from the crater, in ascending and dispersing in the higher region, would produce clouds. Nov. 8th. — The volume of steam has abnormally increased since yesterday. It is nearly the same as it was three weeks after the eruption. This morning the amount is still greater. Although its height is perhaps not greater than that observed THE ERUPTION OF BAXDAI-SAN. 139 within ten days after the eruption, its volume does not seem to be less. Oct 7th. — The new lake Akimoto rushed out cutting through the mud-field, after a heavy rain-storm. In the lower course of the river Nagase, the water level suddenly rose 9 ft. above the ground, causing great uneasiness among the people. Oct. 20th, 6h. 32 min. 3 sec. a.m. — A slight shock was ex- perienced. Nov. 19th, Oh. 25 min. a.m. — A slight shock was felt; motion horizontal, direction X. S., duration 1 mill. 30 sec. On this day snowfall was first observed on the summits of Obandai and Kushi- gamine. Nov. 30th. — Thunder-claps were heard toward Bandai-san and Azuma-san, lasting about 20 minutes; have seen lio-htnino- twice. Dec. 1st. — First snow-fall on the ground. Dec. oth. — Sound of distant thunder heard. Dec. 30th. — Great storm swept the accumulating snow over Bandai, and with it the ashes on the mountain-flank. Peculiar rumbling noises heard in the mountain. People believing this to be a sign of a fresh outburst of Bandai were very much frightened. Jan. 1st, (1889). — A slight shock was experienced at 7h. 10 min. p.m., lasting 2 seconds. Jan. 23rd. — A halo observed, probably due to the refrigera- tion of the vapours of the steam from Bandai. Feb. 8th. — Rumbling sound heard in Bandai, probably due to the fallino; of the crater-wall. Feb. 17th, — Hibara-village was deserted on account of the encroachment of the new lake. Feb. 18th, 5h. 30 min. a.m. — A slight shock. 140 S. SEKIYA AND Y. KIKUCHI In his last communication Mr. Kobayashi says that during the winter the action of rain, water, and snow, has greatly eroded and smoothed down the rugged points, the vertical walls and the precipi- tous hills in and round the crater, reducing them to much gentler inclines, and thus the grandeur and picturesqueness of the scenes have been considerably lost. This fact was confirmed by recent visitors to Bandai-san. From this it may be expected that such a bold scene as that shown in fig. 2, PI. XXI, which is a sketch taken from the western edge of the crater just above Kaminoyu, about 20 days after the eruption, would not be long preserved. The Character of the Eruption. The explosion of Bandai has furnished us with an example of a volcano, long dormant, bursting with terrific force. So far as we know the last great explosion took place more than ten centuries ago. Some minor eruptions are said to have occurred in subsequent years, but seem to have been of a local character. During the long period of rest, the original crater now known as Xumano-taira, has in large measure lost its crater-like form by the disintegration of the surrounding walls. In such cases it often happens that subsequent eruptions take place at other parts of the mountain, breaking open new chasms along other lines of weakness. Such has indeed been the case with this last eruption of Bandai; the line of weakness lying to the north of the old crater so that the mass of the mountain was thrown toward the north into the Nagase Valley. There is now seen in the new crater a great fissure running X. 20° W. from the bottom of the horseshoe nearly to its mouth. Along this fissure runs a long row of steam jets, large and small, puffing and hissing and emitting immense volumes of white watery vapour. PL XXII THE ERUPTION OF BANDAI-SAX. 141 is a distant view of the mountain from a photograph taken by Prof. W. K. Burton, nearly a week after the eruption, showing the linear arrangement of the steam-fissures. It was alon» this line that the eruptive power of steam rent Ko-bandai into pieces. It is manifest that the immediate cause of the eruption was the sudden expansion of steam pent up within the mountain. Of lava or pumice there is no trace. The grey coloured ashes which form a chief product of the explosion are evidently the powder of pre-existing rocks decomposed by the action of fumaroles, and have not been deriv- ed from fused magma. The character of the explosion was thus comparatively simple, being to all appearance a sudden shattering of part of the mountain flank. The work of the explosion has been practically to tear off a portion of the side-wall of the old crater. It has often been observed that the first action of some volcanic outbursts is characterized by extreme violence, large masses of super- incumbent materials which have accumulated in the crater being thrown out, or the side- wall being blown away, by the expansive force of steam. This first stage of the eruption is usually followed by minor ones accompanied either by lava flows or pumice ejections. Thus the finer particles or ashes ascending into the air are different in character in the successive outbursts; those which are ejected dur- ing the first stage of the eruption consisting largely of fragmentary materials, while those of the later eruptions are found to be more or less pumiceous, a fact showing that the latter are derived from the fused matter. The explosion of Bandai, characterized by its sud- denness and violence, and effected in a very short time, may be likened to the first stage of eruption. But no subsequent discharge took place, nor were any signs of farther disturbance discernible. We have already indicated that the volcanoes in the vicinity of 142 8. SEKIYA AND Y. KIKUCHI Bandai, though classified as active, have never, within historic times, shown a true lava eruption. And this may be said to hold good for most of the active volcanoes on the Main Island. In fact, most of them, which had their climax of activity during Tertiary times, are now verging to extinction. This is due to the general and gradual abatement of volcanic force since the end of the Tertiary Era. Any great geologic changes which have taken place since then, have had to do with the general rise of the land surface above the sea-level. At present the Tertiary strata, some three or four hundred metres above the sea-level, cover a large part of the whole area of Japan. They con- sist principally of tufaceous deposits, are mostly of marine origin, and are often found surrounding or even underlying the volcanoes. These volcanoes would thus seem to have attained their maximum intensity at a time when proximity to the sea greatly favoured their activity. But the subsequent gradual rising of the land surface has had the effect of shifting the volcanoes further and further from the sea-board, thus greatly mitigating their action. The eruptions which are usually experienced in these volcanoes in modern times seem to be essentially superficial. The recent cata- strophe of Bandai may indeed be taken as a grand example of this kind of volcanic manifestation and may well be called an explosive eruption. The great horse-shoe-like chasm opened toward the north may be called an explosion-crater. Its appearance from the north side (fig. 1, PI. XVI, and tig. 1, PI. XXI) presents a striking resemblance to the deep chasms which are often characteristic of certain volcanoes, e.g., the 'Val del Bove' of Etnn, the Caldera of Palma, &c, and at the same time suggests similarity of origin. Perhaps it may be of interest here to refer to a volcanic outburst which occurred in Japan early last century and which seems to have strongly resembled the present Bandai éruptions It is the great lateral THE ERUPTION OF BAXDAI-SAX. 143 eruption, which took place on the south -eastern flank of Fuji-yama in the 4th year of Höyei (1707 A.D.). By this outburst a great chasm very similar in its character to that of Bandai, was opened on the mountain side. It is an explosion-crater not inferior in magnitude to the latter. Höyei-zan is really a prominence on the outer rim of this great chasm. It has, however, hitherto been regarded as an example of a 'parasitic cone.' It is to be observed that, in case of the Fuji explosion, there were produced numerous vol- canic bombs, which are now found scattered all around the crater, showing that there was ascent of lava. Another striking phenomenon is the presence of a magnificent series of numerous vertical dykes which traverse the side- wall of the crater. On the other hand there may be seen on the perpendicular walls of the newly opened crater of Bandai, bands of volcanic strata exposed, consisting of an alternation of old lava-streams, and of agglomerates of fragmentary materials, a singular feature of these strata being the absence of any noteworthv dykes passing through them. This fact would show that since the formation of the Bandai group, the mountain has been very little at- tacked by the intrusion of molten magma from the interior. Again, as we have already seen, among the products of the late eruption there is no lava. There are, however, among the debris, especially near to the spot where the steam-jets are issuing within the crater, some greenish coloured rock-specimens with disseminated patches of iron-pyrites, and a white coloured sintery rock having a bleached appearance, almost wholly consisting of silica as shown by the analysis given in the sequel. These materials indicate the action of fumaroles and hot springs in altering the rocks in the interior of the volcano, and in depositing mineral matter while per- colating through the crevices. Among the debris that ran down to the north, are often found porous or scoriaceous rocks of a reddish 144 S. SEKIYA AND Y. KIKUCHI or black colour, which some observers have referred to a molten origin. It is more than probable, however, that these rocks are the fragmentary materials which, having been ejected during the pre- vious periods of activity in the history of the volcano, had accumulated in the form of volcanic strata, and taken their share in the building- up of the mountain. The destruction of Kobandai, which was itself a part of the side-wall of the volcano, reduced these strata to a powdery state, and scattered the scoriaceous materials imbedded in them abroad on the mud-field. Survey of the Crater and the Volume and Weight of the Mountain destroyed. After great volcanic eruptions, surveys have sometimes been made with the view of estimating the dimensions of the parts blown away, or the amount of material ejected during the outbreak. The difficulties attending these attempts are obvious, for it happens very often that the seat of the outburst is inaccessible on account of lava flows or other dangerous obstacles that beset the way. In estimating the dimensions of the newly opened explosion- crater which was formed by the destruction of Kobandai, and in deducing therefrom the volume and the weight of the mass that was thus removed, we encountered one serious difficulty, viz., igno- rance of the original topography and former contour of the moun- tain. x\n unexpected advantage was however derived from the fact that, since in the case of Bandai the explosion lasted a very short time and no subsequent outbreaks took place, we could walk into nearly every nook and corner of the newly formed crater, and thus with comparative ease complete a survey which might otherwise have been almost impossible. The inaccessible or dangerous parts were the THE ERUPTION OP BANDAI-SAX. 145 fissure lines whence steam issued with great violence, the gorges filled with water (new lakes), and the regions at the base of vertical cliffs down which earth and rocks, and even slabs measuring one or two hundred metres, were constantly thundering. Otherwise the conditions were as favourable as could reasonably be expected in the circumstances. Not that the inside of the crater was anything like a level plain. Far from it ; it was cut up by precipitous hills, deep chasms, wild depressions of all imaginable shapes and sizes. PI. XIX shows a prominent rock-hill inside the crater, on which was located one of our stations. The crater-bed was full of these boulder-mounds, and was so irregular that we spent nearly two days in fixing upon a suitable base-line. We were, however, fortunate in finding a nearly straight narrow band in the bottom of a valley in which to locate the base-line. For five successive days Mr. I. Toya laboured indefatigably, until all the important points were measured from prominent heights in and round the crater. When this was done, we removed to Ottate, a spa on the south flank of 0 bandai, and mapped the results of the triangulation. The form of the crater, as deduced from the plan, is semi-ellip- tical, or like a horse-shoe. It opens toward the north, its bed gently inclining outwards. The plan in PI. XXIII shows the general outline. It is surrounded by precipitous walls and steep cliffs of great height especially on the southern side, where a part of Kobandai still 'remains with rugged edges, and where Kushigamine exposes a clean section. The heights of the wall as may be seen from the numbers indicated on the plan gradually become smaller as we approach toward the mouth, at last reaching to the same level as the crater-bed. The heights in and round the crater were measured by taking altitude angles and were afterward referred to sea-level. The 146 S. SEKTYA AND Y. KIKUCHl position of the steam fissure running N. 20° \V. is marked with star- like signs ; prominent hills, depressions, valleys, ponds, etc., are also indicated. Roughly speaking the crater measures 2,463 metres or 8,080 feet across its mouth, which is the widest part from east to west. From the bottom of the horse-shoe to its mouth it is 2,274 metres or 7,460 ft. The total area of the crater-bed is 3*83 square kilometres or 946 acres, or nearly I/o square miles. In the estimation of the volume and the weight of the mass blown away the chief difficulty encountered was our ignorance of the origi- nal contour. The original height of Kobandai, or Little Bandai, was assumed to be equal to that of Obandai, or Great Bandai, (1,840* metres or 6,037 ft.). It was generally believed that Kobandai was a little lower than Obandai, probably owing to the fact that the former was situated further away from the more populous districts in the vicinity and was partly screened by the latter. We were told, however, by those who knew the district well that although Kobandai looked smaller in bulk than its sister peak, there was no appreciable difference in height, and that snow used to fall on the former earlier. This perhaps might be due to its more northern position. Lieut. Y. Nakashima of the Surveying Bureau of the Army, who subsequently surveyed Bandai-san, confirmed this view ; and after a thorough examination of the topography of the district and the forms of the various peaks, he concluded that Kobandai had been equal to, if not a little higher than, Obandai. On the sound judgment of this specialist we can safely place our confidence. We obtained permission to see the map he constructed. It was made with characteristic painstaking care, and every detail was ad- * The barometric measurements by Messrs Y. Wada and X, Ötsvika, of the Imperial Me- teorological Observatory. The eruption oi1 bandai-sax. 147 mirably carried out. We are glad to say that our measurements' and those of Lieut. ïsfakashima agree well ; some difference existed in regard to the heights of crater- walls, but this was more reasonable than otherwise, as they were daily being reduced in altitude by the falling in of the perpendicular edges and steep prominences. The height of the crater-bed above the sea level was 1,170* metres or 3,839 ft. The south-western part of Kobandai which was left undestroyed, exposed an almost perpendicular wall of 505 metres or 1,658 ft. overlooking the crater. From these and from the other data already mentioned, it was found that the part of the mountain that broke away had an altitude of 670 metres or 2,198 ft. above the crater, and that 164*6 metres or 540 ft. had been sheared oft* from its top. On PI. XX11I is given a rough vertical section through the line AB, that is the section passing through the original summit and the highest fractured edge of Kobandai. The doited line shows the parts blown off. It will be seen that the vertical line passing through the summit of the mountain was situated about 442 metres or 1,450 ft. N.E., from the fractured edge, its position being indicated on the plan by a cross. The main body of Kobandai and the north-eastern side were completely blown away, leaving a portion of the south western Hank. The flank of Kushigamine sloping S.W. met that of Kobandai sloping N.E., practically forming one connected mass, and the consequence was that the latter' s destruc- tion was shared by the former in parts where they were so united. Kushigamine now stands overlooking the crater with a bared per- pendicular wall 452 metres high. It is also to be observed that the great fissure line along which the mountain was rent nearly passed through the summit vertical, * Also from Messrs. Wada and Ötsuka's measurements 148 S. SEKIYA AND Y. KIKUCHI or in other words, the fissure-line lay under the summit of Ko- bandai. For ease of calculation the shape of the mountain was assumed to be conical, a near enough approximation. Then, subtracting the parts left undestroyed and making other allowances, we found the total volume of the mountain blown away to be 1*213 cubic kilo- metres or 1,587 millions cubic yards. This is equivalent to say that if the devastated area extending over 70 square kilometres or 27 square miles had been evenly covered with a stratum of earth, rocks and boulders, this stratum would have had an average thickness of 17#4 metres or 57 ft. The cubic content above given represents the gross total of the volume of the mountain destroyed, including not only the débris of earth and rock that descended the mountain -sides, but also the dust, ashes and boulders which were hurled into the air. To calculate the weight of the material corresponding to this cubic content, we determined the specific gravity of different kinds of rocks and earth obtained from Bandai-san. The specific gravity of the pyroxene-andesite composing the mass of the mountain differed more or less with the different varieties met with, ranging from 2*58 to 2*71. On the average it may be taken as 2*65. But as the mountain consisted of much looser materials than these rocks such as pumice, scoriœ, &c, the density of the mountain would be much lower than this value. The mean specific gravity of the earthy materials thrown down by the eruption, as determined by Prof. J. Sakurai, was 2*172. We may suppose without much error, that the mountain mass of Bandai consisted of rock and earthy materials in the proportion 1 : 2, and then we obtain from the foregoing numbers, as the density of Tilt: ERUPTIOX OF BASDAÏ-SAN. 149 the mountain, 2'33.* From this number and from the already es- timated volume, the weight of the mountain destroyed was found to be 2,826,290 million kilogrammes or 2J8'2 million tons. Volcanic products of Bandai-san. The volcanic rocks that compose the mass of Bandai-san, are Volcanic rocks, comparatively of uniform character, and belong to that class of an- désite which is of wide occurrence in Japan, viz., the Pyroxene- andesite which is now to be briefly described. This rock under its various modifications, may be seen piled up in layers in the side-wall of the newly opened crater, alternating with accumulations of loose products ejected from the volcanic vent in former times, such as pumice, scoria\ fragments of obsidian, Occ. Prof. B. Kotô considers tha£ there are six of these principal layers. The rock-layers which doubtless consolidated from lava-flows, are divisible into two main types; the one being lighter coloured, and the other darker coloured, evidently more basic than the former. The first kind is well ob- served as a great band on the eastern side- wall down Kushiini-mine, more than 10 metres thick, overlaid by reddish coloured loose layers. Lower in position and separated by layers of agglomerate, is found the darker variety. The fragments that compose the agglomerate are also usually of the latter kind. These rocks have, however, essentially of identical mineral composition, and are probably to be considered * Prof. J. Milne in calculating the weight of Japanese volcanoes has assumed the average density to be 2-5. (The Volcanoes of Japan — Trans: Seisin. Soc: Japan Vol. IX, part II 1886). Prof. T. C. Mendenhall, formerly of the University of Tokyo, in determining the force of gravity at the summit of Fujiyama, took the density of that mountain as 2-12, which was the mean of the densities of pulverized and porous rocks. (On pendulum experiments mi i he summit of Fujiyama for the purpose of ascertaining the force of gravity at that point — Trans: Seism. Soc. Vol. II, 1880). In the authors' Preliminary Report of the Bandai-san Eruption published in the Official Gazette of September 27th, 18S8, the value of the density was taken a little higher, but it was since altered to the present figure. 150 S. SËKlY AND Y. KIKUCHI the different nicies of the same magma that supplied the materials for their formation. The microscopical examination of these rocks shows that the ground-mass is microcrystalline, with a very little of colourless glass- basis in the lighter coloured variety between the microlithic plagio- clase, while in the darker variety a brown coloured glass is found more abundantly. Numerous magnetite crystals and grains are in both cases always scattered within the general mass, and as enclosures. The micro- porphyritic mineral components are Plagioclase, and Pyroxenes, which are represented by the monoclinic and the rhombic. The most frequent accessory component is Apatite. Among second- ary minerals of \es;i frequent occurrence may be mentioned Tridymite and Iron-pyrites. The principal characteristics of the porphyritic components are here given : — ♦ Play ioclase.— The porphyritic crystals are found generally in lathe-shaped out- line, having characteristic twin-lamellae of the albite-type. The extinction-angle of the plagioclase examined in the sections exhibiting these twin-lamellae varies to a considerable extent ; the range being from 2Ü3 to 30°. In general character the plagioclase is quite fresh and transparent, often with numerous glass-enclo- sures, which sometimes till up the entire space of the crystal and are arranged in distinct zones. In polarized light, the phenomenon of zonal structure is often very typical ; the extinction angles differing in the inner and the outer zones, thus show- ing a difference in the chemical composition of these layers. The specific gravity of the plagioclase as determined by Thonlet's solution gave as a mean 2-ÖS6. These characteristics indicate that the chemical composition should approximate to that of Labradorite. Sanidine. — Although this mineral under the microscope is so difficult of detect- ion, we are justified in claiming its existence, since we observed cases in which the basal cleavage face of a glassy felspar devoid of twiu-lamellaa, exhibited straight extinction. The following chemical analysis of the 'feldspathic' components isolated from a THE ERUPTION OF BAXDAT-SAW 151 rock of Öbandai, has boon made in the laboratory of the Geological Survey Depart- ment. Somehow or other, the amount of foreign ingredients is so large that wo cannot tell from it the true nature of its composition. Si 0, 6126 % Al. 03 1955 Fe.,03 3-3<; FeO 4-06 CaO 8-20 MgO 2-54 Na.>0 3-42 K20 1-22 H,0 1-77 100-38 .4 ugite. — This mineral is found in well-defined forms or sometimes in grains. The common type of a twin witli the face gcPx> as a twinning face, is frequently met with. It is usually greenish-yellow in colour, showing a feeble dichroism. It is usually fresh, with glass enclosures, and occasionally small needles of apatite. Hypersthene. — The occurrence of the rhombic form of Pyroxene has received considerable attention in recent years. This mineral winch lias hitherto been re- garded as rare, was found to be of wider occurrence than we had expected. v The rhombic pyroxene generally appears in more slender sections than the augite ; breadth to length being as 1:3 to 1:5. Under polarized light the inter- ference-colour is weak, and always shows a straight extinction. The pleochroism is very marked; ||' y = light green, JL y = greenish brown. On examining the macropinacoidal section, under a convergent polarized ray, we can often very distinctly observe a biaxial interference-figure, which, however, on account of the thinness of the section appears elliptical. The dark cross which traverses the middle of the figure under crossed niçois, passes into an hyperbola as we rotate the section. The rhombic pyroxene is often found in a cross shaped twin, the vertical axes of the two individuals making an angle of 00° with each other. This twin is there- * An interesting Japanese occuri'enee of this mineral is in the Bonin Islands. Sei' the preceding article in this Volume. — Y. 7v. Iôi; S. SEKIYA AXD Y. KTKUCHI fore that which has often heen met with in other localities, the twinning plane being Pöb . Parallel-intergrowth of the rhombic pyroxene with augite is , also found. This has been dwelt on at length in the description of the Bonin specimens." Though in specimens from Bandai this is not so clearly defined, yet we have found some cases in which this phenomenon was very characteristically developed, the crystal of the rhombic pyroxene being surrounded on both sides, or flanked on one side, by an augite band. The rhombic pyroxene isolated from the rock by means of Thoulet's solution and then lightly washed with hydrofluoric acid, was analyzed by Mr. T. Shimizu. It was impossible to separate the rhombic pyroxene from the augite. The follow- ing analysis is therefore that of a mixture. SiO, 51-80 ° , A1.,03 trace Fe203 1-89 FeO 18-8G Mn30, 103 CaO 7-0(1 MffO 18-84 100-88 This leads to the composition nearly ecmivalent to 2 Fe SiO, + BMg Si (Vf- Ca Si O,. Äfagnetite. — It is very abundant in grains, and as enclosures, especially in the pyroxenes. Apatite— The crystals of this mineral sometimes occurs in a very character- istic form. It is dichroic; || c = brownish, and J- c — yellowish. The crystals usually with fine longitudinal striae, and with transversal cleavage -fissures. This kind of apatite is very abundant in the lighter coloured rock which was found in the bottom of the conical hole at Mine-yama. Tridymite. — This mineral is seldom found in the fissures in microscopic form. In the rock imbedded within the hole at Mine-yama just referred to, it was found in well-defined crystals visible to the naked eye, nearly -~)—l mm. in size, lining the cavity of the rock in the manner of a druse. Some of them were found in * 1. c. p. 78. THE ERUPTION OF BAN DAT-SAX. 153 characteristic twins. When taken out of the cavity and examined closely the crystals were found in hexagonal plates, in combination with the faces (when re- ferred to the hexagonal axes) oP, ocP, P, œP? , flattened along the basé. Under crossed niçois, however, the plate was seen to consist of innumerable patches crossing at an angle of nearly 120°, and showing oblique directions of extinction. The specific gravity of the mineral determined by swimming it in the Thoulet's solution was 2272. For the chemical analyses, given below, of the Bandai rocks and ashes we are indebted to Prof. T. Wada, the Director of the (Geo- logical Survey. They are the result of a very careful analysis by Mr. T. Shimizu. I. II. Ill SiOo 59-66 % .... ALO, 15-51 Fe20, 3-76 FeO 5-40 Mn, 0, 1-40 CaO 6*56 MgO 3-67 Xa2U 2-50 K,(> 1-08 S -59 WA)-, -IS Loss bv ignition — -44 1'35 II. 59-56 " 16-10 6-28 3-02 1-80 6-32 3-08 3-09 •80 •18 •44 59-47 % 17-12 2-33 5-69 7-21 4-04 9-93 L00-31 L00-67 99-77 Insoluble portion in HCl. Insoluble portion in HCl. = 75-34 % = 85-34 % 154 S. SEKTYA AND Y. TCTKUOHT T. is a greenish-black rock taken from the rugged cliff of the partially destroyed wall of Yugeta-yama, containing a small quantity of iron-pyrites. II. is a reddish coloured rock, very frequently met with within the débris, especially in the conical mounds, somewhat powdery at the surface, due to the fmnarolie action to which it was subjected within the volcano. There is more sesquioxide of iron in this rock than in the others. The sample for the analysis was obtained from the ejected masses near Tokoro-sawa, a ravine on the eastern side of Bandai. III. is the analysis of a rock from Bandai given by- S. Nishiya- ma in his report to the Geological Survey, in 1887. The exact locality of this rock is not mentioned. From these results it will be seen that the rocks of Bandai are nearly identical in their composition. We may, nevertheless, distin- guish two types as we have already staled. ( )ne is a lighter coloured rock, the structure being usually porphyrinic ; the porphyritic com- ponents being pyroxenes and microtine plagioclase within a greyish coloured ground-mass. A typical example of the first type may be seen exposed on the side- wall of the new crater, as an extensive sheet in the o-reat cliff forming now the western side of Kushigamine. A similar kind ot rock was also found near the summit of Obandai. Microscopically examined the greyish white groundmass is micro- felspathic in character, with a small amount of colourless glass-basis. All the mineral components are quite fresh. The plagioclase of this rock was found to have a. specific gravity of 2'682. The other type of the pyroxene-andesife is a darker coloured rock evidently more basic than the first and resembling Basalt in its outer appearance; the glassy plagioclase being interspersed within the THE ERUPTION OF BANDAI-SAN. 155 dark, often compact, and resinous ground-mass, thus presenting a marked porphyritic structure. Besides occurring in the form of a sheet of solid rock, it is also abundant among1 the fragments which compose the agglomerate, being, in this case, often «coriaceous in appearance. Under the microscope, the ground-mass is micro- crystalline, with a more or less brownish coloured glass-basis. In the scoriaceous rock just referred to, the hypersthene crystal is often highly pleochroic. The specific gravity of the plagioclase determined with Thoulet's solution is found to be little higher than that contained in the first type, being on an average 2* 691. To this type also belongs the rock which is exposed in a rugged cliff of the half destroyed Yugeta-yama, on the western edge of the new crater, and the analysis of which is given above (I). In this rock there is a very light brownish coloured glass-basis among the microlithic ground-mass. The black colour of the rock is, however, in this ease, also due to another cause, viz., to the alteration which the crystals of pyroxenes have undergone. These crystals are seen to have altered into a peculiar dirty green coloured fibrous substance like viridite, the alteration proceeding from the fissures or from the periphery and often forming a pseudomorph. The altered product, diffusing itself into the surrounding ground-mass, imparts a peculiar dark greenish hue to the appearance of the rock. The alteration also often results in the formation of a reddish coloured iron-oxide. Some- times the crevices of the rock are filled with iron-pyrites. It is highly probable that these phenomena have been greatly due to the action of fumaroles and hotsprings, the rock being found in close proximity to the Kaminoyu spa. The following analysis of the white coloured rock, already referred to (p. 113), found near to the steam-fissures, together with that 156 S. SÈKIYA AND Y. KIKUCHI of the water which has accumulated within the crater-basin from the condensing vapours, appears to show that corrosive gases have been active within the volcano, decomposing and altering the rocks. These acid gases have no doubt affected the rocks, by combining with bases to form soluble salts, which are now dissolved in the waters of the new lake within the crater, leaving as residue the white coloured sintery material, which consists almost wholly of silica. Microscopically examined this white rock is seen to have been entirely altered in substance, but the original form of the por- phyritic crystals is more or less preserved. A completely isotropic colourless substance is found filling up the spaces of all the crystal- sections, which nevertheless exhibit the characteristics of the original minerals. The decomposition of the andesitic rock by the action of acid gases had doubtless the effect of decomposing the mineral matters ; the silica first separating out in a gelatinous form, and then consolidating to a porodine-amorphous state. Thus the isotropic substance mentioned above seems to be silica in the form of opal. The tissures of these crystals are often filled with crystals of sulphur which may frequently be detected by the naked eye as well as under the microscope. Under crossed niçois it may be recognised by the fact that certain faint yellow coloured portions within the section exhibit vivid interference-colours, while the general ground is entirely dark. The crystals of plagioclase sometimes retain their original substance, which the pyroxene crystals never do. In the latter case it is interesting to observe that the alteration of the original substance into a fibrous viridite-like product had began before it was completely decomposed into its present state, as we can still detect in it the characteristic trace of this alteration in the manner already described. THE ERUPTION OF BASDAI-SAN. 157 The analysis of this white rock found within the crater gave : Si02 91-66 % A1A 2-88 FeA 1-20 CaO -36 MgO -10 S -50 Loss by ignition 3*00 99-70 The following is the analysis of the water of the lake at the crater-basin, by Mr. M. Hida, 1,000,000 parts of water containing these substances : — XaCl ] 54*07 Na2S04 34-46 K2S04 30-00 CaSO, 1136-57 MgS04 286-09 MgCOi 178-58 Si02 86-40 A1A + Fe203 ... trace ILS 1906-17 Total evaporation-residue = 1947*40 The general description of the volcanic ash or dust that fell Volcani during the eruption has already been given (p. 111). It has a bluish- 158 S. SEKIYA AND Y. KIKUCHI grey colour, is usually very fine grained, but .sometimes mixed with coarser rock- fragments. The microscopic examination of the dust shows that it is essentially made up of the same mineral components as the Pyroxene-andesite already described, proving that it was deriv- ed by the mechanical trituration of this rock. It consists of minute particles of the microfelsitic groundmass, mixed with crystal-frag- ments of Plagioclase, Sanidine, Augite, Hypersthene, Magnetite and Apatite needles with a very small amount of glass. A specimen of the dust was brought from the town of Miliar u, 38 kilometres to the east of Bandai-san. It is essentially the same as that which fell in the immediate neighbourhood of the volcano, only that it is somewhat liner grained. In mineralogical composi- tion also, it is almost exactly similar, being chiefly made up of the finer particles of the rock, the crystal-fragments of plagioclase, the pyroxenes, and magnetite. These ashes, being comparatively heavy, do not seem to have fallen to any very great distance, thus differing from the fleecy pumiceous materials produced from molten lava at the time of other volcanic eruptions. The following chemical analysis (I.) of the ash obtained by Mr. S. Otsuka of the Geological Survey, from Hikage in Biwa-sawa, when compared with those of the andésite rocks already given, will show how close is the agreement. This ash, when treated with a mixture of hydrochloric acid, mixed with nitric acid in order to dissolve out the sulphur, gave a residue amounting to 52*92 °/0 . This, when analyzed, gave the result as in column II. THE ERUPTTOX OF BANDAI-SAN. 159 I. IT. III. Insoluble portion. Soluble portion, (by difference.) Si02 59-70% 40-11 19-59 A1A 16-68 6-75 9-93 FeA 5-43 1-44 3-99 FeO 2-05 2-05 Mn30, '98 trace -98 CaO 5-20 1-75 3-45 MgO 2-35 1-08 1-27 Xn20 2-67 1-25 1-42 K20 -99 -71 -28 S 2-25 2-25 SO;i -95 -95 P205 -15 -15 Loss by ignition .... '90 -90 100-30 53*09 The ash contained some soluble matters which on being extracted with cold water gave traces of lime, chloride, and sulphate. The analysis of another, and a somewhat more stony sample of the ash was made by Mr. H. Yoshida of the Imperial University, with the following result : — Si Oo 61-82 % FeA; + A1A.... 28-10 CaO 5-73 MgO -79 K20 1-10 Na20 2-42 99-96 160 S. SEKIYA AND Y. KIKUCHI Reports. Under this heading, we arrange in tabular form the answers to inquiries made to school-masters and others living in the provinces near Bandai-san. We have already referred to the particulars obtained in this way, and the great help we derived in the preparation of our paper from the answers so received. In the Table when the phenomena under consideration were re- ported not to have been observed, the space is marked with a dash ; when it was not known whether certain phenomena occurred or not the corresponding space is left blank. Unless otherwise stated, the date refers to the 15th of July, 1888 — the day of the eruption. 1 Shaku is nearly equivalent to 0\°>03 metres or one foot. 1 Sun = Y10 of 1 Shaku. 1 Bu = yi00 of 1 Shaku. 1 Ri = 3-93 kilometres or 2'44 miles. BEPORTS. 162 S. SEKIYA AND Y. KIKUCHI Name & Address op Observers. Direction and Distance prom Bandai-san. No. 1. T. Uda, Kokai village, Yama-köri, Prov. Iwashiro. No. 2. U. Hayakawa, Sasakawa village, Yama-köri, Prov. Iwasliiro. No. 3. S. Yamamoto Kitagata, Yama-köri, Prov. Iwashiro. E. S. E. kilometres. 32 miles. \Y. 30 kilometres. 18 miles. W. N. W. Dr7 kilometres. 9-8 miles. Sound. A black column of smoke ascended high into the air ; the sound of the explosion was something terrific, lasting for about 30 minutes. Ashes. A rumbling sound was heard by some. Sound like distant thunder. Ashes fell in abundance in the neighbourhood of the mountain for about an hour and it was not until about 4 o'clock p.m. that it entirely ceased. I was bathed by ashes. The thickness was 1 shaku near Shirakijö, and 6 sun near Hi- nokuchi. The ashes were pow- dery, but occasionally mixed with rocky fragments which struck upon our bodies. No ashes fell. THE ERUPTION OF BAXDAI-SAX. 163 Earthquakes. No earthquakes on pre- vious days. A few minutes before the eruption curious rumblings were heard, and while we were wondering what was the matter, a violent shak- ing of the earth occurred about 8 times. At the end of the 3rd shock, a black column of smoke ascended from Bandai- san. The character of this earthquake was different from that usually experienced, heavy up and down move- ments predominating. No earthquake on previous days. On the 15th a slight shock of very short duration was felt. It passed unheeded by most people. A slight shock at 4 p.m. on the 12th. Meteorological Conditions. Very fine. No cloud streak- ed the heaven. It was how- ever somewhat hotter than usual. No wind before the eruption. Soon after the erup- tion, a great whirling wind suddenly swept over the east- ern part of the mountain with great violence, destroying Shi- butani, Slhrokijo, Ojigakura, etc. Very fine. Thermometer SS3 F.; atmosphere calm. At the time of the eruption a gentle breeze blew from the west. Very fine ; calm in the morning ; it was little hotter in the afternoon, when the thermometer stood at SB P., and the northern wind began to blow. Miscellaneous Notes. Some people say that they have seen |fire on the mountain ; I saw two flashes of lightning from the rising steam. At Oda village more than 2 ri away on the easfern side of the mountain, a spring became as warm as hot bathing water while the ashes were still falling ; but at about 4 o'clock p.m. it became cold again. This is evidently due to the effect of heat imparted by falling ashes, which often scalded the naked parts of the body. The ashes also had the smell of sulphurous acid. Thunder claps were heard, followed by heavy rain continu- ing 3 or 4 minutes, which was also warm having the temperature of at least 70-80° Fah. Two travellers who were at the time of the outburst, resting at a pass of a neighbouring ridge, observed a black cloud rising from behind Bandai, among which were occasionally seen reddish co- loured objects. The phenomena lasted for a few minutes; after this a white column of steam was seen ascending from these places ; at the same time Kobandai en- tirely vanished from view. Black columns of smoke were seen rising to a great height; neither lightning nor fire were seen. There was observed also a light reddish coloured cloud sp- reading at right angles to the columns of the smoke. 161 S. SEKIYA AXt) Y. K1KTJCHI Name & Address op Observers. No. 4. T. Kami zu ma, Taj i ma, Aizu-köri, Prov. Iwasliiro. No. 5. H. Kotö, Kawaguchi, Önuma-köri, Prov. Iwasliiro. No. 0. N. Yaniasaki and T. Mafune, Fukuyoshi, Asaka-köri, Prov. Iwasliiro. No. 7. S. Sugeno, Harimiclii, Adachi-köri, Prov. Iwasliiro. No. 8. T. Katô, Naganuma, Iwase-kori, Prov. Iwasliiro. Direction and Distance from Bandai-san. s. w. s. 473 kilometres. 29-3 miles. w. s. w. 47*1 kilometres. 29-3 miles. S. 21*6 kilometres. 13-4 miles. E. 45"2 kilometres. 28 miles. S. E. S. 35"3 kilometres. 22 miles. Sound. Bumbling sound heard twice. It was, however, so slight that it was unnoticed by many. At 8h. 40 min., and 9 h. 15 min. a.m., sound was heard. The detonation of the ex- plosion was very intense, con- tinuing for nearly 5 minutes. It was also heard for several hours afterwards. People regarded the rumbling as a premonitory sign of the earthquake. People inside the houses did not notice the sound. Peasants who were working in the field heard it. It continued for nearly 5 min- utes. The screech of the pheasant which is generally heard during an earthepaake, was also noticed. Low rumblings noticed. Ashes. No ashes. No ashes. No ashes. THE ERUPTION OF BANDAI-SAN. 165 Earthquakes. About 5 days before the eruption somewhat strong earthquakes occurred twice. A slight shock at about 7 a.m. on the day of the explosion. At about 0 h. 30 m. a.m. on the 11th of this month, an earthquake was felt. Meteorological Conditions. No earthquake on previous days. At about 8 a.m. a very severe shaking experienced continuing for 5 minutes. It was very violent and the people felt as if the ground was suddenly upheaved. Nobody felt any shock. At 7 a.m. a slight shock, before we heard the rumbling noises. Very fine ; some patches of cloud seen in the eastern sky. \\ estera breeze at the time of the eruption. Very fine before the erup- tion ; the wind was northerly ; after the eruption it appeared as if it had changed to N.W. Very fine, and no cloud, calm, temperature reaching to 90° F. in sunshine. N.W. wind at the time of the out- burst. Very fine ; calm. Weather somewhat changed in the af- ternoon. Miscellaneous Notes. No lightning or smoke seen ; no change in rivers, springs, wells, etc. From the ascending columns light- ning was emitted. The smoke soon covered the top of Bandai. N.W. wind blowing at the time ; ashes fell at the village of Yokohama, nearly 2 ri N.E. from this place, coating the vegetation with greyish powder. While the heaven was clear, a very peculiar cloud (which I afterwards learned to have been steam) appeared near Adatara-san. There was neither lightning nor fire. A black streak of cloud appeared, with beautiful stripes of purple, red, yellow, and green, probably caused by the reflection and refraction of sunlight falling upon steam. 166 S. SEKIYA AND Y. KIKUCHI Name & Address OF Observers. No. 9. Motomiya, Adachi-köii, Prov. Iwaslnro. Direction and Distance from Bandai-san. No. 10. Wakamatsu, Aizu-köri, Prov. Iwaslnro. No. 11. Fukusliima, Shinobu-kôri, Prov. Iwaslnro. No. 12. K. Mita, Onoshinmachi, Tantura-köri, Prov Iwaki. E. S. E. 29-5 kilometres. 18-3 miles. S.W. 157 kilometres. 9-8 miles. E. N.E. 373 kilometres. 23-2 miles. E. S. E. 71-8 kilometres. 1U-2 miles. Sound. At about 7 b 30 minutes a.m., a roaring sound. Thundering noises were beard accompanied by the as- cent of black smoke. Distant sound was beard. Ashes. On the north-western mount- ains, a peculiar cloud of sharply defined form appear- ed, which gradually spread out in a circular form, like an um- brclla, at the same time becom- ing lighter in colour. As the cloud spread wider, grey col- oured ashes of the size of mil- let grains began to fall at about 8 a.m. This was fol- lowed by the fall of powdery ashes ; all the vegetation wore a grey colour. No ashes. At about 8 a.m., the sound The fall of ashes began at was heard thrice at short in- about 9 a.m. and continued tervals by those who were ! till about 12. outside of bouses. THE ERUPTfON* OF BANDAI-SAN. 167 Earthquakes. Strong shock was felt. Meteorological Conditions. Miscellaneous Notes. There was a smart shock shortly hefore the eruption, then it was followed by violent Leavings of the ground. The houses were observed to sway. A moderate earthquake. Ascent of smoke observed. The umbrella-shaped cloud disappeared at about 10 a.m. Fine weather, very gentle breeze. Smoke ascended very high. A shock of short duration on the day before the erup- tion. Very fine ; 82' F. at noon ; somewhat cool at 7-8 a.m. On account of ashes the atmosphere became very dim and heavy ; in the direction of Bandai a thick cloud ap- peared. IfiS S. SEKIY.A AXD Y. KI EUCH I Name & Address op Observers. No. 13. M. Murata, Miliar a, Tamura- köri, Prov. Iwaki. No. 14. Taira, Iwamai- Icöri, Prov. Iwaki. No. 15. N. Oishi, Kohama village, Naraha- kori, Prov. Iwaki. No. IG. F. Kurosawa, Shimo-Ishii village, Shirakawa-köri, Prov. Iwaki. Direction and Distance prom Bandai-san. E. S.E. 38 kilometres. 24 miles. S. E. 80-4 kilometres. 53-7 miles. E. S. E. 86*4 kilometres. 53-7 miles. S.S. E. 78*5 kilometres. 48-8 miles. Sound. Tolerably loud sound was heard. Peculiar sounds were heard by some continuing for about 2minutes. Ashes. Noises heard for about 3 minutes. Ashes began to fall from about 9 h. 30 min. a.m., and continued until 2 h. 30 min. p.m. During this interval the atmosphere had a misty ap- pearance; the film of the ashes turned vegetable and other objects into greyish hue. Ashes fell at one time so thickly as to fill the eyes and nostrils of the passers-by. The street was almost impassable. [A sample of the ash was sent] . At about 10 a.m. finely powdered ashes fell, coating the vegetation and roofs. A greenish white coloured powder or ash fell forming a thin coating over mulberry leaves, &c. In the village of Kawauchi, 5 ri west of this place, the ashes fell in the form of lumps about the size of a pea. THE ERUPTION OF IUXDA1-SA.V. 169 Earthquakes. Earthquake occurred before the eruption, continuing for about 20 seconds. It felt as if something had fallen in the next room. In general char- acter it differed from the shocks usually felt. Feeble shocks at about 8 a.m. on the 14th, and at about 4 p.m. on the 15th. Meteorological Conditions. Miscellaneous Notes. Very fine ; gentle N.W. breeze, 85° F.; it was a little stronger after the eruption. Very thick mist in the morning, but gradually began to clear away from about 8 a.m. It was quite fine at noon. Thermometer 75° F., at G a.m., 90° F. at 12 a.m. About 8 a.m. W. wind pre- vailed but afterwards turned to S. At 11 a.m. it became S.E. Very fine and calm early in the morning. At about 9 a.m. a black cloud appeared on the west, and became dark, but it cleared again in the evening. Therm. 85D F. Very fine ; temp. 95° F. at noon ; a gentle north wind. A dark coloured cloud seen in the N.W., gradually spreading as it ascen- ded. At about 10 h. 10 min. a.m. that part of the heavens became so thickly covered with misty cloud that we could not see the mountain for 20 min. The thick smoke rising from Bandai- san, which was generally regarded as cloud, looked black and somewhat red- dish. During sun-set sparkling rays of red light were emitted from the cloud. In the direction of Bandai, cloud of an elliptical form was seen. 170 S. SEKIYA AND Y. KIKUCHT Name & Address op Observers. Direction and Distance prom Bandai-san. Sound. Ashes. No. 17. N. E. 80-5 kilometres. 50 miles. Watari, Watari- köri, Prov. Iwaki. No. 18. S. Chika & E. Ashi- kawa, Yonesawa, Okitama-köri, Prov. Uzen. N. 33-4 kilometres. 20-7 miles. At about 8 a.m. thundering noises were heard in the south- ern mountains. No ashes. No. 19. W. 53 kilometres. 33-2 miles. T. Kusaka, Shikase village, Higashi- Eanbara-köri, Pvov. Ecliigo. No. 20. W. S. W. 1G4 kilometres. 102 miles. It was reported that a pe- culiar detonating sound was heard on the morning of July 15th ; it was attributed to the roaring sound sometimes au- dible from Asama-yama. Takai-köri, Prov. Shinano. No. 21. K. Nakashima, Kinbokn-sau, Island Sado. W. N. W. 161 kilometres. 100 miles. While I, in company with others, was ascending Kinbo- ku-san in the island of Sado on the morning of July 15th, we heard curious dull rum- blings which we thought tobe the firing of heavy guns in the neighbouring harbour. THE ERUPTION OF BAXDAI-SAN. 171 Earthquakes. At 11 h. 50 min. a.m. a slight earthquake lasting for only about 10 seconds was felt ; motion horizontal. A feeble shock at 3 p.m. on the 14th ; at 7 h. 30 min. a.m. and at 8 h. 20 min. a.m., on the 15th feeble earth- quakes ; the former was a lit- tle stronger than the latter. At about 4 p.m. on the 14th a smart shock was felt and was soon followed by strong shakings of the ground which lasted less than 3 min. Meteorological Conditions. Cloudy, a warm wind, 78° F. at noon. Extremely fine. 90° F. at noon. W. wind. Very fine ; 85° F. at noon ; gentle E. breeze. Miscellaneous Notes. While the weather was extremely fine, we observed the rise of a peculiarly grey coloured cloud-like smoke in the southern sky over Azuma-san. After the eruption, the water of Aka- gawa diminished and became turbid. 172 Letters of inquiry were sent to the following places, but we received answers to the effect that nothing which could certainly be referred to the influence of the volcanic eruption was noticed in these districts. Localities. Sanjöj Prov. Echigo Mizuhara, Prov. Echigo Muramatsu, Prov. Echigo Nagaoka, Prov. Echigo Mito, Prov. Hidachi Takahagi, Prov. Hidachi. Noki, Prov. Shimotsuke Nikkö, Prov. Shimotsuke Karasuyarna, Prov. Shimotsuke. Tochigi, Prov. Shimotsuke. . Shiobara, Prov. Shimotsuke. .. Namiye, Prov. Iwaki. Haranoinachi, Prov. Iwaki. Ishikawa, Prov. Iwaki , Yamagata, Prov. Uzeu Iwahashi, Prov. Uzeu Seudai, Prov. Rikuzen Direction from Baudai-sau. w. W.N.W. W. w. S.S.E. S.S.E. s, s.s.w. s. s.s.w. s. E. E. S.E. E.N.E. N.W. N.B. Distance from Baudai-san. j 96-2 ) 50-8 j 76-Ü ( 47 6 j 76-6 I 47-8 j 106-0 I 65-9 j 135-5 I 84-2 ( 106-0 ) 65-9 fl-53-2 { 95-2 j 106-0 \ 65-9 j 104-1 ) 64-7 ( 133-5 \ 83-0 i 70-7 \ \:V\) X 82'5 I 51-2 j 78-5 \ 48-8 j 62-8 \ 39-0 ( 79 \ 49 j 86-4 \ 53-6 S 103 ) 63-4 kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm, miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. kilm. miles. PLATE XV. Plate XV. Map of Bandai-san district. The area devastated by the débris of rock and earth caused by the destruction of Kobandai is distinguished by the grey colour. The position and form of the crater are indicated. ^«X^ Large and small conical mounds standing1 out from the surface of the debris in immense number. ** Principal steam- fissures in the crater. {5 Hot springs. \. Marks the direction of the hurricane- blast and the area swept by it. Heights are given in metres. Jour. Sc. Coll. Vol.111. PL. XV. ^ PLATE XVI Plate XVI Fig. 1. — Distant view of Bandai-san, from N.W. side as seen from the hill ridge of Nagamine. . 7 (The rugged and the highest "• Ob^nd-ai, \pyt «»f the crater-wall. c. Klishigamine. & Marumori-yama. 1'. Kawageta-yama. fe. Mud-field. The forest in the foreground being situated on a ridge, escaped destruction by the debris. p^ # _Yiew of Numano-taira near the edge of the new crater ; the perpendicular cliff of Öbandai facing this spot. Large hollow depressions (k) are found partly filled with water (p. 124.) The ground is covered with grey coloured ashes and smaller rock fragments. For other letters refer to Fig. 1. [From sketches by Y. Kikiiclii.] Jour. Sc. Coll. Vol. Ill PL. XVI. PLATE XVII. Plate XVII. Fig. 1. — Distant view of Bandai-san as seen from the outskirt ot Inawashiro, with the village of Miné in front. n. Obandai. h. Kushigamine. c. Akahani-yama. (I. The smaller mud-stream, which ran down through the Biwa-sawa (the ravine between b and c ), descending upon, and burying a part of the village of Miné (p. 107-108). [From photograph.} Fig. 2. — Sketch taken from Biwa-sawa. (t . Obandai with rugged and precipitous wall on the northern side. b. Knshigamine. c. Akahani-yama. (1. Upper course of the mud-stream toward Miné. f. Futatsu-iwa (p. 98). [From sketch by Mr. H. Hirauchi.] Jour. Sc. Coll. Vol. Ill PL. XVII itf ßg. ? PLATE XVIII. Plate XVIII. Fig. 1. — Extensive and nearer view of mud- field of Miné. Fig. 2. — Example of large boulders carried down along with the mud-current and forming conical mound near Kawakami. Mounds of this kind were formed on the débris in great mem- bers (p. 110-111). [From photographs.] Jour. Sc. Coll. Vol. HI PL. XVIII. I^R ^ɧ#wll '*&& ]^-*tl_ ^M Êlà'&^^^ '^ :sÄSß^^ir PLATE XIX. Plate XIX. A rock mound prominently standing out in the inside of the crater, that formed one of the stations of the survey. Encircling crater wall on the background; Kushigamine on the left with its fractured side exposing volcanic strata; on the right a part of the rugged cliff seen through the stream. The oiuantic block on the right foreground is one of the many boulders scattered about in the crater. Jour. Sc. Coll. Vol. Ill PL. XIX. Prom photograph taken by Prof W K Burton PLATE XX. Plate XX. Mud stream in Nagase Valley below Kawakami facing S. E. The débris of rock and earth" descending in a swiff torrent left wave- like traces on the side of the hill (p. 108), and tilled up the valley perhaps not less than 40 metres deep. Jour. Sc. Coll. Vol. Ill PL XX. M 'i M ,WÈ i\> 'iiwMi. ml m Et •« \ »"iff; k ( mm iîfl ; ■'' ' '/ lot a , Q 180 146 146 0 - 270 + 194-4 + 192-8 + 1-6 225 183-5 184-1 — -6 180 1 64-3 1 73-6 9-3 135 137-3 150-7 13-4 90 98 1228 23-8 45 40 84-9 44-9 o — 20-4 291 55 '5 + 45 81-5 - 37-8 43-7 90 119 97-3 21-7 135 154-1 138-3 15-8 180 171-4 102-8 8-6 225 187-2 178-9 + 1-7 270 191-6 191-6 0 178 C. G. KNOTT Table I • (Continued.) Total Twist. Longitudinal Intensity. Magnetic Lag. ' - 360° + 210.8 + 211-4 - -6 270 191-0 206-2 15-2 180 138-6 187-2 48-6 90 41-5 153-8 112-3 0 - 73-8 88-0 161-8 + 90 151-3 - 31-0 120-3 180 187-9 1 33-9 5-4 270 198-1 190-3 7-8 360 213-3 2133 0 - 8vr ] + 244.6 + 190-7 + 51-9 7 213-9 192-8 21-1 6 144-2 191-3 - 47-1 5 53-9 190-0 136-1 4 - 12-7 186-6 199-3 3 65-7 182-9 248-6 2 111-6 174-5 286-1 1 147-3 163-7 311 0 172-1 147-6 319-7 + 1 189-7 122-5 312-2 2 202-1 93 295.1 3 210-5 49-6 260-1 4 215-8 0-3 216-1 5 217 - 67 150 6 218-6 143-5 75.1 7 218-6 198-4 20-2 8 219-2 219-2 0 The curves corresponding to these numbers are shown in Plate XXV Fio\ A. The arrows at the ends of the curves indicate the directions in which they are to be gone round. Thus the curve for twist ± ^ is to be gone round clock-wise. In it, as shown by the MAGNETIC EFFECTS OF TWIST. 179 algebraic sign of the numbers in the fourth column of Table I., there is true magnetic lag. That is, the magnetic change lags behind the strain which causes it. The curve is distinctly open. Passing to the twist ± 7T, we see that the magnetic lag is sometimes positive, sometimes negative, but is in all cases very small. Hence the curve can hardly be called open, the going and returning branches being. near- ly coincident. On the whole, however, as we may see by adding the " lag " numbers, the lag is still positive. This twist may be regarded as being almost exactly the critical twist at which the magnetic la«- changes sign ; for in the next curve, that for twist ± — T, the "lag" numbers are nearly all negative and markedly so. In fact for this and for higher twists the phenomenon ceases to be one of lao-gino-, but becomes really a case of what might be called magnetic "priming." That is, the magnetic change, as it were, runs ahead of the straining that causes it. In other words, the rate at which the longitudinal magnetic intensify, whether positive or negative, falls off during un- twisting from either limit to zero is greater than the rate of growth during twisting, It will be noticed that the curves for the higher twists are of the same character as the one given by Wiedemann (Annalen d. Phys. u. Chem., Bd. XXVII., Taf. III., Fig. 8.). In Figure A, as given, the curve corresponding to the twist ± 8 ir is not shown to the same scale as the others. It is represented to the right as a dotted curve, with the scale of twist diminished to one- eight and the scale of intensity diminished to two-fifths. The greater openness of the curves for the higher twists, after the critical twist has been passed, is very striking. At the same time it is to be noticed that the range of intensity reaches a practical limit at a very moder- ate twist. These peculiarities are well shown in Table II. and in Figures V* and C which illustrate it. 180 C. G. KNOTT In Table II., the first column gives the different cyclic twists, the second column the total ranges of intensity, and the third column numbers proportional to the areas of the closed cycbc curves. These areas are easily calculated from the differences given in the fourth column of Table I, and are in fact the best indicators of the nature and magnitude of the magnetic lagging or priming. Table II. Twist. Range. Area -*- m ±1 + 7T ± 2 % ± 8 7T 150-7 289-5 385-2 424-4 137-2 + 11-5 + 1-8 58-9 2(30-3 - 2804-7 Curve B shows how the range depends on the twist. The in- crease is rapid and nearly steady for the small twists ; but for higher twists than ± 2 7T, the rate of increase of range becomes very small. It should ba mentioned that this point was very fully investigated by Mr. Imagawa ; and that the conclusion just stated does not rest merely on these five particular cases. Curve C shows the march of the area of the cyclic curve with the twist. Thus the area begins positive at the low twists, indicating true magnetic lagging. For a twist a little greater than ± tt, the area changes sign, so that we have magnetic priming. For higher twists the area goes on increasing at a very rapid rate. So far as the experiments were carried there seems to be no limit to the increase of MAGNETIC EFFECTS OF TWIST. 181 this area. An incomplete cycle for ± 20 7r indicates an approximate area of 10,000 w. At these very high twists, the wire must of course be severely strained ; the great regularity of the magnetic effect is, in the circumstances, all the more remarkable. With these results before us, a very natural enquiry was as to the existence of similar peculiarities in the case of longitudinally magnetised nickel wire. Mr. Imagawa had time only for a few experi- ments, but these were enough to establish the existence of the pecu- liarity. The most complete results of these experiments are given in Table III. The first column gives the twists in ordinary degrees or in multiples of tt ; the second and third columns give, as in Table I, the going and returning measurements of the magnetic condition of the wire ; and the fourth column gives the differences between the corresponding numbers in the second and third columns. The num- bers are not given in absolute measure, but simply in terms of the scale unit.* The wire used was 0*85 millimetres in diameter and 52 centimetres in length. Thus the twists as given correspond to twists per centimetre of ± 0°'58 , ± 0°-87 , ± 1°-16 , ± l°-73 , ± 2°-32 , ±3°-47 , ±6°-94 , ±17°-3. The results are graphically shown in Plate II., the twists being taken as abscissae and the relative intensities as ordinate«. All the curves are double-looped, and are nearly symmetrica] for the higher cyclic twists. For the first five cycles, namely, ± 30° , ± 45° , ± 60° , ± 90° , ± 120° , there is true magnetic lag, the return curve at either limit always being above the other. In the curve for ± 180° , ± 2 TT , and ± 5 tt 3 howrever, the magnetic lag becomes negative. This change in algebraic sign is also shown by the signs of the numbers in column four of Table III, as will be at once seen * Mr. Imagawa has, indeed, left no record of the constants of the magnetising coil which he used, so that it is impossible to reduce the numbers to absolute measure. So far as the present discu-sion is concerned, this is, however, of no real importance. 182 CG. KNOTT Table III. (Date of Experiment. May 31st, 1887). TOTAL Twist. Longitudinal Intensity. Magnetic Lag. + 30° 256-2 255-8 + 0-4 20 256 255 + 1*0 10 255-9 256-8 - -9 0 257-2 258-4 - 1-2 10 258-4 261-6 - 3-2 20 260-9 264-2 - 3-3 30 265-7 265-7 0 + 45 273-5 274-4 - 0-9 30 2726 267-9 + 4-7 15 271-5 268-8 + 2-7 0 273-4 275-2 - 1-8 — 15 275-3 282-7 - 7-4 30 281-6 288-9 - 7-3 45 292-2 292-2 0 + 60 290-1 288 + 2-1 40 286-4 274-7 + 11-7 20 280-8 272-4 + 8-4 0 274-7 281 - 6-3 — 20 277-9 292-9 - 15 40 282-1 303-1 - 11 60 308-5 308-5 0 + 90 316-1 317 - 0-9 60 309-9 300-9 + 10 30 294.4 271 + 23-4 (I 272-6 278-2 - 5-6 — 30 278-7 305-1 - 26- 1 60 306-4 322-6 — 16-2 90 329-9 329-9 0 MAGNETIC EFFECTS OF TWIST. Table III- (Contimwd j 183 TOTAI , Twist. Longitudinal Intensity. Magnetic Lag. + 120° 333-4 333-5 — 0-1 90 327-2 319-5 + 7-7 60 312 296 + 16 30 284-6 268 + 16-6 0 259 259-6 — •6 — 30 272 288-5 — 16-5 60 300-6 317-7 — 17-1 90 325 333-4 — 8-4 120 339-2 339-2 0 + 180 339 338-1 + 0-9 135 328 329-5 — 1-5 90 297 318 — 21 45 254-5 274-6 — 20-1 0 244-5 242-2 + 2-3 — 45 278-8 251-5 + 273 90 313 298-7 + 14-3 135 332 329 + 3 180 341-5 341-5 0 + 360 345-6 344-3 + 1-3 270 310-6 344-5 339 180 214-5 340 95-5 90 247-8 324 — 76-2 0 297-4 292 + o'4 90 331 248 + 83 180 347 248-5 + 98-5 270 352 315-5 + 36-5 360 352 352 0 + 5 TT 340 344 — 4 3tt 247 346-5 — 99-5 % 334-5 344-7 — 10-2 — ■K 351 324-4 + 26 6 3tt 349 240 + 109 5 IT 347-6 347-6 (J 184 C. G. KNOTT if the numbers for ± 120° and ± 180° are compared. Thus, some- where between these two twists just named, there is a critical twist for which the wire would probably show no magnetic lag at all. It will be noticed how the curves open out as the twists are taken higher and higher. They open out much more rapidly than they appear to do in the Plate ; since, for convenience of representation, the curve for ±360° is plotted to half-scale as regards twist, and the curve for ± 5 7T (shown dotted in the diagram) to a quarter scale. Here also, as in the curves for the circularly magnetised wire, the range of variation of the mag- netic intensity increases rapidly with the twist for small twists, but comes practically to a limit just about the critical twist for which the magnetic lag vanishes. In a series of similar experiments on per- manently magnetised nickel wire, Mr. Iwagawa failed to obtain any indication of change of sign. The permanent magnetism rapidly diminished as the twistings were taken larger and larger ; the range during twisting reached a distinct maximum for a twist of ± 90°, and then rapidly fell off, so that for a twist of ± 360° it was hardly mea- surable. But throughout, the magnetic lag was always positive. I now pass to the results for iron. Some of these are given in Table IV. arranged exactly as were the numbers for nickel in Table I. It hardly seems necessary to give the curves. Enough to say that they are very smooth, very similar in general outline to the corres- ponding curves for nickel, but differ from these in being so to speak their inversion, such as would result from reflection in a plane mirror. The numbers in the fourth column are obtained by subtracting the second column from the first, so that for true lag the differences are positive. The wire used was 58 centimetres long and 0*64 millimetres in diameter. The current along the wire was 1*4 ampères. The in- tensities are given in approximately absolute electromagnetic units (C. G. S.) MAGNETIC EFFECTS OF TWIST. 185 Table IV. (Date of Experiment. March 4-7, 1887.) Total Twist. Longitudinal Intensity. Magnetic Lag. - 90° - 184 - 183 + 1 60 180 135 45 30 170 51 119 0 149 + 93 242 + 30 92 151 243 CO + 61 173 112 90 184 184 0 - 180 - 318 - 314 + 4 120 315 262 53 60 306 141 165 0 258 + 184 442 + 60 16 295 311 120 + 233 320 87 180 325 325 0 - 270 - 348 - 345 + 3 180 352 309 43 90 340 193 147 0 270 + 190 460 + 90 + 120 329 209 180 289 352 63 270 - 351 351 0 - 540 - 384 - 387 Q 360 391 386 + 5 180 307 362 — 55 0 + 224 199 - 423 + 180 359 + 314 - 45 360 382 388 + 6 540 383 383 0 - 720 - 392 - 384 + 8 540 395 386 + 9 360 391 382 + 9 180 + 215 373 — 588 0 363 332 - 695 + 180 386 95 - 481 360 392 + 311 - 81 540 392 394 + 2 720 386 386 0 186 C. G. KNOTT As in the case of nickel, so here, the magnetic lag is positive for the smaller and negative for the higher twists. The change of sign occurs, however, at much higher values of twist. In Table V. we see the relations between the twist, the range of intensity, and the area of the cyclic curve, clearly shown. Others are included than those given in Table IV. Table V. Twist. Eange. Area ~ 7T ± .45° 94 + 14-9 ± 90° 368 127 + 135° 530 247 ± 1 80° 041 353 ±225° 000 387-5 + 270ü 704 407 + 540° 779 - 513-5 ± 720° 789 - 1821 Thus, exactly as in the case of nickel, the range very soon ap- proximates to its limiting value, while the area, once the critical twist is passed, seems to grow rapidly as the twist increases. An interesting feature of the iron curves at the high twists is that the greatest and least intensities do not occur at the limiting twists. In other words, the curves have a distinct S shape with true maximum and minimum points. This feature appears first in the experiment for the twist ± 135°, for which the full numbers are not here fiven ; but it does not become distinct and undoubted at both limits of twist until the twist of ± 270° is attained. Thereafter it is an invariable feature. In seeking for an explanation of these phenomena of magnetic la"\ we must bear in mind the essential difference between experiments in which the mechanical strain is the cause of magnetic change and those in which magnetising force is the cause. Take for example the MAGNETIC EFFECTS OF TWIST. 187 well-known case, so frequently discussed, of a wire subjected to a continuously varying longitudinal field. Here in virtue of magnetic retentiveness, there is always true magnetic lag as the field is dimi- nished from its highest value. The magnetic force is, in part, simply removed. But when the wire is twisted through a large angle, the effect of untwisting the wire is no mere removal of twist, but is really a superposition of an opposite twist. Now, we know that the induced magnetism due to a given field is apparently destroyed by a reversed field of smaller value. It is this effect, rather than the effect of mere diminution of the field to zero, that is to be compared to the effect of untwisting, especially if it is untwisting from a large twist. If the twist is small, however, that is, not greatly beyond the limits of tor- sional elasticity, the untwisting will be aided by the elasticity of the wire, so that it will have something of the character of a mere undo- ing. From this point of view, then, small twistings and untwistings will be to a certain extent comparable to applying and removing mag- netising force : while large twistings and untwistings are to be com- pared rather to applying first a given magnetising force and then a small reversed magnetising force. Hence for small cyclic twistings, the magnetic lag is a true lagging effect ; but for large cyclic twistings it becomes really a "priming" effect. The fact that the limits of tor- sional elasticity for iron are much greater than the same for nickel fits in admirably with the result established above that the critical twist at which the lag changes sign is much higher for iron than it is for nickel There is, however, another and perhaps a simpler explanation of the phenomenon. It is suggested by some results of experiments of the same nature which I have carried out very recently. The experi- ments are not quite completed ; but enough has been done to show that, in the case of nickel (and in certain circumstances, iron also) 188 CG. KNOTT magnetised circularly by a current passing along it. the positive lag is eliminated if the wire is tapped or kept in a state of forced vibration. Thus it may be that the tendency of cyclic twisting is to produce negative magnetic lag ; but that the eifect is modified by the mag- netic retentiveness of the undisturbed wire, a retentiveness which is generally explained in terms of molecular friction. The two explanations suggested are not altogether rivals. It is quite possible indeed, that both may be true, being simply somewhat different ways of looking at the phenomena involved. Effect of Twist on the Magnetization of Nickel and Iron. By H. Nagaoka, Rigakushi, of tho Imperial University. Plate XXVII. The present paper is n continuation of one published some- time ago under the title 'Combined effects of torsion and longitude nal stress on the magnetization of nickel.'* The results there de- scribed were only for a single specimen of nickel, and moreover the range of twist to which the wire under examination was subjected was always the same, namely, ± 4?5 per cm. length of the twisted wire. The main facts established in thai paper were as follows : — When the nickel wire is twisted in a constant magnetizing field, the magnetization curve obtained during torsion and detorsion under- goes a gradual transformation as the amount of longitudinal stress acting on the wire is increased. Inder low tensions, the -magnetiza- tion increases on twisting", and decreases on untwisting;. The nature of the hysteresis is such that the ascending branch of the curve is always above the descending branch, and the curve is quite symmetrical on both sides of zero twisting. On increasing the longitudinal stress, the magnetization curse becomes gradually deformed, while at the same time, there is great decrease of magnetization. Eventually when the loading reaches a certain value, one end of the wire acquires * See this Tournai Vol. TT. pg. 2S3. 100 II. NAGAOKA. reversed polarity, and the curve which was double looped transforms into a single looped curve. Now the question naturally suggests itself, can this curious phenomenon of the reversal of polarity as well as other remarkable characteristics in the magnetization of nickel be still observed in wires taken from different sources, and also for different angles of twist? The objects of the experiments now tobe described are to fill in these gaps, and also to extend the investigation to the case of iron. The method of procedure was exactly the same as in the former experiments. The intensity of magnetization was measured by the direct magnetometric method from the amounts of deflection of a magnetometer mirror. The wire was placed vertically within a solenoidal coil, and its upper end was magnetic east of the magneto- meter mirror. A constant magnetic field was maintained within the solenoid ; and as the wire was subjected to the given cyclic twisting, the deflections on the magnetometer scale were noted. The twist was applied by means of the twisting apparatus described in the former paper. The effects were examined for two specimens of nickel wire obtained from different manufacturers. They were of different diameters, being 0"86 nun. and 1 mm. thick respectively. The latter was the one used in earlier experiments. Each wire studied was first annealed by gently passing it over the flame of a spirit lamp three or four times. As the object of the investigation was merely to examine the changes of magnetization, T have not thought if necessary to reduce the observed readings to absolute units; and consequently the num- bers given are simply in terms of the scale unit. The reversal of polarity and other accompanying changes are so complicated that l have examined only a few particular cases, and most of the experi- EFFECT OF TWIST OX THE MAGNETIZATION OF NICKEL AND IKON. l!Jl inents for nickel were performed under low tensions. The effects of torsion on magnetization are much simpler for iron than for niche], vet in both there exists quite a curious change in the hysteresis as will a (»pear hereafter. Effect of Twist on the Magnetization of Nickel. The effect of cyclic twisting; on the magnetization of nickel wire is itself a function of the magnetizing- field and of the range of twist- ing. and passes through a distincj gradation of changes as this range is gradually increased. For small twists these changes are marked by many peculiar and interesting characteristics; and most of my ex- periments are limited to cases in which the twist is not very large. To examine the effect of longitudinal stress on the magnetization for each range of twist and for each magnetic field would be a matter of immense labour. I have therefore examined only a few cases in which the load was varied, and have noted some of the general features. Changes of magnetization for (he Hoist of ± 0°.86. — The smallest amount of twist producing changes in magnetization, which could be measured with certainty, was about 0?9 per cm. A nickel wire, 0.43 mm. in radius and 35 cms. long, was well annealed, and placed in posi- tion, the longitudinal stress acting being only the weight of the small twisting rod and the brass wire connecting this rod to the lower ex- tremity of the nickel wire. The first experiment was performed in a field of Ö.34 (C.G.-S. electro-magnetic unit). The wire was twisted through ± 30° (0?86 per cm.) on both sides of the initial unstrained position. After the changes in magnetization became cyclic, the following readings of the deflections were taken. 1 92 II. XAGAoKA. Twist (positive). Deflections. Twist (negative). ; Deflections. 109.5 0° 112.5 0° ]() 107.4 10 1 10.4 20 110.1 20 1 20.0 30 121.1 30 132.4 20 117.9 20 128.8 10 1 13.0 JO 122.2 0 100.5 0 113.7 The examination of the curve (Fig. 1, Plate XXYJI.) will shew that on twisting the wire in the positive direction, there is a slight diminution of magnetization till the angle of twist amounts to about 10°, then the magnetization gradually increases up to 30? On un- twisting, the magnetization diminishes, but at a somewhat slower rate than it had just been increasing, so that the return curve lies above the other until the wire is nearly all untwisted. The same things happen during negative twisting and untwisting. Thus in the present instance, the course of the change of magnetization is quite different from what was obtained in my former experiments under like circumstances. There I found that when the nickel wire was twisted h 1^5 Tier cm., the magnetization rose gradually till the maximum twist was reached, but on untwisting diminished at a somewhat quicker rate than it had just been increasing, so that the return curve lay below the other.* Reversal of hysteresis. — Comparing these two experiments, we lind that there is true lairirinu' or hysteresis in magnetization when the * In my former paper, I spoke of Fig. 5 (PI. XVI.) as being exactly the reversa of Thom- son's curve for iron, but I was mistaken ; had the course of the curve been as for the twist of 0".86, it would have been so, but not otherwise. Compare also Professor Wiedemann's curve (Fig. 1) in his paper HIa(fnctischci Untei'su- chmujai (Wied. Auu, 13d. XX Vil, L8SG). EFFECT OF TWIST ON THE MAGNETIZATION OF NICKEL AND 1E.ON. lUo range oftwist is small, but that when the range is large there is a ' priming ' in magnetization rather than a lagging.* Now we are quite warranted in using the word hysteresis in both senses, just as acceleration in dynamics includes retardation. We may speak of positive and negative hysteresis; and express the phenomenon just described as a reversal of hysteresis, occurring probably at some twist intermediate to the two just named. How this reversal of hysteresis is affected by various conditions such as the quality of nickel, the strength of the magnetizing force, the amount oftwist. &c, is a very obvious matter for investigation. Kuccl in stronger fields. — With the same wire and other conditions being the same, the field was taken stronger and stronger. Very similar effects were obtained as may be seen from Fig. '2, which corresponds to a field § = 11.9. In Fig. 3, however, a change begins to shew itself. The field is now 13. gradual increase of magnetization, but it is always smaller for the same angle of twist, than it was during the operation of twisting. Thus there is hysteresis in the proper sense of the word. Further twisting in the opposite direction produces a similar succession of changes, although the curve is not quite symmetrical about the line * t was nut aware, until my own experiments were all completed, that this i>heuomeiioii liât been observed by Mr. Iinagawa. See the precediug paper, by Or. Knott. 194 H. NAGAOKA. of zero twist. The curve of magnetization for stronger field« (Fig. 5) are similar to the aboves, the range of the change of magnetization becoming1 apparently greater as the magnetizing field is increased. oil JO o o Reversal of polarity for the. twist of ± 0°,86. — The next wire used was taken from the same specimen as before. A weight of 5 kgs. (881 kg. cmr';) was strung to the end of the brass wire attached to the lower extremity of the nickel wire. After the wire was brought to the cyclic condition by repeated torsion and detorsion in a field of 0.34 unit, the following deflections were taken. Twist (positive). Deflect ions. Twist (negative). Deflections. 0 11 0 10 10 •) 10 19.5 20 - 8 20 29 30 -17 30 41 20 — 9 20 33 10 0.6 10 23 • 0 10 0 1 1 .5 These readings shew that, in the given field, the amount of lon- gitudinal stress was sufficient to produce the phenomenon of reversed polarity. The magnetization curve (Fig. G) when compared with the corresponding curve for the twist of ± 4°.5 again indicates how the changes of magnetization differ in these two cases. In former expe- riments, there was always increase of magnetization on first twisting, and decrease on untwisting, the hysteresis being negative. In the present instance, there is steady decrease of magnetization during positive twisting, and increase during untwisting. Besides, as the twist at either extremity begins to be undone, the magnetization lags behind, so that the hysteresis is positive. These experiments sufficiently prove that the reversal of polarity EFFECT OF TWIST ON THE MAGNETIZATION OF NICKEL AND IRON. 195 takes place in different specimens of nickel, and also for different amounts of twist. The curve representing the changes of magneti- zation when the polarity suffers reversal is, as already noticed, single looped, whereas we get double looped curves in cases where there is no reversal of polarity, that is, in cases in which the longitudinal stress is sufficiently weak. Hence in the present case we should ex- pect that, bv gradually increasing the magnetizing force, until the 1-uigitudinal stress of 5 kgs. weight is insufficient to cause reversal of polarity, we shall ha able to transform single looped into double looped curves. Tlv experiment when male fulfilled the expecta- tion, as may be seen from Fig. 7. Here the field is 5.0 units ; the carve has ceased to dip balow the zero line so that the polarity is not reversed. At the same time, the curve, though unsvmmetrical with respect to the line of zero twisting, turns out to be double looped. On further increasing the held the two loops become more and more symmetrical (see Fig. 8). Thus the question as to whether a given curve will be single looped or double looped seems to depend on the , existence or non-existence of the phenomenon of the reversal of pola- rity. What has been stated in my earlier paper with respect to the form of the magnetization curve for the twist of ± 4°. 5 per cm. is also applicable for that of ± 0°.86 per cm. There is, however, a remarkable difference in the change of magnetization which may be expressed by saying that there is reversal of hysteresis between these two different twists. Reversal of hysteresis for the twists of ± l°.-j and ± 1°.S. — In the next series of experiments, the twists were taken larger. A wire 40 cms. long and 1 mm. thick was set in various magnetizing fields, and twisted through a total range of ± 60° or through a twist of ± 1°.5 pea- cm. The only longitudinal stress acting was the weight of the twisting rod and the brass wire attached to the end of the 196 TT. XAGAOTCA. nickel wire. Out of the many curves obtained in this series, I give only three, those namely which correspond to fields of 2.8, 15. 3, 21.1 units respectively. The first two curves (Fig. 9 and Fig. 10) have the some characteristics as those obtained for nickel twisted through ± 0°.S6 per cm. There is still true lagging on untwisting. But the third curve (Fig. 11) for §=21.1 is not similar to the corresponding curves represented in Fig. 7 and Fig. 8. The curve in its general form is like those obtained for smaller twists, hut the nature of the hysteresis has been changed. The hysteresis which has been positive for the two former curves becomes negative for this strength of the magnetizing force. The magnetization diminishes on twisting till the extreme limit of torsion is attained, and on untwisting it gra- dually increases. If the magnetization had tended to lag behind, as was found to be the case for the smaller twist, the return curve would have been the lower. But this is no more the case : the curve on untwisting passes above that of twisting, and thus there must have been reversal of hysteresis for this strength of the magnetizing force. It is at once apparent that the strength of the magnetizing field is one of the principal factors affecting the phenomenon of the reversal of hysteresis. In order to investigate this change more thoroughly, a new wire was taken from the same bundle, and examined in fields §=16 (Fio-. 12) and «0=20.8 (Fig. 13). The former curve shows that there is decrease of magnetization up to a certain amount of twist, but on further twisting there is increase. This increase, however, is so small that it is barely sufficient to attain the original intensity even at the extreme limit of twist. On untwisting, the magnetization gradually diminishes but increases again, passing through a minimum value before the wire is completely brought back to its initial posi- tion. This peculiarity of a minimum on untwisting had never been EFFECT OF TWIST ON THE MAGNETIZATION OF NICKEL AXD [RON. 197 noticed before for the smaller twists. On increasing the field to 20.8 units, another change takes place in the magnetization curve. The points of minimum magnetization on both twisting and untwis- ting have vanished altogether, the magnetic intensity always diminish- ing in the former, while it increases in the latter. At the same time, the original double looped curve becomes contorted, und the going and returning curves cross each other. But as soon as this stage is past, the increase of magnetizing force makes the four loops of the magnetization curve collapse into two, and hereafter the curve wears the genera] aspect as shown in Fig. 11. diminution of magnetization by twisting in strong magnetizing fields. — The fact that the magnetization diminishes instead of rising when twisted in high magnetizing fields, can be explained from the nature of the curves of magnetization for different I wists. In my former paper,* it was shown that beyond a certain strength of the field, the suscep- tibility of the twisted wire diminishes mon; rapidly than that of the untwisted, so that the magnetic intensity is in general smaller for the twisted than for the untwisted. Thus we should be led to expect a diminution of magnetization on twisting, and increase on untwisting, and the curve will have the appearance as in Figs. 7, 8, and 11. But this does not throw any light on the reason for the change of sign of hysteresis between the cases represented by Figs. 10 and 11. Reversal of hysteresis occurs in lower magnetizing fields as the hoist is taken larger. — The preceding experiments for the twist of 1°.5 le>] me to suspect that the strength of the magnetizing force as well as the amount of twist formed a chief cause of the reversal of hysteresis, and is not due to the difference in nickel. To test this, a thin wire 35 cms. long was twisted through :': 1.7 per cm. The curve (Fig. * Magnetization and retentiyeness of nickel wire under combined torsional and longitudi- nal stresses. This Journal, Vol. IT. 198 H. NAGAOKA. 14) obtained in S> = 0.34 shews that the nature of hysteresis is similar to the two former results for smaller twists and is the reverse ofthat for the twist of ± 4°. 5 per em. But in the stronger field of £> == 2.4. the amount of hysteresis was greatly diminished as may be seen from the curve shewn in Fig,'. 15. For £}=5.1, the deflections were as follows: — ■ Twist (positive). Deflections. Twist (negative). Deflections. L22 {) 426 . 10 t52 10 1 II 20 is:; 20 t73 30 503 30 t97 in 515 40 51 2 50 -O.) 50 519 GO •".27 60 52 1 50 521 50 518 to 512 to 508 30 497 30 492 2" 17 1 20 4-66 10 1 II 10 435 o t2G 0 122 The curve as plotted in Fig. 16 shews what change has taken place. There is increase of magnetization on twisting till the extreme limit of torsion is attained. On untwisting there is diminution, and. for the corresponding angles of twist, the intensity of magnetization i< always less than on twisting. In other words, the magnetization has no tendency to lag behind on untwisting, hut rather the contrary. There is negative hysteresis for this particular combination of twist and magnetizing force. In fact the nature of hysteresis is similar to that for the twist of ± 4°. 5. hut reverse to any so far obtained for smaller twists in low magnetizing fields. As the magnetizing field is increased, the same order of things comes in, and the course of the EFFECT OF TWIST OX THE MAGNETIZATION" OF NICKEL AND IRON". 199 magnetization curve is opposite to what holds in weak fields for the same angle of twist, as will be seen from Fig. 17. These experiments sufficiently prove that the reversal of hysteresis not only depends on the amount of twist, hut also on the strength of the magnetizing field, the phenomenon appearing in lower and lower magnetizing fields as the twist is taken greater and greater. The above statement is confirmed by further experiments made with 1 mm. wire, the range of twist amounting to ± 'Ie. The curve (Fig. 18) for £=0.34 has the course usual for small twists, so that there is true lagging on untwisting. But when the field strength is increased to 7.3 units (see Fig. 19), the going curve lies above the returning curve, and thus the hysteresis turns out to be negative. Thus for this amount of twist, the magnetizing force was sufficient to cause the reversal. Form of the eurer when the reversal is about to take place. — One im- portant point still remains to be investigated. We must examine what form the curve takes just as the reversal of hysteresis is about to occur. 1 had the satisfaction of obtaining the desired curve by subjecting the wire to a twist of ± 2°.o per cm. under the action of the vertical component of the terrestrial magnetic force. The curve just on point of reversal (see Fig. '20) is no more double looped, but becomes more complex. Examining the changes of magnetization -, we notice that on twisting, there is always increase of magnetization till the extreme limit of twist is reached, while on untwisting, the curve, instead of going below the former path, goes above it for a certain distance but before returning to the line of zero twisting, cros- ses the course taken in going, and finally passes below it. Thus the curve is four looped. The appearance of four loo;» marks the transi- tion in the nature of hysteresis, and any further increase of the mag- netizing field produces reversal of hysteresis, the curve becoming 200 H. XAGAOKA. again double looped as in Fig. 21. ( 'urvesfor large twists. — When the range of twist is increased to 3° or 4° per em., a reversal of hysteresis by increasing the magnetizing force is not obtained. The curves are generally of the same character as those already described for the twist of 4°. 5. Even for such large twists as '.>° or 10° (see Fig. 22 and Fig. 23) the above remark hold« true, the hysteresis being such that the going always passes above the returning curve. To give detailed description on this point would be merely repeating the earlier paper, and I feel no need of entering into the discussion of the results for these twists. Effect of holst in sinnig magnetizing fields. — Another thing which calls lor remark is the behaviour of nickel in strong magnetizing fields, during cyclic twist ings. It was already noticed for the twist of ± 1°.5, how the magnetization diminishes on twisting, and increases on un- twisting and how a particular complexity in the curve precedes the transference into such a state. The same curve of transition is also observed when the amount of twist is so large that the reversal of hysteresis takes place in a low magnetizing field. This is well ex- emplified in Fig. 24, which was obtained for Jq=20.8 with a thicker wire twisted ± 2° per cm. But owing to the reversal of hysteresis, the course of the curve is different as the comparison of Figs. 13 and 24 will shew. The general character of the magnetization curve in strong magnetizing field, during cyclic changes of torsion, is shewn in Fig. 2ô, which is given, as a type of such curves. It was obtained with a wire of 0.43 mm. radius in a held of 24.7. There is constant decrease of magnetization during twisting, and increase during untwisting up to a certain angle of twist, where if reaches a maximum. It then diminishes, and returns to its former value when the wire is brought to the initial position of zero twisting. The in- tensify of magnetization is alwavs greater during untwisting than EFFECT OF TWIST ON TITE MAGNETIZATION OX NICKEL AND IRON. 201 darin"- twisting. It will be afterwards shown that the curve is analogous to that of iron subjected to large twists. Reversai of polarity for (liferent twists. — With regard to the rever- sai of polarity for different amounts of twists, another set of experi- ments was performed with a thin wire for a twist of ± 1°.7 per cm. ( )n loading the wire with a weight of 5 kgs., the longitudinal stress was sufficient to produce the reversal of polarity when placed in the field 0 = 0.34. The magnetization curve thus obtained is represented in Fi"". 26. This curve is single looped, and has the same characteristic as those obtained for the twist of ± 4°.o in weak magnetizing fields under high tensions. The hysteresis in the present case is negative, and is opposite to that for the twist of ± 0°.86. (Compare Fig. 0 and Fig. 26). The difference of hysteresis for these two twists must be due to the difference of twist. To test this more fully, a new wire was placed in the same condition as before with regard to 1 ho longitudinal stress and the strength of the magnetizing field, and subjected to two different twists of ± l°.2 and ± l.°4 respective! v. The curve obtained from the former was similar to that for ± 0°.9, while the one given by the latter agreed with that for ± Ie. 7. Thus, for the above longitudi- nal stress and ,§=0.34, the reversal of hysteresis must take place between the twists of ± 1°.2 and ± 1 .4. Tins result indicates that the hysteresis is influenced by the difference of twist although the wire is subjected to high tensions. The next point to be examined is how the double looped curve is transformed into a single loop when the opposite polarity is about to appear. This interesting point was examined by increasing the strength of the magnetizing force. For ft — 2.5 andr = ±l°.7, the opposite magnetism is completely effaced, while at the same time the curve (Fig. 27) becomes double looped. Although the loop formed during positive twisting is very thin, it is quite certain that the 202 H. XAGAOKA. curve becomes double looped simultaneously with the vanishing of the opposite polarity. The newly formed loop gradually widens as the magnetizing force is further increased, as will be seen from Fig. 28. The experiments made for the twists of ± 0.°86, ± 1°.7, ± 4°.o all show that the cyclic curve of magnetization is gradually trans- formed from a double looped to a single looped curve, when the wire begins to shew opposite polarity. Large twists applied to nickel do not seem to affect the pheno- menon of reversal of polarity. After observing the changes in mag- netization as shown in Fig. 22, I loaded the wire with 1 kgrm. weight. The longitudinal stress thus applied was already sufficient to cause the reversal of polarity and the curve as shown in Fisr. 29 was obtained. It is single looped and similar in character to those observed for the twist of ± 4°. 5. The results thus far obtained by varying the amount of twist from ± 0°.86 to ± 9° per cm. prove that the phenomenon of the reversal of polarity always takes place however the twist may vary, provided sufficient longitu- dinal stress be applied. To examine more particularly into these intricate relations between twist, longitudinal stress, and magnetization would mean an amount of labour, whir-h the importance of the subject hardly seems to merit, Effect of Twist on the Magnetization of Iron. This subject was first investigated by Sir William Thomson, and t ho experiments to be described are to a great extent merely a repeti- tion of his. The effect of twist on the magnetization of iron is not of so complex a character as it is in the case of nickel. In iron, the effect of the longitudinal stress combined with twist does not produce any reversal of polarity, nor are the changes of hysteresis so intricate as in nickel. But when the twist exceeds a certain limit, EFFECT OF TWIST ON THE MAGNETIZATION OF NICKEL AND IKON. 203 there is the same curious reversal in the hysteresis. Repetition of Thomson s experiments. — I shall first of all describe some of the curves obtained for small amounts of twist, being merely a repetition of Thomson's experiments. An iron wire 0.G6 mm. thick and 40 cms. long was well annealed by means of a spirit lamp, and arranged in the same way as the nickel wires had been. The wire was then twisted through 180° in both directions, the strength of the magnetizing field being kept constant. The curves thus obtained during cyclic changes of torsion are shown in Figs. 30 and 31. The former was obtained under weak longitudinal stress in the magnetizing field J § — 2.9. The latter is plotted from observations made in the same field, the longitudinal stress amounting to 1800 kgrms. per sq. cm. The effect thus produced by twisting iron is not so complex as for nickel, and the change in magnetization is but a small fraction of the whole. The curves obtained under various loads are nearly all similar in shape, and for the most part agree with those given by Thomson. New experiments were made by using a wire of the same length taken from the same specimen, and twisting it through ± 360° (±9° per cm.). The curves thus obtained in different magnetizing fields are plotted in Tigs. 32, 33, 34. The hysteresis in these magnetiza- tion curves is such that in the returning branch the curve is lower than in the going branch. The comparison of curves subjected to twists of ± 4°.o and ± 9° will shew that everything remains the same except in a single point. As the twist is increased, the ascending and the descending branches of the curve tend gradually to approach towards each other, although the hysteresis has the same sign in both cases. Thus the amount of hysteresis in iron is affected by different twists, while the increase of the magnetizing force seems to cause no 204 H. XAOIAOKA. essential change. If the twist be still more increased, two eases are quite possible; either the going and the returning branches of the curve Avili continually tend to approach, or the going will ultimately lie below the returning branch, so that reversal of hysteresis ^vill declare itself. Reversal of hysteresis. — To solve this question, a wire 27 ems. long and 0.6b' mm. "thick was treated in the usual way, and twisted through ± 360°. The twist thus applied was very large ami amounted io 13°"3- per cm. After a number of twistings and untwisting«, the fol- lowing reading's of deflections were taken for ti = 2A). i -"■■ . . — Twist (positive). Deflections. Twist (negative). Deflections. 0° 487.2 0° 489.8 40 481.7 J" 484.0 80 476.8 80 170.3 120 473.4 J 20 476.0 160 170.0 160 472.2 200 467.2 200 469.8 2 JO 464.1 2 10 467.0 280 402. 0 280 40 1,8 320 450.0 320 462.9 360 457.3 360 460.8 320 400.0 320 463.3 280 463.8 280 467.0 210 409.2 2 10 472.0 2 00 470.8 ] 00 478.7 1G0 487.5 120 480.6 120 497.6 1 20 499.7 80 5 00. 7 80 502.4 40 495.9 40 496.8 0 489.8 0 400.1» EFFECT OF TWIST OX THE MAGNETIZATION OF NICKEL ANi> IKON. -05 lîy examining the curve (i ig'. 35) we see that on twisting the wire, there is constant decrease of magnetization, while on untwist- ing, the magnetization rises steadily till it reaches a maximum, after which it returns to its original value at the initial position of zero twisting'. Thus in the returning branch the curve is no more lower than in the going1 branch, so that the course of the magnetiza- tion curve is opposite to that for the twists smaller than and up to ± 9°.0. As was already noticed in nickel, there is also reversal of hysteresis in iron. So far as my experiments go, the strength of the field or the amount of longitudinal stress does not seem to produce ;iii\ remarkable variation on hysteresis, as the curves (Fig. 3o*, 37, 38) will shew. It is thus established that the hysteresis in iron becomes reversed when the range of twist exceeds a certain limit. This must evidently lie between the twists of± 9° and ± 13°^ per cm. Bui ii is not easy to determine the exact critical twist, even by experimenting for twists intermediate between the two. For a twist of \'2° per cm. the difference between the going and returning branches of the curve diminishes to a certain extent as will be seen from Fig. 39. For a twist of 10. °4 (Fig. 40) the two brandies of the curve are separated by a still smaller interval ; but still the former lies below the latter, so that the hysteresis is opposite in character to that for the twist of 9° per cm. Judging from these experiments, it can safely be inferred that the curious phenomenon of the reversal of hysteresis in iron takes place when the range of twist amounts to about ± 10° per cm. Comparison of results for iron and nickel. — Comparing these results for iron with those obtained for nickel, we find some analogy between the two. As was formerly remarked, the changes in magnetiza- tion in nickel due to cyclic twistings are opposite in character to those of iron. But when the amount of twist as well as the 206 H. NAGAOKA. strength of the magnetizing force is varied, we find gradual trans- formations taking place. The magnetization curves in nickel for small twists and weak longitudinal stresses are opposite in charac- ter to those in iron for moderate twists. When the twist in nickel exceeds lc.5 per cm., and the magnetizing force is sufficiently strong, the magnetization curve loses its opposite character to that of iron under moderate twists, and acquires a form similar to that for iron subject to very large twist, as the comparison of Figs. 25 and 39 shews. Summary. — \ shall in conclusion summarise the results obtained in the present experiment. In nickel under feeble longitudinal stress, there is reversal of hysteresis when the range of twist is moderate. This phenomenon takes place in lower magnetizing fields as the twist is increased. When the amount of twist exceeds 3° per cm., nothing of this nature is observed. The reversal of polarity, on the contrary, is a phenome- non common to all twists, provided the longitudinal stress applied be sufficiently great. In iron, so far as my experiments go, the magnetizing force does not alter the character of hysteresis, but reversal of hysteresis takes place when the twist exceeds 10° per cm., and the magnetization curve bears a close resemblance to that of nickel in strong ma^netizim«' fields for moderate twists. The effects of twist in iron and nickel, and especially in the lat- ter, are so complicated that it is impossible to give anything like a true explanation coordinating all these facts. The magnetic qualities of these two metals with regard to stress are generally opposite. Not only do twist and. longitudinal stress produce opposite effects in mag- netization, but by twisting the wire in the magnetizing field we ob- tain transient currents in opposite directions for these two metals and EFFECT OF TWIST ON THF MAGNETIZATION OF NICKEL AND IRON. 207 it can scarcely be doubted that these sets of phenomena are closely linked together. The occurrence of a maximum current for moderate amounts of twist and other peculiarities will be discussed in a future communication, Oxyamidosulphonates and their Conversion into Hyponitrites. By Edward Divers, M.D., F.R.S., Prof. and Tamemasa Haga, F.C.S., Asst. Prof. Imperial University, Japan. In our paper on the Reaction between Sulphites and Nitrites of Metals other than Potassium, (J. Oh. Soc. 51, 659), we gave notice of our intention to work upon the reactions of the sodium salts. This intention we have carried out. and already find ourselves in a position to materially extend and modify the knowledge of the chemistry of the sulphazotised compounds contained in the writings of our prede- cessors in the investigation, Fremy*, Ad. Claus and Koch+, Berglundll, and Raschigï. We propose to publish our contribution to this large subject in a few short papers, such as the present one, each complete in itself. Oxyamidosulphonates, the subject of the present paper, are the sulphazidates of Fremy, the sidphydroxylamates of Claus, the hydroxyl- aminc-monosulphonatcs of Raschig. Between the one set of terms — * Ann. Chim. Phys. [3], 15, 403. t Ann. Chem. Pharm. 152, 336; 138, 52 and 10-4. || Bull. Soc. Chim, 25, 455; 29, 422; Ber. Ch. Ges. 9, 252 and 189G. X Ann. Chem. 241,161. 2\2 E. DIVERS AND T. HAGA nitrilo—, imido-, and amido-, oximido- and oxyamidosulphonate, and the other set — amine- and hydroxyamine-tri-, di-, and monosidphonate, there is not much to choose by. But as it is desirable on the score of consistency, to employ exclusively one set or the other, the already prevalent use of nitrilo-, imido-, and amidosulphonate makes it proper to use oximido- and oxyamidosulphonate rather than lujdroxijamine-di- and monosidphonate. Oxijamidosulplionie acid, known only in solution, was first pre- pared by Fremy, who found that potassium oximidosidphonate, (neutral- sulphazotatej, sooner or later decomposed into acid-sulphate and the oxyamido-salt, doing so at once when its solution was boiled. Altering his atomic weights to those now accepted, and writing em- pirically his formula"1, to which he attached no constitutional signifi- cance, his equation becomes — S5QS0K4HÄ= S208K2H6N2+2KHS0*+H2S04* Clans has shown, however, that these formulas incorrectly represent the composition of the salts, and Haschig has confirmed Claus's state- ment. The two formulas corrected and halved stand as So09lv2H5K and SOJvH.X, or H( )X(S( ),lv),(OIIÔ)2 and HONH(S03K), accord- ing to Clans. The latter formula we have further to modify slightly. To get the oxyamidosulphonic acid pure for preparing its salts, Fremy neutralised with ammonia the hvdrolysed solution of potassium oximidosulphonnte. added barium chloride, filtered oil' sulphate, and then, by adding baryta- water, precipitated a dibarium oxyamidosulpho- nate. This salt, when washed, he decomposed by adding just enough sulphuric acid fo combine with the barium, and then used the filtered solution of the new acid for combining with bases. The acid appears He recognised the peculiarity and importance of this first instance of what we now style the hydrolysis of a suiphonate into a sulphate. ON OXYAMIDOSULPHOXATES. 913 to be the only sulphoxyazotised acid possessing any stability. Claus introduced the slight modification of Fremy's process for preparing it of omitting the preliminary neutralisation with ammonia. Having discovered a second barium salt, neutral and soluble, we proceed differently in treating the barium precipitate, and thereby avoid the contamination of the salts with sulphite, always present in Fremy's barium precipitate (see the section of this paper headed, the decomposition of oxyamidosulphonates by alkaline bases). The dibarium salt is, as found by Freiny. Aery alkaline to litmus, and we add to it only enough sulphuric acid to get a neutral solution, along with barium sulphate equivalent to half the barium and such barium sul- phite insoluble as may have been present in it. The solution contains only nionobarhun oxyamidosulphonate and from it can be prepared the acid and its salts by adding its equivalent of sulphuric acid or a sulphate, to determine which, an estimation of barium is made in a portion of the solution. Raschig prepares an impure acid from Fremy's solution obtained by boiling potassium oximidosulphonate so as to hydrolyse it into potassium sulphate and oxyamidosulphonate. To do this he removes the potassium sulphate by alcohol and then concentrates the solution of the acid to a syrupy consistence. Sodium oxyanndosulphonate} prepared by Freuiy and by us, is a clear gummy liquid as thick as molasses, which exposed over sul- phuric acid (in a partial vacuum) never solidifies or shows any sign of crystallisation. It is neutral in reaction. Potassium oxyamidosulphonate, prepared and analysed by Fremv. by Claus, and by us. It occurs in six-sided plates as stated by Fremy, when crystallised from its hot solution, but the plates are more often square, while by spontaneous evaporation of its cold solution thick tables and bold prisms are obtained. Clans found the 214 E- DIVERS AND T. HAGA crystals to be anhydrous, and Fremy's analysis and formula agree with this Unding. But it is now known that Fremy's analytical results cannot be relied on, and we have already had to give an instance of this in the present paper and shall have to give others. Differing from Claus, we find all the crystals of this salt to effloresce slowly over sulphuric acid, and to give on analysis results indicating the presence of one molecule of water. Solutions show a great ten- dency to supersaturate, and it becomes often quite difficult to crys- tallise them. When thoroughly dry the crystals can be kept for months without undergoing much change, but moist they are unstable, hydrolysing and becoming acid to litmus. The acidity developed is that of hydroxyammonium sulphate, hardly showing with methyl-orange. Heated they suddenly intumesce below 100° and thoroughly decompose. To determine the sulphur and nitrogen we hydrolysed the salt in a sealed tube with hydrochloric acid at 130° C, here following Easchig's process, which gives, however, somewhat irregular results as we afterwards found (see the analysis of the dibarium salt). The hydroxyamine thus [»reduced was measured by iodine after addition of potassium acid-carbonate. Water could not be removed by ex- posure over sulphuric acid at the ordinary temperature and pressure rapidly enough to be convenient for analytical purposes, this and other sulphazotised salts retaining, according to our experience, part of their crystallisation-water with great tenacity. Nor coidd the water be well expelled in the oven, because of the decomposition of the salt at about 95°. But we made a fairly good estimation of it, by moderately heating the salt in a Sprengel-pump vacuum, in a long bulbed tube containing also sulphuric acid. The following is a table of our results— OX OXYAMIDOSÜLPHONATES. 215 honh(so3k),oh2 a h b Potassium 23.08 22.53* 23.45 Sulphur 7.58 7.34 7.69 7.56 Oxyamidogen, HOXH. 18.94 17.96 19.76 17.50 Water 10.39 9.29 Sample a was in prisms, and b in tabular crystals. We have given much consideration to Claus's apparently carefully obtained results, but are unable to öfter any explanation of their difference from ours. We have prepared the salt in winter and in summer (when he found it difficult to work), by evaporation of cold solutions and by cooling hot solutions, and have always obtained crystals which slowly effloresced in the desiccator. Dibarium oxyamidosulplionate, prepared by Fremy and by us, is a crystalline, alkaline, nearly insoluble salt. It dissolves in hydro- chloric acid, and then shows by the odour evolved, the presence of sulphite, from a trace only to even much, as an impurity. The only analytical datum given by Fremy is that the salt is formed from one equivalent of acid and two equivalents of baryta. AVe have analysed it and found for it a composition agreeing with the formula given by Fremy less the Ho by which his formulae generally exceed those now adopted. In this analysis and that of the following salt we slightly modi- lied the method of hydrolysing, so as to get uniform and higher numbers for the hydroxyamine. The modification consisted in heating for some time with hydrochloric acid only to 100", before raising the temperature to 130". Simple hydroxyammonium sul- phate may be rapidly heated with acid to 130° or even higher with- out getting low results, from which it would appear that at the moment of its formation at 130° from its sulphonic derivative, hy- * Slight loss of potassium sulphate during cooling known to have occurred. 216 E. DIVERS ÀXD T. H AG A. droxyamine is less stable than when already formed. The consti- tution of the dibarium salt is expressed by the formula — HO- -N-S03 Ba / \ Ba, OH, HO — s-so. Calculated. Found Barium 53.31. 53.13 Sulphur 12.45 12.41 Oximidoeen, HON 12.0G 12.02 Barium oxyamidosulphonate, prepared by us, in solution first, as already described in this paper, by adding just enough sulphuric acid to the dibarium salt to remove half its barium. The neutral liquor thus obtained yields by evaporation over sulphuric acid crystals of the salt. It is a very soluble salt and forms small hard brilli- ant square tabular crystals, intermixed with minute square prisms; When long kept it decomposes. Its crystals contain water. Heated nearly to 100°, it suddenly and violently decomposes into gases and barium sulphate. In analysing it the barium was determined in one case by ig- niting with sulphuric acid (a) ; in two cases the salt was slowly heated with dry sodium carbonate, whereby oxygen was absorbed from the air, after which the heat was raised to fusion of the mixture, and the barium and sulphur both determined (î>) ; while in another case, (he salt was hydrolysed by heating with hydrochloric acid, the separated barium sulphate (representing all the barium and half the sulphur of the salt) weighed, the other half of the sulphuric acid precipitated with barium chloride, and lastly the hydroxyamine ti- trated with iodine (c). The results were the following — ON OXYAMIDOSULPHOXATES. 217 (HONHS03)2Ba,OH2 a. b. I). C. Barnim 36.15 36.17 36.27 35.88 35.53 Sulphur 16.88 16.84 16.65 16.49 Oxyamidogen, 16.88 16.50 The hydrolysis of oxyamidosulphonic acid. Although oxyamidosulphonic acid possesses relative stability, the fact that its solution does decompose was fully noticed by Fremy. Rasch ig has found that the decomposition proceeds sharply and, in presence of hot acid, rapidly, according to the equation — 2HONHS03H+2H20 = (NH3OH)2SO,+ H2S04 and with this important finding we fully agree from experience, and to it have nothing to add. Fremy stated that when boiled with water the acid decomposes wholly into acid-ammonium sulphate and oxygen or hydrogen -peroxide. His finding ammonia and oxygen (or any gas) cannot be explained. He appears to have tested for hydrogen-peroxide by adding manganese binoxide which would ac- count for his finding it, since an effervescence of nitrons oxide could easily pass for one of oxygen. Clans expressed his hesitation to accept Fremy's equation (modi- fied)— 2HONH(S03H) + 2H,0 = 2(NH4)HS04+02, as quantitative, but at the same time admitted that he had also obtained (besides sulphuric acid) ammonia and oxygen (or nitrous oxide.) Raschid has got other results, as already stated, and in these both ammonia and either oxygen or nitrous oxide, are absent. These inexplicable differences have their parallel in what is contained in the next sec- tion of this paper, only there the differences announced are between ourselves and the other workers. 218 E. DIVERS AND T. HAGA In presence of potassium- acid- carbonate oxyamidosulphonates react with iodine solution like a hydroxyamine salt only very much more slowly, so that without previous hydrolysis their amount can be titrated in this way though only with difficulty. Decomposition of oxyamidosulphonates by alkaline bases. The decomposition of oxyamidosulphonates by a solution of potassium hydroxide had apparently been fairly well worked out when we came to give attention to it. Fremy had observed that when heated with excess of this reagent the potassium salt disen- 2'ao-ed ammonia, as well as oxvçren of which, as he savs, he had established the absolute purity by analysis. Hence it seemed that oxyamidosulphonates undergo the same decomposition when heated with alkali as when heated with acid. Lossen had just discovered hydroxyamine when work upon the sulphazotised bodies was taken up by Clans, and this circumstance led the latter to see in the reaction between potassium hydroxide and Fremy's sulphazidate, as observed by Fremy and by himself, most convincing evidence that the salt is constituted as a sulphonic de- rivative of hydroxyamine, and decomposes with alkali into this base and potassium sulphate. He did not, he admits, succeed indeed in isolating the hydroxyamine or any of its salts, but he found all the sulphur of the sulphazidate converted into sulphate, and just the other bodies and in just the proportions which Lossen had found hvclroxyamine to yield when heated with alkali, namely, ammonia equivalent to between a third and a half of the total nitrogen, and gases which neither extinguished nor rekindled a glowing match, wdiich were, therefore not the pure oxygen of Fremy's finding, and which might well be nitrogen mixed with nitrous oxide, as required 05 OXYAMIDOSULPHOXATKS. 9]q on the supposition piade. Added to tliis was the fact of the réd u ciné" action which the alkaline mixture exerted upon copper and silver, and the proof seemed complete. Rasehig, in his recent paper, went further in the matter than Claus and. with or without experimenting we cannot decide from hi* words, concluded that an (unhealed) alkaline solution of oxyamido- sulphonic acid is actually a solution of five hydroxyamine in the quan- tity calculated from the amount of the acid taken, and therefore just such a solution of hydroxyamine as is used and wanted for preparing aldoximes and acetoximes. Now with the exception of nitrous oxide beinçf «riven off, and that of reduction of copper and silver salts — an action to be treated of in the following section of this paper, none of the observations of these chemists have we been able to confirm. This decomposition of oxyamidosulphonates by alkali is of another and still more in- teresting character than (/laus and Rasehig conceived it to be- That the oxyamidosulphonates are hydroxyamine derivatives, which hydrolyse in acid solutions into hydroxyamine and sulphate, is in- deed certain, as ascertained by Rasehig. But, nevertheless, in alkaline solution they give neither sulphate nor hydroxyamine, nor the de- composition-products of hydroxyamine. Oxyamidosulphonates decompose with potassium hydroxide, and similar reagents, exclusively into sulphite and hyponitrite, and the decomposition-products of a hyponitrite. Xo ammonia and, so far as we could judge, no sulphate and no nitrogen — or if any only un- important quantities of nitrogen and sulphate — are formed. The difficulty of keeping a sulphite solution for days free from sulphate, and of detecting small amounts of nitrogen in presence of nitrous oxide are well-known facts sufficiently explaining any uncertainty in 920 E* DIVE:RÖ -.AX-D T. H AG A our assertions. The total absence of ammonia peremptorily forbids any admission of the generation of hydroxy amine. ;■.;-,■;.■ Cold dilute alkali or alkaline-earth hydroxide suffices to partly effect the change under consideration. Consequently every attempt to form di potassium or sodium oxyamidosulphonate corresponding with the dibarium salt has failed in our hands because of this re- solution of the salt into simpler ones on adding alkali. Also for the same reason Fremy's dibarium salt, described in this paper, although an almost insoluble salt, cannot, we find, be prepared quite free from sulphite, and when kept for any considerable time becomes charged with it and with traces of hyponitrite. To effect the complete or nearly complete conversion of these salts into sulphite and hyponitrite they may be either lefc for days in the cold with the very strongest potassium-hydroxide solution, or be heated to boiling for a short time with strong alkali. In both cases effervescence occurs, due to the decomposition of hyponitrite. The gas is not the feeble supporter of combustion met with by Claus, but so far behaves as oxygen, as Fremy had observed. Only it is not oxygen but nitrous oxide, soluble in water. The highly alkaline liquor when acidified gives abundance of sulphur dioxide, and when only neutralised gives with barium chloride a precipitate which might of course be taken for sulphate by a mind prepossessed as Claus's almost admittedly was, and which does, as is well-known, rapidly change into sulphate on the filter. A\ ^hen partially or fully neutralised with acetic acid the solution gives on treatment with sufficient silver nitrate much silver hyponitrite together with a very little reduced silver owing to the never quite complete destruction of the sul phonic salt. At first the silver nitrate is consumed in forming potassium-silver sulphite, and this consumption can be avoided if desired, either by using the barium salt instead of the potassium salt, ON OXYAMIDOSULPHONATES. 221 îor by adding barium hydroxide, and then filtering off barium sulphite before adding silver. This decomposition actually furnishes much the most productive •method of preparing hyponitrite yet discovered. The following are the results of some trials we have made, the silver hyponitrite hav- ing been purified by the authors' method (J. Ch. Soc, 45, 81J of dissolution in nitric acid and reprecipitation with sodium carbonate. Generally the silver hyponitrite was directly weighed, but in one or two cases it was converted to chloride for weiu'hino- : Digestion of 0.5772 gram of crystals of potassium oxyamido- sulphonate for twenty-four hours with a saturated solution of potassium hydroxide and a bit undissolved, still contained a very small quantity of the sulphonic salt undecomposed. But the yield of hyponitrite came up in this case to 76 %. of the full amount ; Boiling 0.9370 gram of crystals with concentrated potas- sium hydroxide for a short time was attended with copious efferves- cence of nitrous oxide, and left still a little undecomposed salt, but the yield of hyponitrite still reached 30 % of the equivalent of the salt taken. Merely to prepare hyponitrite from nitrite in this way there is no necessity of getting first a pure oxyamidosulphonate, a well-pre- pared solution of either alkali-salt sufficiently concentrated is quite serviceable if treated with solid potassium hydroxide. Working in this way we found — 0.4545 gram sodium nitrite,* the final treatment of which, after conversion to the sulphonic salt, was in the cold with the most concentrated potash for twenty-four hours, gave hyponitrite * Measured off for analysis as oxyamidosulphonate solution produced from a large quanti- ty of nitrite worked uPon. 222 E- DIVERS AXD T. HAG A amounting to 40 % of the full yield, had nil the nitrite been utilised ; 0.5833 grain .sodium nitrite,* by final cold treatment for twelve days with the potash, when still a little undecomposed salt remain- ed, gave a yield of hyponi tri te equivalent to 33 i "/„ of the nitrite; 0.417.1 gram sodium nitrite* by final first cold treatment for twenty-one hours and then at 100° for a quarter-hour yielded hyponitrite amounting to 49 ^ "/„ of the calculated quantity. But in order to get such results as these, referred to the nitrite taken, our modification of the process for getting the oximidosul- phonate must be followed, an account of which we reserve for the paper on these salts. Here we need only mention that we can get at least 85 nj<> of the calculated quantity of oximidosulphonate from nitrite, a proportion far higher than previously got by Raschig, the only quantitative worker. In consequence of the decomposition of much of the potassium hyponitrite into hydroxide and nitrous oxide, the measure of the hyponitrite does not of itself serve to prove that the formation of this salt is the onlv decomposition of the oxyamidosulphonate. But it docs make this deduction highly probable when taken along with the occurrence of so much nitrous oxide and sulphite, and with no ammonia, nitrogen, or sulphate. The determination of sulphite is, however, what seems sufficient of itself to prove the singleness of the decomposition, although here too any very close approach to the calculated amount cannot be expected, considering the ready oxidisa- bility of sulphites to sulphates, and that not quite all the sulphonate is ever decomposed. In consequence of the près. Mice of hyponitrite and its reaction with iodine, a volumetric estimation of the sulphite with iodine was not possible. We therefore availed ourselves of the * See not on p. 221. ON OXYAMTDOSULPITOXATES. 223 reaction of sulphurous acid with stannous chloride for its estimation. Stannous chloride has no action upon hydroxyamine (Divers and Haga, J. Ch. Soc. 47 624). and none upon oxyarnidosulphonic acid. Our procedure was to put in a pressure-bottle the diluted solution of the salt decomposed by alkali and neutralised, mix it with excess of stannous chloride, and almost fill the bottle with water. Some nrsenious oxide was also added in order to fix all hydrogen sulphide in presence of the necessary excess of hot acid, which was. however, not so concentrated as to lead to reduction of arsenic by the tin-solu- tion. The tightly closed bottle was kept in nearly boiling water for an hour, and then left to cool. The washed precipitate of stannous sulphide was heated with hydrochloric acid and potassium chlorate until all sulphur had oxidised, and the solution evaporated to dryness, and again evaporated with hydrochloric acid to dryness. Finally, after removing the tin by hydrogen sulphide, the sulphuric acid in the filtrate was estimated as barium salt. In this way, from 0.7470 gram of salt, which by long keeping had slightly hydrolysed, we got sulphur equivalent to 88.63 per cent, of all in the original sulphonic salt. Another sample, freshly crystallised, in fine plates, was boiled with the potassium hydroxide, entirely out of contact with air, by keeping it in an atmosphere of hydrogen. Thus treated, the product, with the tube containing it,?was dropped into the bottle of stannous chloride. In this case, 0.2007 gram gave 89 % (89.05) of the sul- phur as sulphite. These results render it clear that sulphite and therefore hyponitrite are the two and only primary products of the change. The reaction by which hyponitrite and sulphite are formed con- sists probably in the substitution of the hydrogen of the oxyamiclo radical by potassium, and then of self-decomposition of the potassium compound. There is no hydrolysis or saponification, simply dis- 224 E. DIVERS AND T. ITAGA sociation or chemical fission — HONHS03K + 2K0H = KONKS03K + 21FO 2KONKS03K = (KON)2 + 2K,S03 Raschig lias observed a closely similar decomposition of Fremy's potassium sulphazite by strong potash into sulphite and nitrite. AVe would gladly account for the differences between the results found by the other chemists and our own, but we can do little in this direction. AVe have to face the tact that Claus's work was quan- titative. The only suggestion we can offer is that Fremy and Claus's originally pure preparations were not treated with alkali until they had been kept long enough to undergo their usual decomposition (fully in Claus's case) — 2HONH(SO:,K),OH2=(HONH3)2SOI + K2SO, into hvdroxvamine and sulphate. Such a mixture would behave just as they found. As for oxygen, Fremy must have mistaken ni- trous oxide for it, and in making this supposition we have evidently the support of Clans and Raschig. Lastly, as for Claus's nitrous oxide diluted with nitrogen, dilution with air and steam may perhaps have been the cause of his not having got such a gas as Fremy and we ourselves got. Oxyamidosulphonates evaporated to dryness on a water-bath with either potassium or sodium carbonate evolves carbon dioxide1 during the last stages of the evaporation, and yields much sulphite. No hyponitrite can remain undecomposed in such circumstances. Even a solution of the oxyamidosulphonate left in the cold for a day with the carbonate shows evidence of the presence of a little sulphite. Repeated evaporations with potassium acetate leave an alkaline mixture containing a minute quantity of sulphite. OX OXYAMIDOSULPHOXATES. -;-; ZZ.) Oxidation of oxyamidosulphonates by basic reagents. Fremv observed that manganese dioxide dissolved as manganous salt in oxyamidosulphonie acid with effervescence of oxygen. Also that the same reagent caused a lively effervescence in a solution of the potassium salt. In these observations he was right save in mistaking nitrous oxide for oxygen. Finally, he found that the potassium salt immediately reduces salts of silver, copper, and gold. We must ex- cept copper from this statement, when alkali is absent Claus found that in the presence of potassium hydroxide the potassium salt reduces salts of copper and silver in the cold, just like hydroxyamine, as we have already had occasion to mention, but he made only qualitative experiments. Raschig, who holds that alkali converts oxyamido- sulphonates wholly into their equivalent of hydroxyamine, records no experimental determination of this point, though he quantitatively estimated the hydroxyamine produced by the action of an acid. The reaction which takes place in our hands is the conversion of the oxyamidosulphonate into sulphite and sulphate and the re- duction of a quantity of metal-oxide equivalent to the oxidation of the oxyamide residue, and not of hydroxyamine supposed to be pro- duced. That is to say, the cuprous oxide obtained is half what it would be were hvdroxvamine first formed as believed by Claus and Raschig. The equation, therefore, will stand thus — 2 HONH(S03K) + 2CuO + 2KOH - l\,S0;.+ KoS04 + Cu20 + X,0 + 3H,O which shows the potassium hydroxide taking the two sulphonic re- sidues to form sulphite, sulphate, * and water, and the copper-oxide oxidising to water the one hydrogen of each of the oxyamido residues, thus leaving hyponitrous acid to resolve itself finally into nitrous * Hyposulphate was searoliod for, and could nut bo found. 22 6 E- DIVERS AND T. H AHA oxide and water. After the reduction, addition of hydrochloric acid liberates much sulphur dioxide. The reaction does not quite complete itself, ceasing when the solutions become very dilute. Thus, if to a solution of one gram of the oxyamidosulphonate in a liter, only a few drops of a dilute solution of copper sulphate and then of potassium hydroxide are added, the mixture shows a permanent blue opalescence, but deposits no cuprous hydroxide, even when preserved for hours in a closed vessel. This observation may serve to show that although, wdien very dilute, alkali does not produce much sulphite and hyponitrite, this is not because hydroxyamine and sulphate are produced instead, for if such were the case the hydroxyamine would then act upon the cupric hydroxide. The fact that the alkaline solution contains not hydroxyamine but a sulphonic derivative of it, which gives sulphite in its reactions with reducible bodies, and that it has only half the action of its equivalent of hydroxyamine, are serious, if not fatal, objections to re- sorting to it ;is a reagent on organic compounds, as Kaschig has suggested may economically be done. This chemist, notwithstanding that he has pointed out (his memoir, p. 182) that the reason that oximidosulphonates do not possess any of the reducing powrer of hydroxyamine, is that in them the two active hydrogens of hy- droxyamine are replaced by sulphonic radicals, and that oxyamido- sulphonates by retaining still one of these hydrogens remain as equally easily reducible as hydroxyamine itself, has yet failed to see that, his contention being well-founded, it will be the oxyamido- sulphonate and not hydroxyamine which exerts the reducing power of its alkaline solution. That it is so, is shown by the fact, determined by us, that in the absence of reducible agents, alkalis to a small ex- tent do not decompose oxvamidosulphonates and for the rest change OX OXYAMIDOSULPHOXATES. <227 them into sulphite and hyponitrite, neither of which give« a cuprous precipitate in presence of alkalis. We have vet to supply particulars of our quantitative work. The amount of sulphite produced was imperfectly estimated unavoid- ably, partly because of the great oxidisability of the very dilute alkaline sulphite by air, and partly because of the always incomplete decomposition of the oxyamidosulphonate. To measure it, the mother-liquor of the copper precipitate was run into excess ofhalf- decinormal iodine-solution (mixed with acid enough to more than neutralise the alkali in the mother-liquor), and the unconsumed iodine titrated with sodium thiosulphate. All the water used was previously freed from air by boiling. Of the salt, 1.0967 grams, treated with copper sulphate and potassium hydroxide, gave in this way 40 % of the sulphur of the salt as sulphite, and that was our best result. Theory, as given by us, indicates 50 %, while on the other view there should be none at all. Other portions of the mother-liquor of the copper precipitate were acidified for hydrolysis of any re- maining sulphonate, and concentrated by evaporation. One of these then gave quite a distinct further reduction with the copper mixture, due to hydroxyamine, while another measured portion showed on titration with iodine in presence of potassium-acid-carbonate the pre- sence of hydroxyamine, equivalent however to only one-twelfth of the whole salt. To measure the amount of copper reduced we added to 0.2913 gram of the salt (already very slightly hydrolysed, by keeping), dis- solved in water, a sort of Fehling's solution, much stronger than usual and with much less alkali in it, in slight excess, heated to boiling to collect the cuprous oxide, filtered, rapidly washed, and weighed the reduced oxide as black oxide. We thus obtained cupric oxide equal to 48 % of the weight of the salt instead of 47 % , 22$ E. DIVERS AND T. HAGA calculated from our equation. On the other theory, twice as much should have been got. We then applied the stannous- chloride process, avoiding all ex- posure to air by treating the salt in the pressure-bottle to be after- wards used in the analysis, with the alkali and copper salt in a cur- rent of hydrogen. To the resulting mixture, and without removal of the cuprous oxide, we added, still in an atmosphere of hydrogen, the necessary stannous -chloride and acid. Only then and for a moment was the bottle opened in order to replace the cork and gas tubes by the stopper of the bottle. Heated, as before, 0.4298 gram öf freshly crystallised salt gave 44 % (43.93) of the sulphur as sulphite, a re- sult confirmatory of our theory, and better than that (40 %) got by iodine- titration. It thus appears clear that the sulphite formed when the oxy- amidosulphonate is oxidised by eu prie oxide is half what is produced when the salt is decomposed by alkali alone. That only nine-tenths of the reckoned sulphite is obtained in both cases is partly if not entirely due to two causes. One of these is that, as already pointed out, in each mode of decomposition, a little oxyamidosulphonate (or a body like it) is always left at the end of the reaction. The other and main one is that the tin reaction is incomplete, for working upon sulphite of a known degree of purity we have got only 91 and again 93 l"/o of the sulphite indicated. In alkaline solutions, silver and mercuric hydroxides act quite similarly to cupric hydroxide, qualitatively at least, and yield much sulphite. Constitution of hyponitrites, as revealed by the decomposition of oxyamidosulphonates. The decomposition of oxyamidosulphonates into sulphite and hyponitrite sets at rest any doubt as to the constitution of hyponitrites. OX OXYAMIDOSTTLPHOXATES. 229 For coming in this case direct!}' from a substituted hvdroxyamine a hyponitrite must have its oxygen between the nitrogen and metal. Berthelot and Maquenne have recently published papers (Compt. rend. 108, 1286, 1305) containing analyses of calcium and stron- tium hyponitrites. These analyses, as they point out, establish the ac- curacy of the empirical formula given by one of us (Divers) to hvponi- trous acid, upon which doubt had been cast by the previous work upon the silver salt by Berthelot himself and Ogier (Compt. rend. 96, 30, 84). To this salt the latter chemists gave the formula Ag4N405, the correctness of which was afterwards contested by us (J. Ch. Soc. 45, 78). Berthelot now admits this salt to be not obtainable.in a pure state, thus also confirming us, as against Zorn, van der Flaats, and Menke. all of whom claimed to have got it pure without difficulty. Zorn's opinion that the molecule of the acid contains two atoms of each of its elements, already generally accepted, is now en- dorsed by Berthelot and Maquenne. Lastly, Maquenne is disposed to deny that nitrous oxide can be the anhydride of hyponitrous acid, even to the extent that carbon monoxide is the anhydride of formic acid, but on grounds which to us seem quite insufficient. Even the facts recorded in this paper can leave hardly a doubt that it is so. The formula of hyponitrous acid may now confidently be written as HOjSt2OH or (XOH),, that is, the acid is hydroximidogen, of which NOH is the radical. lour. Sc. Coll ko/. /// P/.XXk. ^c 0 2 . n t. % IS( -. n an «à 20 t. V f 1 - 40 ( • -»_[, t ■211 S^^=; eo \ S --'TI > — t2 \ rr s.. A ' \Q^N — -2 n U ist T + r n 2 \ \ 2 <- ^ M ■too 1 B 300 \ > v A "B ttO \ ^X \ \ \ C v &5 700 V o> 1 a 200 ^5l twist ^ i lour . Sc. Coli Ko/. ///. //. XX K/. Magnetic Lag in twisted Nickel longitudinally magnétisa . \ V < s N r ^ " y \ v . j / / '» ( > || l \ V . 2 / \\ \ \ \ l . / / \ \ \ / 7 / / \ \ \ . -/ \ \ i / \ \ \ i f -> \ \ \±360' \ '' / \ \ \ / II K \ ( \ i \ \ \ 1 ! \ V \ \ \ \ / il \ " \ 1 1 \ \ \ \ \ / I \ ^ \ \ \ \±90 1 / \ \ vT \ \ \ / / \ \ / 1 \ \ ) \ \ \ / / \ \\ 1 1 \ \ ±60'/ \ y / / ±i 5' W y 1 1 1 \ \ ' l ! \ v_ \ II A*'20f f . \ \ 1 / A K \ ±3 0' \ \ \ 1 1 / /\ _.' <: \ fr ) \ * tao 1 1 '. i \ \ 1 / \ \ ! / 1 < I Jour, Sc. Coll. Vol. Ill PI. XXVII. On a Condensation Product of Acetone and Aldehydammonia, lYIitsuru Kuhara, Ph. D. Smiic time ago A. Mantzsch1 obtained, by the condensation of acetone and aldehydammonia, ;i complex mixture of basic substances boiling between L50° nnd 360°. from which he «'mild only separate collidine which was formed as a direcl condensation product of aldehydammonia. I have, likewise, attempted to carry on for the last four years the -nine line of investigation, in a somewhat different manner and liave succeeded in separating a basic substance besides collidine from that complex mixture. The results of the author's experiments have already been communicated to the Tokyo Chemical Society in June, 1888. In September of the same year E. Durkopf2 published the results of his researches on the same mbject, in which he gives an account of a basic substance which was formed by the condensation of acetone and aldehydammonia. His compound is, however, probably isomeric with that obtained by the present author, as both compounds arc similar in some respects and have the same molecular formula. In the author's experiments, a mixture of 1500 grams of acetone and 750 grams of aldehydammonia was continually heated at 120°- lo0° for 25 hours in n cast iron digester over an oil bath. When the digester was opened, its contents were found to lie a dark brown 1. Bur. a. cleat, chem. Gesel. 14. n; 17. 2 „ „ m 21,2713. 232 M. KU HAT,' A liquid of thick oily consistency, with a strong ammoniacal odour. The whole was transferred to a retort and subjected to distillation over a water bath, by which means a greater part of the unchanged acetone was removed. The remainder of the liquid was then distilled over an oil hath to get rid of the remaining traces of acetone and the water formed in the reaction, until the temperature reached 120°. The retort was then heated above the last temperature over a direct, flame, but a very small quantity of an oily liquid distilled over until the temperature attained 170°. from 170° to 230°, however, a large quantity of a pale yellow oil, and above 230° some thick syrupy liquid distilled over. The residue in the retort was of a dark brown resin- ous matter. On subjecting the liquid boiling between 170° and 230° to fractional distillation, an amber-coloured oil which boils at 173°-176° was separated. It soon turns brown in contact with air, even slightly in the process of distillation. The oil has a strong stupifying odour somewhat resembling that of conine. When inhaled, it causes a headache. It is poisonous ; for a drop of it will instantly kill a large fros". It ü'ives white fumes with hydrochloric acid, and combines with common mineral acids with the evolution of heat, it is slightly soluble in cold water, but on warming, it precipitates. As it soon changes in contact with air. it was thought best to convert into a platinum double salt, which is more stable. To do this the oil was dissolved in dilute hydrochloric acid, and on adding platinum chloride to the solution, a double salt was precipitated as a yellow crystalline powder. The precipitate was again dissolved in hot water in which it?is moderately soluble, and from the solution on cooling laro-e reddish-yellow feather-shaped crystals of the doublesalt were deposited. On repeating several times the crystallization of the double 'salt from the hot water, the pure salt— for such it was judged OX A CONDENSATION PRODUCT OF ACETONE & ALDEHYDAMMONIA. 233 to be from its appearance — was obtained. The crystals, after being dried over sulphuric acid3 were analysed. I. 0.3754 gram of the double salt on combustion gave 0.1502 gram ILO and 0.111 gram C02. [I. o.:i7f>."> gram of the double sail gave 0.1419 gram ILO and 0.4245 gram CO«. III. 0.4451 gram of the double salt gave 0.1778 gram FLO and 0.4636 -ram CO-, IV. 0.3364 çram of ihr double salt gave 0.149 gram ILO and 0.3488 gram CO,. V. Al first the double -all was reduced by sodium amalgam and then the chlorine estimated in the usual manner ; upon which 0.3248 gram of the double salt gave 0.1243 gram AgCl. \ I. 0.295 -ram of the double sail, when burnt with soda lime, gave 0.212 gram 2X1 1 .CI.LtCI,. The double sali, on ignition, gave the following1 amount of platinum : I. 0.1974 gram the double -alt gave 0.0594 gram ?t. II. 0.3756 .. .. .. „ .. 0.1137 ., „ III. «».22 11 „ .. 0.0680 „ IV. 0.256Ö .. .. .. .. „ 0.0769 „ „ Y. 0.2020 „ „ „ .. „ '0.0611 „ „ i. ii. nr. iv. v. vr. Carbon 30.07 28.39 ."ICSI 28.27 — Hydrogen ... 4.44 4.48 4.10 4.91 Nitrogen ... — — — 4.51 — Chlorine ... — — — — S2.24 T. II. III. IV. V. riatinum 30.06 30.27 29.76 29.76 30.24 234 M. KUHABA Mean (found). Carbon 29.38... Calculated as (CgHttN'.HClJaPtCl*. 21). 19 yaroiren 4.25 Nitrofen 4.51 4.25 Chlorine . Platinum. ..32.24... ..30.00... L00.G3 100.00 The Hydrobichromate, (C1,sl I1;N)L, I L.( 'rJ )r. — To the base dissolved in dilute sulphuric arid, some crystals of potassium bichromate were added and the mixture gently heated. On cooling the solution, the hydrobichromate crystallized out in large brilliant yellow prismatic plates, and upon reerystallization it was soon obtained pure. It is sparingly soluble in cold, but easily in hot water. When heated, it deflagrates and leaves a green oxide Cr203. The crystals of the hydrohichromate, after being dried over sulphuric acid, were analysed : I. 0.3557 gram the salt gave, on combustion, 0.1853 gram H20 and 0.5491 gram C02. II. 0.2504 gram the salt, when ignited, gave 0.0831 gram Cr203. III. 0.1886 „ „ , ,. 0.062 „ Found. Calcul; I. . ir. nr. itedas (C8Hl3"S)2H2Cr207. ( 'arlton 41.88 ■ — ■ — ... ...41.31 Hydrogen 5. 79 — — ■ ... ... 6.02 Nitrogen — — — ... ... 6.02 Chromium — _ J.oO 22.54... ...22.54 Oxygen - — ■ — ' — ... ...24.09 The Mercury Double Chloride, CJl1;X.IlC1.2HgCL+H,0.— On adding the solution of mercuric chloride to the aqueous solution of the hydrochloride of the base, fine hairy white crystals separated out in ON A CONDENSATION PKODUCT OF ACETONE & ALDEHYDAMMONIA. -35 ;i Few minutes, and upon recrystallization from absolution in hot water. the pure double salt, for so it was judged to 1)'.' from its appearance, was obtained in large brilliant scaly plates. The sail melts at 148°- 149°. Ir partially volatilizes at above 1(10°. It seems to contain one molecule of the water of crystallization, but on account of its volatility its amount cannot be accurately determined by the usual method-. The crystals of the double salt, after being dried over calcium chloride, were analysed : (I.) 0.3957 -ram the double sail gave 0.3941 -ram AgCl. (II.) 0.3645 .. „ „ „ „ 0.2353 .. U.S. 1 ' iutl- ,.i i . i , _ < alculated as I II. C8Hi3N.HC].2HgCl2 + Ha0. Chlorine 24.64 24.66 Mercury 55.63 5 r>. ô i) Water — 2.50 The crystals of the double salt, finely pulverized, were heated over an air bath al 130° for some time, by which the powder became compact, having apparently lost its water of crystallization through partial volatilization. Thesalt was then analysed. 0.3887 gram the anhydrous sail gave 0.2593 -ram HgS. Found. Calculated as C„Hl3N.H« 1 2B~gCl2. Mercury 57.83 "»7.2 On comparing this analytical result with that given in the preceding analysis, we may infer that the salt contained one molecule of water. The Gold Djuble Salt, GaE13N. HCl. AuCl3.— This was prepare«! by adding an alcoholic solution of gold chloride to an aqueous solution of the hydrochloride of the base. On driving off the alcohol from the solution, the double chloride separated out as a crystalline precipitate, which consisted of microscopic yellow needles. It was 236 M. KUH A IIA redissolved in hot water and repeatedly recrystallized. It melts under hot water to a sticky oily mass and ''solidities again on cooling. The salt, dried over calcium chloride, was analysed : 0.3386 gram the salt gave 0.1452 gram Ai Gold... Found. ...42.88.. Calculated as C8 H 1:i X . U C 1 . A n C !„. 42.54 The Hydrocldoride, CÖH13K.HC1. — Un mixing concentrated hy- drochloric acid and the hase together, they combined energetically with the evolution of heat, and the mixture solidified. The whole mass was then dissolved in alcohol,and on concentrating the solution nearly to dryness the hydrochloride crystallized out, in an impure slate, in (he form of long needles. The impure sali thus obtained was purified by dissolving it in the smallest possible quantity of alcohol, and then adding ether to the alcoholic solution, by which means the hydrochloride was precipitated in line colourless needles, as it is insoluble in ether. It is exceedingly deliquescent in air. The crystals seem to decompose under a desiccator, as they soon lose their transparency, probably giving off a part of hydrochloric acid. The crystals, dried over sulphuric acid under a desiccator, were analysed : I. 0.1540 gram the salt gave 0.1300 gram AgCl. II. 0.2112 III. 0.1500 0.1819 0.1276 fourni. Chlorine. . I. II. 20.88 21.25 Calculated as CgHVX.HCL 22.25 The low percentage of chlorine, in the above analysis, may possiblv be due to the decomposition of the salt during desiccation. The Hijdrubroniidi'j C8H13K.HBr. — On adding bromine to the base dissolved in ether containing some alcohol, the hydrobromide ON' V CONDENSATION PRODUCT OF ACETONE & VLDB-HYDAMMOXIA, 287 separated oui in colourless transparent prismatic needles. The salt, dried over caL iuin chloride, was analysed : 0.2946 gram the salt gave 0.27 gram AgBr. Pound. Calculated as C8Hl3N.HBr. Bromine 39.13 39.21 The Acetyl Compound, CsrT]2N(CH3\\üO). This was prepared in the usual way, by dissolving the base in anhydrous ether and adding acetyl chloride toit. The hydrochloride of the acetyl com- pound thus precipitated was purified by repeated recrystallization from the mixture of alcohol and ether. It crystallizes in fine needles, in appearance very much like the hydrochloride of the base. It is very deliquescent. The free acetyl compound separates oui as an oil, on adding caustic potash to its hydrochloride. The salt, dried over calcium chloride, was anal ysed. 0.2649 gram the hydrochloride of the acetyl compound gave 0.1906 gram AgCl. I-', unci. i ulated us < VipV (CÏT3.C1 l i I • i Chlorine 17.7!) 17. CI The Nitroxo-eompounl, <'JI,\.\'<). — On adding ['irre- of solid potassium nitrite gradually to the aqueous solution of the hydro- chloride of the hase. ;i large quantity of a thick brown <>ilv substance separated. [t was extracted with ether, and shaken with water, and again extracted with ether. After repeating the same process of washing several times, the oil, dried over calcium chloride, was once distilled, h is a pale yellow oil, having the characteristic odour of nitrosamine. It gives the well-known Liebermann's nitroso-reac- tion : consequently the base itself is a secondary amine. The si »lui ion left after separating the nitroso-compound, was treated with caustic potash, by which a small quantity of an oil was 238 M. EUHARA precipitated. The hitter was then changed into an hydrochloride, and consequently into a double salt, with mercuric chloride. The double salt thus prepared was found to be identical with the double salt of the original base, agreeing- in its crystalline form, melting point, and analytical results. The double salt, dried over calcium chloride, was analysed, and the results were as follow : I. 0.2396 gram the double sail gave 0.2353 gram AgCl. II. 0.2000 „ ., „ „ „ 0.1958 „ III. 0.3165 ., „ „ „ „ 0.2037 „ llgS. IV. 0.3378 ,. „ ., ., „ 0.2200 „ Found. Calculated us I. IT. III. IV. 0SU!;,\ .HC1.2HgCl,H H.,0. Chlorine 24.30 21.22 24.66 Mercury — 55.-17 55.50 55.59 It is evident, therefore, that a greater part of the base changed into anitrosamine, and the oil separated from the solution with caustic potash was simply the base unchanged. According to the results of analysis thus far, il is highly probable that the author's base has the formula CSH13N and differs in many respects from the base isolated by Durkopf,* although thev are isomeric. The Oxidation and Decomposition of the Base. '. )n heating 10 grams of the base freshly distilled, with the chromic acid mixture consisting1 of 20 grams of potassium bichromate and 20 grams of sulphuric acid diluted with its own volume of water, the mixture gradually became green with the evolution of OIL The * Durkopf's base is stated to change into collicline on treating it with KN02. — Ber. d. cleut. dum. Gesel. 14,271 :v OX A CONDENSATION PRODUCT OF ACETONE & ALDEHTDAMMONIA . 239 gre in solution thus formed was mixed with a large quantity of water and subjected to distillation ; and in the distillate was found a considerable quantity of acetic acid and a little formic acid. Next, the base was subjected to oxidation with a potassium per- manganate solution containing 1 part of potassium permanganate to . c are special values for the atoms of the constituents A, B, C In all known instances a. J3, y. &c, as well as F, are very simple functions. 242 K. IKEDA It is the endeavor of the present paper to «how that capillary- attraction of organic liquids belongs to the above mentioned category of physical properties, and can be represented by the same general formula. That there is a well defined relation between capillary constants of liquid substances and their molecular composition has been put beyond all doubt by the extensive and elaborate experimental inves- tigations of R. Schiff. He has chosen boiling points as the temperature of physical comparability, and expressed the relation between capillary constants at these points and the chemical composition by the formula r.a-bH N = - H ' in which N stands for the relative number of molecules raised in a capillary tube, H is the sum of the special values for the constituent atoms referred to hydrogen as unity, e is the base of the Napierian logarithms, while a and b are constants. This equation accords fairly well with the experimental results ; but it is hard to see why such a relation should exist between N and H, or rather the meaning of the latter term is very difficult to make out. The expression, indeed, claims no higher title than that of being strictly empirical; still it is better to use those terms only, to which some probable meaning, at least, can be assigned. The two formulre about to be proposed are in no way superior to Schiff's as far as purely theoretical consideration goes, for they pretend to no theoretical ground whatsoever. Still they seem to be more rational inasmuch as the terms involved have some meaning-, and besides a^ree better with the observed results, as will be shown later on. The new equations are of the same form as the general expression already given, and show the relation between capillary constants and molecular volumes in a very clear manner, although it is rather difficult to understand why they 2 CA.PILLA.RT ATTRAC TION" IN" RELATION* TO CHEtftCAL COMPOSITION* Ac. 243 should be of these particular forms. The formula? may be written thus : ,r_ma-\-nbA-pc+ , or KV*=ma + nb+pc+ (I„) and JÔ VlVi = ma' + nb' +pc' + {II). Here K stands for the capillary rise of a liquid multiplied by its specific gravity, both being taken at the boiling point. Fis the molecular volume, or the molecular weight divided by the specific gravity ; O .- - a '< 5 g s § t r S I i1 •: ^5 Normal )., .. Ilexane \-''u 140.02 t-530 (.650 2.772 2.845 + 12') + 2.0 Normal ) ,-, TT Heptane ) ■ "' 162.56 5336 5400 2.575 2.606 -j-64 + 1.2 Diisobutyl C8H18 1 8 1,89 6059 6150 2.410 2.446 + 91 + 1.5 Diis'oamyl ClfllL, 231.80 771 s 7650 2.192 2.168 -ils -1.1 Amylene C5H10 110.18 3564 3750 3.082 3.242 -:- L8 i + 5.2 OctyleDe C8H1C 177.6] 6074 G000 2.572 2.541 -74 -1.2 Dially] CCH10 120.10 t259 L350 3.008 3.072 + 01 + 2.1 Benzene C6R6 '."'».17 t012 L050 1.25 1 t.295 + 38 + 1.0 Toluene C7H8 11 s.J.-. 17 1. s 3.692 3.733 + 52 + 1.1 Xylene (1:2)C8H10 139.91 5551 5550 3.354 3.353 -1 0.0 « (1:3) „ 139.69 55 i !• 5550 3.358 3„362 + G + 0.1 „ (1:4) „ 140.21 55 1 1- 5550 3.340 3.343 + G + 0.1 Ethyl- ) benzene ) " 138.96 5 60 2 5550 3.420 3.388 -52 -1.0 prpyl" an,, benzene ) J l- 161.82 6427 6300 0.122 3.061 -127 -2.0 Kthyl- ) toluene ) " 161.95 6373 6300 3.092 3.057 -70 -1.1 Mesitylene ,, 1G2.I1 6234 6300 3.012 0.044 + QG + 1.1 Cyniene C10H14 184.46 G970 7050 2.782 2.814 + 80 + 1.1 24*6 K. IKEDA TABLE I. — Continued. Substance. V Molecular Volume. 3. KV* B + t> O u ■j U 4 g (3 'S ■? 1) 0) g « I £ " + tt )■> 196. ? 7166 7200 2.612 2.624 + 34 + 0.4 Methyl Alcohol {CH40 42.72 1066 l<)7o 3.818 3.832 + 4 + 0.4 Ethyl Alcohol JC2H60 62.19 1731 1820 3.530 3.710 + 89 + 5.1 Propyl Alcohol |03H80 81.29 2583 2570 3.524 3.506 -13 -0.5 Isopropyl Alcohol 1 - 81.69 2513 2570 3.404 3.482 + 57 + 2.3 I so butyl Alcohol JC4H100 101.61 3287 3320 3.208 3.240 + 33 + 1.0 Isoamyl Alcohol JC5H120 122.74 4172 4070 3.068 2.994 -102 -2.4 Dimethyl- ethyl-curbiuol) ': 121.27 4140 4070 3.100 3.047 -70 -1.7 Allyl Alcohol |0„H,0 74.11 2495 2420 3.910 3.793 — 75 -3.0 Methyl Formate Jc,H402 62.65 2452 2400 4.944 4.839 -52 -2.1 Ethyl Formate |C3H602 84.57 3074 3150 3.952 4.050 + 76 + 2.5 Methyl Acetate ! - 83.66 3076 3150 4.020 4.117 + 74 + 2.4 Propyl Formate Jcjia 106.15 4058 3900 3.710 3.565 + 158 + 3.9 Ethyl A cet ate „ 105.78 3854 3900 3.542 3.585 + 46 + 1-2 Methyl Pre pionate 1 - 104.27 3845 3900 3.612 3.664 + 55 + 1.4 Isobutyl Formate |CSH1002 130.74 4829 4650 3.230 3.110 -179 -3.7 * The molecular volume of this compound s lems to be abnormally low, being 10 or 9 units less than the calculated, which accounts for the discrepancy as shown in the calculation for T;ib!e II. Still it is but fair to confess tint several other substances of similar constitution have molecular volumes quite as anomalous. See Lessens calculation of molecular volumes in Liebig's Annaleu(254. 54). CAPILLARY ATTRACTION IX RELATION TO CHEMICAL COMPOSITION. 247 TABLE I.— Continued. Substance. V Molecular KV* o : + + «3 3 5 CttM s &» - Î ■< I 2 = "2 - 2 a?« be i S 3 Volume. S 5-1 + o 3 s = A 3 | 8P Propyl Acetate JC5H1002 L 28.56 Mil 4650 3.184 3.190 + 9 + 0.2 Ethyl Pro- pionafce f " 127. » 152.51t 5610 5400 2.978 2.867 -210 -3.3 Propyl Pro- pionate » 150.70 5 | i i 5400 2.! »22 2.919 -0 -0.1 Ethyl Butyrate >■> 150.25 5356 5400 2.908 2.932 + 14 + 0.8 Ethyl Iso- butyrate ?» 150.68 5246 5100 2.836 2.919 + 154 + 2.9 Methyl Valerate >) 148.33 5430 5400 3.006 2.989 -30 -0.6 Isoamyl Acetate C7H1402 1 7 1.60 6372 6150 2.762 2.666 -222 -3.3 Isobutyl | Propionate' >> 173.55 6051 6150 2.618 2.690 + 96 + 1.6 Propyl j Butyrate j }> 1 73.85 6189 6150 2.700 2.683 -39 -0.0 Propyl Iso-j butyrate j »? 174.20 6056 6150 2.634 2.675 + 91 + 1.6 Ethyl j Valerate j >i 172.99 6139 6150 2.698 2.703 + 11 + 0.2 Isoamyl ) Propionate) CâHigO;, 196.90 6977 6900 2.524 2.49G -77 -1.1 Isobutyl j Butyr.ite ) )> 197.66 67S6 6900 2.442 2.483 + 114 + 1.7 [sobutyl ) Isobutyrate) " 198.21 6814 6900 2.442 2 173 + 86 + 1.3 *In Lossen's paper above alluded to, this value is put=l51.7. The determination of Elsässer seems tobe eve a lower than this, aud as Schiff lauds very highly the purity of this chemist's investigation materials, it is highly probible that Schiff's own substance in this instance is not as pure as his usually are t The same remark applies to this also. 248 K. IKEDA TABLE I.— Continued. Substance. V Molecular Volume. s. »o : + + + T3 a» m O Ü 2 BÏ4 3 a S * S & s 5 a ■ PvPiyl f {c8HI60. Valerate ) 8 16 - 196.82 6969 6900 2.524 2.499 -09 -1.0 Ml7\ * C-H80, Acetate ) " a - 121.5 4520 4500 3.370 3.361 -20 -0.4 Ethyl In TT n Oxalate \^n^ 138.79 5108 4950 3.124 3.027 -158 -3.1 Benzoate ) 6 s - 151.65 6653 6600 3.563 3.534 -53 -0.8 Benzoate p»ni«u2 174.65 7335 7350 3.178 3.184 + 15 + 0.2 Ethyl-oxide C4Hl0O 106.27 3441 3600 3.142 3.287 + 159 + 4.6 Methvl I Amyl C6HuO 148.13 5060 5100 2.807 2.829 + 40 + 0.8 Ether ) Paraldehyde CcHlL,0;, 150.74 5708 5850 3.084 3.160 + 142 + 2.5 Acetone ■ C3H60 77.10 263!.; 27UU 3.894 3.988 + 64 + 2.4 dehyde ) 5 10 118.27 1303 4200 3.345 3.265 -106 -2.4 Acetic An-)n ,j f) hydride ) * 6 ä 109.5 4462 4200 3.896 3.66.6 -262 -5.9 Dimethyl Lg Q Acetal [ 11U.9 4081 4050 3.470 3.443 —31 -0.8 Difh;y1, CtHuO, Acetal j ' 159.91 5 144 5550 2.692 2.746 + 106 + 2.0 ; Pinacoline C6H120 138.25 5017 4950 3.086 ;3.040 -67 -1.5 Ani sol OHbO 125.21 5503 5550 3.928 3.966 + 47 + 0.9 Phenetol C8H10O 149.4 6142 6300 3.370 3.455 + 158 + 2.5 Methoxycresol „ 147.8 6345 6300 o.O-jI 3.506 -45 -0.7 Dimethoxy-) cresol ) s 10 - 157.6 7)87 7050 3.598 3.582 -37 -0.5 Cnminal- )n jj r. delude \C^120 188.9 7750 7700 3.019 3.000 -50 -0.7 CAPILLARY ATTRACTION IN RELATION* TO CHEMICAL COMPOSITION. 249 TABLE I.— Continued. Substance. F Mol ecu In r Volume. AT'* + + 5? A Observed. K Calculated. "0" g a "2 + 4) « S ■= S I * ■a g per centoge Difference. Formic Acid JCHA 41.08 1558 1650 5.917 ? ? ? Acetic Acid CoH^O, 63.40 1853 2400 3.670 ? ? •? Propionic Acid |C3H602 85.94 2558 3150 3.212 ? ? p Butyric Acid JC.HA L08.10 •".217 39 10 2.886 V ? V Isobutyrio Acid 1 •• 109.87 3162 3900 2.746 ? 9 ? Valeric Acid |0äH10O2 Pin. 2 7 3821 1650 2.570 J> ? V Ethylene Chloride jc,H4Cl2 85.35 381G 3700 4.8 10 4.692 -116 -3.1 Ethidoue Chloride I - 88.08 3427 3350 4.104 1,012 — 77 -2.3 Propyl Chloride JC3H7CI 91.58 3263 3425 3.732 3.908 + 162 + 4.8 Propylene Chloride JC3H6C1, In 7.- V.i 4533 1 tso 4.062 3.988 -83 -1.8 Isobutyl Chloride jc4HaCl 1 1 L20 4068 4175 3.333 3.421 + 107 + 2.6 Isoauiyl Chloride C5HUC1 134.40 4859 4925 3.118 3.161 -{-66 + 1.4 Chloroben- zene }CCH,C1 114.4 5053 5075 4.128 4.146 + 22 + 0.4 Chloro- toluene C7H7C1 L34.9 5848 5825 3.734 3.719 -23 -0.4 Chloroform CHCI3 84.65 3 1 13 3450 4.420 4.430 + 7 + 0.2 Carbonte- )r,,,. trachloride}^'4 103.77 4313 4300 4.080 4.068 -13 ~o:3 Bromine Br2 53.52 2792 3000 7.132 7.663 + 208 + 7.4 Ethyl Bromide |c2H5Br 77.07 3004 3075 4.443 4.543 + 71 + 2.3 Ethylene Bromide }c2H4Br2 97.01 4900 4500 5.128 4.710 -400 -8.1 250 K. IKEDA TABLE I. — Continued. Substance. Molecular Volume. KV* + + + > OQ Q 5 4) 1 rt o "3 o c4tt 2 b 'S * 2 S 5 * 1 1 1 0) hi a> P£Wl |c3H7Br Bromide ) A 97.05 3830 3825 4.007 1.001 — 5 -0.2 Isopropyl ) Bromide ] " 99.2 3814 3825 3.861 3.871 + 11 + 0.3 ^lide | 5 T3 'S O g a f + = ! 1 it « 8 « SP Aniline C6H7N 106.08 5168 5075 4.730 4.6 1 -5 -93 -1.8 Pyridene C5H5N 89.39 4137 1325 4.895 5.116 + 188 + 4.5 Piperidene C5HnN ] 08.76 4693 I77Ö 4.138 4.208 + 82 +1.7 Quinoline CaH7N 139.75 7355 7175* 4.452 1,3 13 -180 -2.4 Nitr?; }oh3no2 methane ) 59.5 2390 2375 5.207 5.175 -15 -0.6 Nitroefchane CjjHjNO, 80.25 3070 3125 1.271 1,317 + 55 + 1.7 Isoamyl- jCH NOs nitrate ) ' " û 152.59 5923 5825 3.142 3.091 -98 -1.6 Acetonitrile G.H^N 57.23 1873 1875 4.327 4.331 + 2 + 0.1 Propionitrile C3H5N 78.28 2649 2625 3.825 3.790 -24 -0.8 Capronitrile C6HUN 141.1 5345 4875 3.189 2.909 -470 -9.0 Eth*yl (CHS Sulphide p»11"0 122.2 t622 4650 3.420 3.440 + 28 + 0.6 Allyl Thio- LHN carbiuiide ) 4 " 123.13 5046 1925 4.193 4.092 -121 -2.4 Phenyl thio- jc 143.7 7021 7025 4.077 4.081 + 4 + 0.05 Methyl thio-CHNS cyanate \ - 6 78.96 3329 337Ô 4.745 4.816 + 46 + 1.5 Ethyl thio- xs cyanate ) A ° 99.84 4166 4125 4.176 4.135 -41 -1.0 Carbon )pq Bisulphide) 2 62.06 2842 2900 5.813 5.930 + 58 + 2.0 »9x600 + 7x75 + 650=0575 6575 + 2x300=7175. But this mode of calculation is pro- bably incorrect. 252 K IKEDA The Atomic constants (a* b' c') for the various elements used in the calculation of the second table are as follows : for carbon 44 when there is no double linking. „ hydrogen 21.5 in all cases. „ oxygen 31 in alcohols* 44 when combined with two carbon atoms. 55 when connecting a carbon atom to a benzene ring, chlorine 119 when only one atom of chlorine is united with a carbon atom, chlorine 110 when two or more atoms of chlorine are united with a carbon atom, bromine 162 when only one atom of bromine is united with a carbon atom, iodine 215 when only one atom of iodine is united with a carbon atom, nitrogen 54 in all cases where it is a triad. 55 55 11 11 11 JJ 55 sulphur 148 when combined with two carbon atoms. When two atoms are united together by more than one bond, the value of 7t F1'ls is increased by 19 for each additional bond. Thus : group (NC = C/) = 107 = 2x44 + 19 group (XC7=0) = 118 =44+55 + 19 group (-C=tf)=*136=44 + 54 + 2xl9 group (-iV = C = S)=284 = 44 + 54 + 19 + 148 + 19 group (SCN) - 270 ) j-these have not been resolved. group (NO,) =172 ) CAPILLARY ATTRACTION IX RELATION TO CHEMICAL COMPOSITION. 253 TABLE II. Substance. Normal Hexanê CcH^, Normal Heptane C7Hlc Diisobutyl (C4H9)f Diisoamyl Amylene Octylene (C,Hu)f C5Hio CsHig Diallyl USC=C. H,C-CH2-C=CH2 H <" H IK' « H Benzene He \ Cfl Toluene IIC=C-C=C-C^C-CH H H H II II Xylene (I : 2) „ (1:3) „ (1 : 4) Ethyl Benzene Propyl Benzene Ethyl Toluene c„h »1-l10 C6H..C2H, C6Hj.C3H7 Crllj.CH^.CoH: Mesitylene (1 : 3 : 5) CcH3(CH3)n Cymene (1 : 5) CGB4(CH3)(C3H7) Ternpnp ' C3H7 xeipeue j H H H | H (citrene)JH3C-r-c=c-c=C-( H K. 2.772 2.575 2.410 2.192 3.082 2.572 3.008 1,20 1 3.G92 3.354 3.358 3.340 3.420 3.122 3.092 3012 2.782 2.748 V Molecular Volume. 140.02 162.56 184.89 231.80 110.18 177.61 126.10 96.17 118.25 139.91 L39.69 1 1,0.2] 138.96 161.82 161.95 162.41 L 84.46 18C.3 * 9, o co' + 1 1 i 8 '- + V u He» -- + ° u a o a M - Si* g + .fj _ 567 565 — 2 -0.35 652 652 ± 0 +0.00 734 739 + 5 + 0.68 915 913 — 2 — 0.22 ! D 1 45 1 + 3 + 0.66 723 715 — 8 -1.11 521 517 — 4 -0.77 451 450 — 1 -0.22 536 537 + 1 + 0.19 624 624 ± 0 + 0.00 623 624 + 1 + 0.16 624 i!21 J- 0 + 0.00 624 624 a_ 0 + 0.00 714 711 — 3 -0.42 712 711 — 1 -0.14 705 711 + 0 + 0.85 787 798 + 11 + 1.40 791 822 + 31 + 4. * For the peculiarity of the molecular volume of this compound, see the foot-note annexed to the foregoing table. 254 K. IKEDA TABLE U.— -Continued. Substance. K. V Molecular Volume. 00* + a & "= I.5 g*r : + + + be « ce o ö § o u W Carbinol j^^is" 3.100 121.27 506 509 + 3 + 0.60 AUyl Alcohol H2C=CH-CH2OH 3.910 74.11 318 311 — 7 -2.20 Methyl Formate HCO.,CH3 4.944 62.65 293 292 — 1 -0.34 Ethyl Formate HC02C2H5 3.952 84.57 374 379 + 5 + 1.34 Methyl Acetate H3C20,CH3 4.020 83.66 372 379 + 7 + 1.88 Propyl Formate HC02C3H7 3.710 103.15 474 466 — 8 -1.68 Ethyl Acetate H3C202C2H- 3.542 105.78 461 466 + 5 + 1.08 Methyl Propionate HaC30,CH3 3.612 104.27 457 466 + 9 + 1.97 Isobutyl Formate HCO:,(CiHa)'3 3.230 130.74 565 553 12 — 2.12 Propyl Acetate H30202C3H7 3.184 128.56 550 553 + 3 + 0.60 Ethyl Propionate H6C302C2H3 3.168 127.86 545 553 + 8 + 1.45 Methyl Butyrate H7Ce02CH3 3.250 126.36 544 553 + 9 + 1.65 CAPILLARY ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 255 TABLE II.— Continuel. Substance. K. V Molecular Volume. -CI + o : If g o . bt s - -j "S S a - . So Methyl Isobutyrate : (H7C402 CH 3.190 126.4 ! 540 553 + 13 + 2.41 Isoamyl Formate HCO,(C,Hu)' 3.080 103.22 665 640 -25 -3.76 Isobutyl Acetate HSCA(CA)' 2.978 152.51 651 -11 -1.69 Propj 1 Propionate BROACH, 2.922 15 '.7o 635 640 + 5 + 0.79 Ethyl Butyrate H7CAC2H5 150.25 632 640 + 8 + 1.27 Ethyl Isobutyrate (H7CA) '.11 2.836 150.68 62 i 640 + 14 + 2.23 Methyl Valerate H9C502CHa 3.006 148.33 632 640 + 8 + 1.27 Isoainyl Acetate H3CJ0,(C,1IU- 2.762 1 74.60 735 7J7 - 8 -1.C9 Isobutyl propionate H5C302(C j.-; is 1 73.55 711 727 + 13 + 1.82 Propyl Butyrate H:C40,C3H7 2 7 170.85 723 727 + ft + 0.55 Pr pyl Isobutyrate (HA< l- _ 174.20 716 727 + U + 1.51 Ethyl Valerate (H.O.OJC.B _ 172.:».' 71s 727 + 9 + 1.25 Isoamyl Propionate jHsCsOAHh 2.52-1 810 814 + 4 + 0.49 Isobutyl Butyrate HAOI Ally] Iodide [ H H, JH,C C-CI Iodobenzene W Propylamine C„H7.NH2 Allylamiue C8HSNH2 Isobutylamine i'^n,) xii. Dietljylamine (CUI,),NH Trietliylaminc (C2Hsï3N Aniline C6H3NH2 1.007 3.831 \ :■•>■■; i 3.581 4.204 •5.007 1.587 1. 1 »so 3.024 3.232 4.358 4.168 3.920 4.290 3.563 3.333 2.750 1,730 07.05 00.2 00.5' 1 18.39 110.88 L- I + be ® a 9 9 ? 7 SK 443 44. 3.773 141.95 63.9 86.12 106.9 446 445 424 420 520 j 531 582 [ 580 671 667 32i» 324 Ml 411 50.1 498 128.28 [585 585 151.05 ; 070 072 100.9 ,483 474 130.55 641 6 II 85.fi 1 "8.38 378 380 356 356 106.76 4071407 109.05 L 53.82 1.63 167 032 0 11 106.08 533 526 + 2 — 1 — 4 + 2 •> 4 + 4 + 0 o — o ± o + 2 — 9 + 3 i •> T - ± 0 ± 0 + 4 + 9 — 7 + 0.45 0.23 —0.94 + 0.38 -0.35 0.59 + 1.2; + 0.00 0.1)0 0.00 + 0.30 -1.80 + 0.47 + 0.53 ±0.00 + 0.00 + 0.86 + 1.42 1 .31 CAPILLAi; Y ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 259 TABLE II.— Continued. Substance. V Molecular \ olume. Pvriilino 11 i I II H H C N-C C-C=C Piperidine {Ir,c^. giggle Quinoline h >i ii / \ ( c . ( ' ■ < i < < -N ^ h h a h t.895 U38 1.452 N/tromethane ('II, (NO.,) 5.207 Nitroefchaue C.Hs(NO,) t.27] [soamy] Nitrate CfHuO(NO I 3.142 Acetonitrile CH3C = N 4.327 Propionitrile CaH5C = N 3.825 Capronifcrile C BnC = X 3.189 Allvl thiocarbiuiideC8H5-N=c=s Phenyl thiocarbimide CBH5-N=C = S U93 4.077 Methyl thiocyanate OH8(S.CN) 4.745 Ethyl thiocyanate C J I S.('\ Ethyl Sulphide (ail , _S LI 76 3.420 89.39 ins. ;•; 139.75 I [ |. 5 1 5 717 59.5 283 281 439 51 1 (390 80.25 152.59 57.23 78.28 1 11.1 113.13 143.7 78.96 99.84 122.2 365 668 247 336 61 I 368 0t 3 245 332 593 5i 543 543 709 713 378 t67 537 379 t66 539 -21 _ 2 + 3 ~ ifcl £5 -1.12 -0.78 -2.93 -0.70 + 0.82 5 +0.75 -0.81 -1.19 -3.42 ± 0 + I + 1 - 1 ±0.00 + 0.57 + 0.26 -0.21 + 2 +0.3/ 260 K. IKEDA TABLE II.— Continued. Substance. K. V Molecular Volume. ^ : + Difference between I^V" and « u a) û Water H20 11.935* 18.74 110 98 ? The average difference between the observed and the calculated values of KVY, and therefore of K, in the first table is ± 1.85% ; i while the similar difference in the values of K*V118 in the second table is not more than 0.90 %> the agreement being much closer. The value of K calculated therefrom is, however, again about the same as that in the first table, viz., 1.8 %• These results are more satisfactory than those obtained by Schiff" s formula, which gives values of N differing from the observed values by 3.5% on the average; and as Nia proportional to K, it may be observed that the average difference between the experimental and the calculated values of K according to the formulae here pro- posed is only one-half of the similar difference which obtains between the values of N when Schiff' s formula is employed. In the first table there are four instances of great discrepancies, ranging from 7.4 to 9.0 %. Of these, isoamyl-formate has an exceptionally large value for I\\ it being about 6 °L higher than the average value of four of its isomers, while its molecular volume given by Schiff seems to be too high by two per cent., so that there is some reason to believe that a careful redetermination of these data might remove this discrepancy. The next case is bromine, of which it is sufficient to remark that this element might possibly have two capillary constants as its brother This is calculated from Frankenheim's determination. CAPILLARY ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 261 element chlorine, whose lower value applies to the cases where two or more atoms of chlorine nre closely united together. Of the remaining two abnormal cases, ethylenedibromide and capronitrile, I have no remark to offer. They must be looked upon as genuine ex- ceptions. In the second table the discrepancies are much smaller except in the case of bromine, but the remark made above about this element applies here also. The acids form an important group of exceptions in both tables as well as in Schiffs paper; this is probably owing to the abnormality of their molecular magnitude at the boiling points, and it would be strange indeed if the capillary phenomena did not show an analogous abnormality. The atomic capillary constants above given must be regarded as only rough approximations. The values for carbon, hydrogen, chlorine, bromine, iodine, etc in the first table may be higher or lower than the true values by 50 units, while the difference may be greater in other elements. The values of atomic capillary constants used in the calculation of the second table seem to be much closer to the true values, and may be assumed to be correct within 5 units. A better determination of these constants, or :i better choice of the coefficient of I', may give much better results than have been obtained from the calculations given above. The first formula can be applied to the approximate determination of the molecular magnitude of a liquid compound, provided its percentage composition as well as the probable nature of its atomic concatenation, the capillary height (//) for a tube of known internal perimeter at the boiling point, and the specific gravity (/?) at the same temperature are known. Suppose the empirical formula of the liquid to be Am Bn Cp where m', ri, p' may be integral or fractional : put this mass .1/ and divide it by p. and designate the quotient by V. Further, suppose the true molecular mass=»M. 262 K. IKEDA Then by the formula we have x(m'a + n'b-\-p'c+ ) Tr I m'a±n'b+p'c + hp = A = — ■ f^ . Hence x = (m'a-\-n'b+))'cJr }2 KW,3 For example, toluene contains 91.3 % °f carbon and 8.7 °/0 of hydrogen ; dividing the one by 11.97, and the other by unity, we get the empirical formula C7m6S Hejl=M— 100. The specific gravity at the boiling point is 0.776, so that V -128.8 and the observed value of K is 3.692. Substituting these values in the above formula we have (7.68x600 + 8.7x75); ftQn V- = : = .aOi). (3.692)2x(128.8)3 Therefore, the molecular magnitude is C7.G3x.939 / /s - . „:;r, r=C7_u H&M. This result is too high by 2 °/0, inasmuch as the calculated and the observed values of A.' show a difference of 1.1 %• If this difference be 2 % as it i8 most likely to be, then the molecular magnitude would differ by 1 °/0, which may be regarded as fairly approximate. But whenever this mode of determining molecular magnitude is applicable, the vapor density method can also be employed, so that this applica- tion of capillary determination is more curious than useful. The second formula seems to be of greater interest to chemists, for it shows the close relationship between the mode of atomic linking and the capillary phenomena. It can be applied with caution to determine the number of double bonds in carbon compounds, for whenever there is one the value ofl K1,18 is increased by about 19 units, which is quite a large quantity compared with the probable i discrepancy between the observed and the calculated values of K V118. Benzene has three double bonds according to this mode of calculation, while ally 1 compounds seem to have only one such bond. This is in accordance with the conclusion of Brühl from optical determination. CAPILLARY ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 263 In the cnse of terpene, it' we assume 1=196, then the result agrees with Bruhl's investigation, hut if F=186.3 then there can he no douhle bond, or at most only one. Quinoline has only five double bonds according to the accepted formula, while the calculation from the capillary heighl indicates six such linkings. Carvol also show s a similar peculiaril v. The capillary value of the oxygen atom varies in a very peculiar manner; in alcohols il is only 31, in ethers and acetals it is 11. while it attains its seemingly normal value in aromatic com- pounds where it is 55. In aldehydes ami acetones h is 7 1 55 + 19. The calculai ion according to this formula seems to justify the com- monly accepted constitution of paraldehyde CH,C _(VII . K*FLM .;:,t 6x44 + 12x21.5 + 3x44. 0 It may he objected that the value for the group I C < I L07 has been obtained by making the value for hydrogen too high, and that trie conclusion about the atomic linking arrived at in this way is not trustworthy. Bui the agreemenl between the observed and the calculated values of /v'-I11"1 for piperidene which has no douhle bond, hut which bas two atoms of hydrogen less than what a saturated compound oughl to have, seems to prove the correctness of the above conclusion. When well worked out the capillary phenomena are likely to atford some insighl into the atomic grouping of a molecule, but this can not be attempted with confidence until more accurate data are amassed. So far the claim of the formulas seems to be fairly made out, still there are many grave considerations which tend to show that the connections expressed by the equations are more accidental than 264 K. IKEDA essential, and warn us not to theorize too freely from any such hasty generalization. That there are more than one formulae, which have no essential relation to each other, to express one and the same connec- tion, seems in itself to be a strong proof of their being of an accidental nature. A thing (A) may depend entirely on another (-£>)■; but if (B) can influence a third (C), or is always (or even usually) accompanied by it, then the proposition connecting (A) and ((,') will sometimes have the appearance of a law. If it holds good for a great number of instances, then it may be assumed to be true within certain limits, ;ind can be used as a sort of a law, though theoretically it is not. Just as, in using a circle oi a certain radius to represent a short piece of almost any curve, we must assign a strict limit to the length of the substituted arc lest the error grow inadmissively large; so here, these accidental laws must not be stretched too far. The formulas developed in this paper do not apply to the case of water. This may be owing to the fact that water has a very small molecular magnitude compared with the substances treated of in this paper, so that it does not lie within the limit of applicability of the formulae. Or it may be due to the fact that water has a critical temperature so ab- normally high that it is not comparable with other substances at the temperature at which the vapor pressure is equal to a certain fixed quantity. Or it may be due to some unknown cause, as is the case with the magnetic rotation of light, where water has a far greater value than can be inferred from its composition. But one thing is certain that this abnormally high capillary constant for Avater is not due to the complexity of its molecular structure, for the larger the molecule, the less must be the capillary attraction, supposing the formulae to be true. This difficulty can, of course, be eluded by giving oxygen combined with two hydrogen atoms a certain arbitrary value, so as to make the calculated value of K agree with the observed. This, CAPILLARY ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 265 however, is begging the whole question, for there is no other com- pound with which to test the validity of this particular value. Schiff' s formula is far better than the new ones in this instance, for the extra- polated value of JV for water seems to agree tolerably well with the actual value. The next thins; to be considered is the mode of measuring1 the constant. The capillary height multiplied by the specific gravity and divided by 2 does not represent the capillary constant or the surface tension ; for according to the theory of capillary attraction pV=Tl cos i or V=a*l eus i. where V is the volume of the liquid raised in the tube, p is the density. / the internal perimeter of the tube, T the surface tension, ar the capillary constant . i the contact angle. When a circular tube is used as in the experiment of Schiff and most other investigators, £(=2lfr) might be measured with fair approxi- mation ; V can also be found oui with tolerable accuracy, but it is next to an impossibility to measure i in the tube method. As has been criticized by Volkmann, Schiff made an unwarrantable assumption that i is always zero in the liquids investigated by him, and calculated out or accordingly. But as the value of i in the various liquids investigated by Schiff appears to be pretty large, especially in the case of chloroform, what he calls the capillary constant of a substance at the boiling point cannot be accepted as such indiscriminately ; and the more so, since, as he himself points out, i changes with rise of temperature, and since the boiling points of many liquids given in his communications are somewhat high, i cannot be zero even where it is so at ordinary temperatures. Volkmann has calculated the value of i for all the liquids contained in the first communication of Schiff's, from the height of the meniscus given in the paper. But, as Schiff replies, the meniscus height is undoubtedly one of the most difficult of quantities to measure. Indeed, the recalculation made by Volkmann 266 K. IKEDA cannot be looked upon as an improvement, for according to it the value of % for hydrocarbons is sometimes very considerable, while that for chloroform is very small. Empirically, Schiff's work is of great value, for whatever may be the value of ?', the capillary height h is a well defined quantity and can be accurately measured, and the relation found between capillary height and chemical composition is valid whether the constant is theoretically correct or not. Still from a strictly scientific point of view, the investigation must be deemed very imperfect, involving, as it does, an incorrect method of calculation. It is, therefore, very desirable that the matter should be investigated anew with an entirely different method, if possible, so as to confirm Schiffs work, and to remove all inaccuracies as far as practicable. The tube method is said to be the most accurate, still there is the drawback above alluded to. The plate and bubble method of Quincke is theoretically good, but practically the results obtained are but rough approximations. The drop method is not well fitted for accurate determinations, and the same may be said of the contact plate method which has been employed, amongst others by Schall, for similar purposes. The ring method used by Duprè and Wilhelmi has been taken up by Prof. Yamagawa of the Imperial University, who has given an expression by which the error due to hydrostatic pressure may be eliminated, and which has been applied to the measurement of surface tensions of various liquids. Timbe'ry* has used a thin platinum ring to investigate the influence of temperature on capillary constants. He gives only one determination by this method for each temperature, while he takes the average of several measurements by Quincke's method. He says that the result obtained by the former method is so accurate and certain that no second determination is required to confirm it. This is an excellent account, and my own experience * Wiedemann's Annalen 30 (1887) s.545. CAPILLARY ATTRACTION IN RELATION TO CHEMICAL COMPOSITION. 267 fully corroborates it. When a ring with a very thin edge is used, the configuration for maximum surface-tension is most probably attained when i—o, so that the correction for the contact angle is not required. I have tested a few liquids by this method with a very thin circular platinum ring", and have obtained the following results : Traction on the Eins m grammes AVater L.668 Alcohol .526 Chloroform .6215 Benzene .6535 Water 1. 1.000 .315 .373 .392 Capillary Height x Specific Gravity t her- mometer Grade ■ Bp. Gr. 136.8 t2.5 48.4 45.1 Contact Angle Water=1.000. Temperature 27<-28\ 1.000 Supposed to be zero. .311 .357 .395 18° Supposed to be zero. These are bul rough determinations, the substances used being only tolerably pure ; still the difference between the results obtained by the tube method and by the ring method is very significant. The capillary tube used in these experiments was a broken thermometer of a good bore, which showed no difference in the length of a mercury thread about the places used for the measurements. As the bore is very small (about l/6 of a millimetre), no correction has been made for the meniscus. The ring had a périmètre of 69.8 m.m., and as it was very thin no correction lias been made for hydrostatic pressure. The contact angle of benzene seems to be zero, while that of chloroform is rather large. Tr seems, therefore, advisable to use this method of measuring surface tension. As air is said to have considerable in- fluence in depressing the capillary height in the ease of water, it is desirable to conduct the experiment in vacuum, and as organic liquids dissolve gases with greater readiness than water does, the error 268 K. IKEDA from this source may be considerable.* The second series of Schiff's experiments is reported to be free from this source of inaccuracy. Another consideration which tends to diminish the value of the formulas is the temperature of comparison. The choice of the boiling points seems to be rather arbitrary, the pressure of vapor chosen depending entirely on the accident of our habitation. The theories of Van der Waal may perhaps give some aid in determining the temper- ature of comparison, but it is also probable that it will not furnish very accurate guidance. The only way to do this properly is to investigate the influence of temperature on capillary phenomena by a thoroughly reliable method, to see whether it has any relation to the changes of density of the liquid and pressure of the vapor and other concomitant phenomena. All these considerations take away much of the apparent value of Schiff's formula as well as of those proposed in this paper, and call for a new and accurate investigation of the phenomena. The physical properties of all substances must chiefly depend on the chemical composition, and the science of chemistry must be regarded as being grievously backward, so long as she cannot predict these properties from the knowlege of the chemical constitution. In conclusion I have to return my best thanks to Professor J. Sakurai, for the great interest which he has taken in my work and for his valuable suggestions. '^^^r^a-2- * But this inaccuracy may be disguised by the smallness of the capillary constants in organic liquids. Determination of the Elements of the Sun's Spin. by S. Hirayama, Rigakushi. Since the discovery of sun-spots by Fabric-ins and Galileo, the elements of the rotation of the sun, thai is, the position of the sun s axis and its period of rotation, have been determined by many astronomers. These results arc tabulated in rlouzeau's Vade-Mecnm de l'Astronomie .and differ great I y with the different observers. The various values of the inclination of the sun's equator to the ecliptic lie between .. § 162 Rotation pp. 4! 1. Carrington, Observations of the spots on the sun from 1853 to 1831 pp.244. (3) Spörer, Beobachtungen der Soanenflecken zu Anclam (1874). (•I-) Wilsiag, Neue Bestimmung der liotationselemente der Sonne 1884. Astr. Nach. Bd. 107 pp. 277. 270 S. HTRAYAMA. The great discrepancies are apparent at a glance. To "find the corrections to the assumed vaines of Prof. Spürer. I took his valuable observations of sun-spots from 1SG1 to 1884, and selected the spots where motions were pretty regular, since , it is clear that for such a determination the duration of the appearance of the spots should be long, and the proper motion regular and small. The number of groups of selected spots amounted to 933, of which the number of regular spots is great for the time of sun-spot maximum. Method of Reduction. The formula" obtained bv differentiating the fundamental eola- tions (»iving the heliographic longitude and latitude of a sun-spot are too complex to be used for the reduction of the materials of many observations. I haveadopted instead the method due to Carrington, in which, since the error arising from the assumed position of the pole of the sun is obviously felt chiefly in latitude, and not much in longitude, the variation in latitude only is considered. Consider the celestial sphere, the centre of the sun being the centre of the sphere. Let A* be the pole of the ecliptic. Let V be the sun's true pole. Let V be the sun's assumed pole. Let PX=I and P'K=I'. Let N' be the assumed node, then the angle N'FS=Û\e computed longitude of the spot from the assumed node— a'. Let S be a sun-spot, and PS=8 ; P'S=Ô'. DETERMINATION OF THE ELEMENTS OF THE SUN'S SPIN. 271 Then considering the nearly righl angled triangle PQlJ\ we have ô'-ô P'Q PP' cos f PP'N + NT'S J PP' cos PP'N cos z'-PP' sin PP'N sin a' .'. (V — (5 =X. cos z'—Y. sin x' if we assume X -PP' cos PP'N, Y PP' sin PP'N. It A and 1 can be found from a series, or from many combined series of observations, then PP' y/X^+ï , tan PP'N ^. Bu1 X PP' cos PP'N PR ( N-NJ' sin I )' PP' sin PP'N= P'R=l-I' lli-iire we have the following formula: giving the true elements / I' + Y N N'+P.cosecl. I he above treatment gives ô'— <5 .V cos z'—Y sin t! as the equatious of conditions to lind A and Y. It is a tremendous piece of work to apply this equation to each individual sei of observations of spots. Hence it is necessary to consider in what way these quantities A' and Y can most advantageously he found from the whole ac- cumulated mass of observations. Since each series of observations yields a certain number of observed values of ô' corresponding to obsen ed \ allies of a', we may a>>um<' i he following series of equations. Ô„'— Ô=-X eus ô — Y sin it ôi' — ô-^X cos b — Y sin h fj / — <5 = X cos c — Y sin c By successive subtractions, we get Oft — b'a—X fcos b — cos a J — Y (sin b — siu a) ôc'— ô\=X (cos c—cos bJ—Y (sin c—sin b) 272 S. HIKAYAMA. as the equations of condition to determine X and Y; a, b, c, the computed longitudes of the spots from the assumed node being known. Also by interpolating other values of ô' for previously selected values of a' at equal intervals, we simplify the calculation exceedingly. I have found the variations of latitude per 10 degrees of longitude by simply dividing the variai ion of latitudes by the difference of longitude. For example, for the solar spot No. 4-1, which appeared in March 1882 (see page 282 of Publicationen des Astrophysikalischen Observatorium zu Potsdam Nr. 17) we lind 1 35 150 163 178 192 20G 222 23(3 250 201 277 90°— ô . -0.32 — 0.40 -0.48 -0.30 -0.39 -0.37 -0.37 -0.58 -0.75 - 0.80 -0.93 Variations ut' 90" — ô' per lu' of a'. — .j — 0 + 7 •> + 1 0 -15 -11 — 4 -M Thus — 0°.05, — 0.°06, are considered to be the variations of latitude between longitudes 110° and 150', 150° and 160", respectivelv. This method of approximation will be sufficient for the majority of the observations. Having found the variations of latitude per 10° of longitude, we then tabulate them in the following form. No. 44. .130° 140° 150° 100° 170° 180° 190° 200° 210° 220° 6 + 7 +1 + 7 -2 +i ; +i DETERMINATION" OF THE ELEMENTS OF THE SUN'S SPIN. 'i ( . ; T shall not reproduce all the numerical values obtained in this way for the whole interval of nearly 24 years. As an example, I shall give the complete method <>f reduction for the year 1884. In discuss- ing in this way Spörer's observations, I shall ofeour.se adhere to the numbering of the different groups he has «riven. Table I. contains the number of each Potsdam group and the variation of latitudes per 10° of longitude, the same result being represented in a somewhal altered form in fable II. on separate sheet. Table I. L884. Spot, a' Lit. Diff. Spot, x Lat. Diff. Spot, a' Lat. Diff. 2 72 12° + 11.1 + 20 240° 0 — 9.2 137 -17.1 — 1 3 58 + 1 1.8 + 29 136 - 9.3 0 152 -17.6 — 8 72 + 12.2 - 6 151 - 9.3 — 8 165 -17.7 88 + 12.1 164 - 9.4 - 6 35a '.".i + 10.7 + 2 16a 109 + 12.3 + 8 180 - 9.5 + 4 112 + 10.8 + 26 122 + 12.1 + 7 193 - 9.45 0 157 + 11.2 179 + 12.8 207 - 9.45 __ 2 356 97 + 9.8 _ •) 28 1 -2 - 8.4 0 235 - 9.5 + 10 1 to + 9.7 -20 117 - 8.4 u 264 - 9.2 1 55 + 9.4 — 25 175 - 8.4 -24 32a 109 + 6.05 -18 167 + 9.1 102 - 8.8 + 11 128 + 6.4 + 21 :'»7 66 + 9.5 -16 206 - 8.6 — i 142 + 6.7 , ,q HO + Id + 8.8 ii 221 - s.7 157 + 6.9 0 124 + 8.8 — 23 (53) 84 - 9.1 + 3 17(i + 6.9 + ; 137 + 8.5 — 5 113 - 9.0 0 198 + 7.1 _20 195 + 8.2 -15 141 - 9.0 — < 213 + 6.8 208 + 8.0 0 155 - 9.1 + 7, 1 ' :;:;,( 65 —16.6 221 + 8.0 169 - 9.0 0 80 -16.6 .g 1" 75 + 12.9 + 20 198 - 9.0 -20 107 -16.95 _21 90 + 13.2 i) 213 - 9.3 + 4 123 -17.3 — 7 102 + 13.2 - 1- ^74 S. H IK Aï AM A. Table I.— Continued. 1884. Spot, a' L'it. Diff. Spot, a' Lat. Diff. Spot. '1 Lit. Diff 159°j +13J» 192° + 12 A -14 143° + 7^8 + 13 38 91 +13.8 + 13 206 +12.2 _2] 158 + 8.0 + 7 105; +14.4 -25 234 +11.6 186 + 8.2 + 21 117 +14.1 + 2 47a 77 + 8.2 - 6 200 + 8.5 + 7 f 175 +14.7 108 + 8.(1 ■ i 2 12 + 8.8 (174 +13.3 -15 122 + 7.9 -13 58 93 - 0.4 + 6 (188 +14.5 1 137 + 7.7 - 7 109 - 0.3 -14 (188 +13.1 • i 166 + 7.5 -Il 151 - 0.9 (203 + 1 1-5 194 + 7.2 _13 60 121 + 12.0 + 18 (203 +13.1 +10 209 +7.0 0 102 + 12.8 .-16 232 +14.1 223 + 7.0 193 + 12.3 - 9 41 97 -21,0 + 14 47 b 163 + 8.0 _]0 248 + 11.8 111 -23.8 - 8 192 + 8.3 _M 63 130 + 10.7 - 7 123 -23.9 0 200 + 8.1 L60 + 10.5 _ ■) 178 -23.9 0 49 88 - 9.7 - 8 218 + 10.4 - 8 192 -23.9 — 7 101 - 9.8 -20 230 + 10.8 207 -24.0 -14 116 -10.1 0 66a 140 + 10.6 - 8 234 -2 1,1 145 -10.1 - 3 152 + 10.5 2 64 110 -25.7 -27 174 -10.2 + 7 193 + 10.4 + 15 140 -26.5 + j 188 -10.1 0 201) + 10.0 195 -26.3 l3 202 -10.1 72 138 — 7.0 + « 2U8 -26.5 g 55« 123 +14.9 + 6 151 _ — 7.5 + 5 250 —26.9 - 8 139 +15.0 — 35 194 - 7.3 + 7 263 -22.0 170 +13.9 -31 200 - 7.2 - 8 U 121 +11.8 0 186 +13.4 -32 222 — 7.3 0 : 134 +11.8 +20 214 +12.3 + 13 237 — / . 3 , 149 +12.1 f]( 229 +12.5 73 95 —16.0 + 2 178 +12.4 ( 56 113 + 7.9 ; - 3 139 -15.9 - 8 DETERMINATION OF THE ELEMENTS OF THE SUN'S SPIN. 275 Table I. — Continued. 1884. Spot. a' L it. Diff. Spot. a' Lat. Diff. Spot. a' I .:U. Diff. e 0 0 0 e 152 — 6.0 — / 756 122 + 1 1.8 ii 179 + 7.9 — •> 196 - 6.3 -in 165 + 1 1 .s — 6 208 + 7. S •jli' 6.6 g 182 + 1 1,7 ii 107 j ■> i + 8.2 -20 22 1 — 6.7 11 L93 + 1 1,7 1 17 + 8.4 h 253 7." i 5 1 17 _l 15.3 _!_ 7 161 + 8.4 h 98 133 — « .;> (i 159 + 1 5.6 12 219 + 8. ! — 20 t is / •> 0 176 + 15.1 ii 2:;i + 8.1 -36 102 / :> - 1 1 188 + 15.4 - 1 3 262 i 8.0 17»; i .■ • • (203 -;- 1 5.0 s 7 11!' — 7.1 ; 21 190 — 7.1 ._.,, i 202 + 15.1 + 7 1 18 — 8.3 - 1 1 201 — 7.0 217 + 15.1 1 < / 8.8 • 25 + 1 228 O.J |:î (216 + 15.5 269 — s. 7 2;;:; 0.6 : 21 [231 + 1 5.0 + 7 112 1 52 — 8.9 h 217 — I 0.3 230 + 15.1 165 — 8.9 7 :\h 189 1.2:. -flC 7^ 1 11 13.0 o 18 i — ■s. s • i 20 h - I,U II I -V _13.0 + 1: 208 8.75 218 I,n ! 08 -12.8 i 'Jh, 219 — 8.9 28 71.- 181 2.1 / 181 — 12. S — t 233 — 8.5 1 '.i 190 •) •> 13 196 — 1 2.9 2J.9 — 8.8 208 2.35 10 79 116 + 12.6 + / 264 — 8.9 i — -'îli 22:1 2.5 + t 131 + 12./ + 15 27s — 9. 1 13 2--Î7 — 2.1 (i 1 1 1 + 12.9 + 7 2'.':; — 9.0 252 — 2.1 159 + ! 3.0 + 23 93 1 11 — s. 7 o 7 on 122 + 3.6 — / 172 + 13.3 1 ■)• i — s. 7 - 105 + ■>.'■'> + o 80 107 + 7. fi — / 17" — 8.8 + 7 182 3.1 — s 1 .)•> + 7.5 + 15 ISI — 8.0 I'.H + 3.3 -10 13-5 + 7.7 + ln (185 — 8.7 -1 1 22:'. + 3.0 164 + 8.0 _ 199 — 8.9 _ 276 S. HTRATAMA. Table I. — Continued. 1884. Spot. Lit. Diff. Spot, a' Lit. Diff. Spot. 7. Lat. 1 Diff. dôb 1)86 996 213 - 9.0 228 - 9.1 243 - 0.1 256 - 0.3 270 - 8.7 280 - 8.7 2 0 2 -11.1 241 -11.2 251 -11.8 20 1 -12.0 220 - 1 1 ..s 238 - 1 1 .s 2 t9 -12.4 203 i i ■. -12.0 i i i ! 50 i i . i - 1 1, 1 171 - 1 t.o 188 -15.2 202 -15.1 21 G -1 1,0 180 -10.2 201 — 18.0 220 -18.4 230 -18.0 244 -10.2 175 -12.4 180 -12.4 203 ! -12.4 - 7 221° 0 231 -15 210 +43 o 280 302 101 182 — 5 100 _60 213 - 1 5 223 2:57 o — )■) + 20 -21 — • >:> -21 + 7 + 1 I. + 10 + 11 -2o - 1.3 0 o -22 00 138 _ « •> 313 208 2 1'.» 203 278 300 320 101 205 221 233 217 201 27^. -12.8 -12.8 -12.8 -13.2 -13.4 -10.0 -20.1 -20.7 -21.2 21.1 - 0.0 - 0.7 - o.i - 0.1 - 0.5 -1:5.5 -1 1.3 - 1 k3 -11,5 -1 1,8 -1 1.1 - 1 1,0 -11,5 -14.2 11.2 14,0 13.8 13.8 18, I 1 0// o o - o -15 -42 -35 -50 + 7 + 1:5 + 21 o -10 -10 o 125« -1:5 -10 + 30 + 7 + 10 Q + 14 + 14 o 0 280 303 317 333 318 210 2 17 278 290 308 225 237 281 309 323 251 2(36 282 295 310 324 1256 201 289 303 177 101 205 12; -13.8 ; -13.8 j -13.7 | -13.8 ' -13.7 - 0.3 — 0.0 — 10.0 -10.2 -10.1 -10.3 - 1 1,0 -14.89 -15.1 - 1 5.2 -15.1 + 8.8 + O.o + 0.0 + 0.0 + 8.8 + 8.0 j + 7.5 + 7.:5 + 7.1 + 12.6 + 12.3 + 12.5 o + ~ - 0 + 7 -21 — 20 -I 1 + « + 13 o o -13 + 7 -11 -17 + 18 + 14 DETERMINATION OP THE ELEMENTS OF THE SUN'S SPIN. 277 Table I. — Continued. 1884, Spot u L at. Diff. Spot. a' Lat. Diff. Spot. a' L •it. Ditï. o 0 o o 0 219 + 12.7 — / 319 — is.:; c 2:51 — 2.»» -1 1 234 _l- 2.6 + 7 •j 3 3 6.3 0 245 — .) .) — 20 249 + ■ 2.5 347 6.3 i 2 i 1 — 2.5 ~ 130« 232 _L i . i -11 351 6.3 Is 27!» — 2.0 — 9 24(3 + i . 5 — / •) — 6.5 2(JU — 2.7 + 8 260 + 7.1 171 213 < .■ > 0 303 — 2.0 + 7 ]:;■/. 250 + 6.4 0 22 7 t .3 -- / 318 — 2,_. 260 + i;.i < 241 7.1 ii 1 10 188 — 9.2 -20 275 + 6.3 2:,:. 7.1 + 7 2nl — 9.5 + 7 134 169 — 7.2 -17 269 i .■> -10 215 — 9.4 + 7 1S1 — 7.45 — :! 298 — 7.0 h 22'.» — '•».0 -11 199 — < .5 0 312 — 7.6 + ' 243 — 9.5 0 212 — 7.5 0 326 7.5 + 2'.» 257 — '.».5 — 7 227 — i .5 — / 343 — 7.u + 18 271 — 9.6 i» 241 — 7.6 — / 35 1 - (5.8 + 7 285 — 9.6 + 7 25-5 — i . < - 1 8 6.7 + 21 300 — 9.5 -11 282 — 7.8 + 7 •)•) - 6.4 31 I — '.'.7 + 10 29U — < . i -19 137 203 - 1.6 — i 329 — '.».5 + 15 312 — 8.0 -15 217 - 1.7 ii 342 — 9.3 325 — s.2 -14 231 - 1.7 2:; I.V.» 196 — 8.0 - 339 — s.i -■ 8 2 17 - 2.0 -21 210 — 8.1 — 5 ••; 12 — 8.5 261 2.:: -21» 22'.i — s.2 0 152 192 — * .•» -- 20 275 - 2.7 - 20 210 — S.2 ii 207 — 7.2 + 13 290 3.0 21 251 — s.2 0 -)•)•) — 7.0 + 5 304 — 3.3 -14 200 — s.2 + 14 292 — 0.7 + 17 318 — o.o + 1:'. 343 — 8.0 - 0 304 — <;.:; 0 333 — 0.0 351. — 8.1 -14 304 — 6.3 o 156 219 - 2.0 0 8 — 8.3 278 S. HIRAYAMA. Table I. — Continued. 1884. Spot. a' Lat. Diff. Spot. a' Lit. Diff. Spot, a' Lat. Diff. 141 179° + 18^4 -23 260 " -ïo.6 + 7 147 186° -15.6 +u 192 + 18.1 0 271 -15.5 -14 200 -15.4 _24 207 + 18.1 0 288 -15.7 +2() 217 -15.8 : _ 4 220 + 18.1 0 303 -15.4 -17 242 -15.9 + 7 234 + 18.1 -13 320 ~15-7 - 8 256 -15.8 o 249 + 17.9 - 1 4 345 -15.9 270 -15.8 -12 263 + 17.7 -14 143« 186 -11.6 -20 287 -16.0 + 13 277 + 17.5 -38 201 -11.9 -14 302 -15.8 290 + 1 7.0 ii 215 -12.0 0 ] 46 192 + 8.3 + 14 305 + 1™ + 6 228 -'-•u + « 206 + 8.5 -21 322 + 17.1 244 -11-9 0 223 + 8.1 + 4 142« 241 + 14.8 _ 6 258 -1L0 + 7 249 + 8.2 _14 2^ + 14.1 |+81 273 -11-8 -15 263 + «s.O — 7 273 + 14.6 0 28Ü - 1 2.0 — / 277 ! + 7.9 0 286 + 14.6 lis! 30 1 -12.1 _12 293 < + 7.9 _13 301 + 14.4 0 318 -12.:'» 0 308 + 7.7 318 + 1 1.4 -14 344 -12.3 ici'» 215 + 9.3 _20 :;j !• + 1 4.0 L44« 1 79 -24-1 -40 230 + 9.0 _25 1425 237 + 15.7 0 194 -24,7, , o1 1+31 216 + 8.6 _]7 250 + 15.7 + 27 207 -24.3 0 258 + 8.4 205 + 16.1 + 8 221 -24.3 -25 277 + 8.2 0 278 + 16.2 0 237 -21,7 286 + 8.2 -13 293 + 16.2 1445 184 -24.3 + 14 301 + 8.0 0 143 188 -15.2 -20 198 -24,1 + 8 316 ; + 8.0 + 8 203 -15.5 + 7 211 -24.0 -20 329 + 8.1 + / 217 -15.4 226 -24.3 ■Wo +8.2 + ( + « 231 -15.3 — 20 239 — -) 1 2 + '/ 118« 189 -15.7 6 246 -15 6 0 251 -21,1 207 -15.8 + i DETERMINATION OF THE ELEMENTS OF THE SUN'S SPIN 279 Table I. — Continued. 1884. Spot. a' Int. Diff. Spot. a' l.r. Diff. Spot. a' L it . Dili'. c o o o o 233 -15.7 0 2'il + 0.7 + 0 3 14 + 13.0 2 17 -15.7 — 7 273 + 9.7 -13 17:! 21'.' + 8.1 - 8 262 -15.8 0 2S0 + 9.5 23 231 + 8.0 + 13 278 -15.8 0 302 + '.'.2 2 16 + 8.2 - 8 293 -15.8 165 213 -12.3 -10 2:.'.' 4- 8.1 0 Lööfl 29 7 + '.».7 - 8 22'.» -12.0 1:; 288 + 8.1 0 310 + 9.6 t 2 1.5 -12.8 + 17 302 + 8.1 — I) 326 + 9.5 257 -12.6 -11 171a 339 + 10.8 + 13 155/, 282 + 11.3 ii 27*', 12.8 33 351 ; 1 1.0 + 50 293 + 1 1 .3 - 1 9 285 -13.1 + 7 8 + 11.7 — 7 309 + 11.0 -13.0 -13 23 + 1 1.6 157 216 + 6.5 + 9 315 - 1 3.2 - 8 171 2 1 1 + 10.4 -14 227 + 6.6 -13 328 -13.3 - 0 258 + 10.2 + 10 212 + 6.4 -20 344 -13.4 -17 271 + 10.5 0 257 + ''».1 ii 356 13.6 28 i + 10.5 2 7:. + 6.1 ) 7 _ / h;:./, 222 13.5 1 2 176 235 + 8.0 + 17 28 + 5.8 -1 1 239 13.7 - 12 2 17 + 8.2 + 7 300 + 13 251 - 1 1,2 ii 262 + 8.3 — 1 315 + 5.8 27" - 1 1,2 — 22 277 + 8.2 -13 L60& 201 + 1 1.3 + < 270 - 1 1, 1 202 + S.'l i) 215 +14.4 n 10s„ 227 + 12.1 -1 1 306 + s.i» + 53 2:11 + 1 1.1 2 11 + 11.0 + 21 321 + 8.8 164 .».>.> •>•>.) + «.'.2 + L> s 260 + 12.:! n 177 235 - 5.S -20 347 + 9.0 - 13 27'» + 12.3 -18 250 - o.l -11 1 + 10.2 ■ l:; 287 + 12.H + 53 208 — 1;.:; -18 16 + 10.4 302 -12.8 + ~ 2 7'.» - 6.5 -1 1 1 80 232 + 10.2 — 1 316 + 12.0 — / 293 - 6.7 + 13 2 17 + 10.1 — 20 331 + 12.8 + 15 308 — 6.5 -15 280 S. HTRAYAMA. Table I — Continued. 1884 • Spot. a.' Lat. Diff. Il Spot, a Lit. Diff. Spot. a L lt. Diff. 321° o - 0.7 + 12 324 + 5.2 - 8 343° + o 4.8 . + ~ 338 - 6.5 -33 ™ + 5-1 j -1 1 357 + 4.9 -14 350 - 6.9 -13 1,9 -13 11 + 1.7 5 - 7.1 -13 + 5.1 li)4d 313 + 6.7 + 2 > 20 — 7.3 M 188/ï 284 + 21 I 3.2 — (i 328 + 7.0 -2:! 34 - 7.0 316 + 3.0 -11 311 + • '..7 + 31 1 78a 300 - 2.7 ,, :!:!l + 2.8 -25 35 1 + 7.1 — 36 3] 1 — 2.9 + 12 :U,; + 2.5 — 29 8 _L_ 0.0 + 39 331 - 2.7 -Il 3ii0 + 2.1 o 20 + 1 .■> + 1 345 - 2.9 -2fl "; + 2.1 o 52 _!_ 7.1 360 3.2 -13 '•' + 2.1 196 2.*»:; - 0. 1 — t ,0 - 3.4 + 14 ,SS''-77 + 3.9 — t 268 — 6.5 — / 30 - 3.2 305 + 3. / -11 282 — 0.0 -18 199 20 1 — 3.0 s ' "' + 3 . • > o 290 — 6,9 -15 20(1 - 3.2 330 _u 3 . • i — 20 312 — 7.1 3<)7 - 3.1 310 + 3.1 — 0 107 35 1 — 5.9 0 317 - 3.1 + i + 3.0 + 1 1 270 — 5.9 + 8 33] _ 3.0 + < + 3 . : > 282 — 5.8 -1 1 346 - 3.] 0 194« 317 + 4.8 + 33 2'. m; — 6.0 + '!. 3 - 3.1 9 + 5.4 — 29 310- ' — 5.8 — 5 14 - 3.2 349 + i + 5.0 +46 329 — 5.9 + 5 29 - 3.1 2 + / + 5.0 351 — 5.8 0 43 - 3.0 . J 1 94tt1 1 8 -r i + 5.2 - 6 12 — 5.8 + 10 57 — 2.9 36 + 5.1 + 12 22 — 5.7 — 7 186« o - - -11.8 + 7 ';- + 5. 1 36 — 5.8 + 7 9 -11.7 + 18 liM' 2" + 5.2 — t 51 — 5.7 0 26 -11.4 314 + 0.1 0 66 - 5.7 188 c 307 + 4.9 HS |*»i + 5.1 -23 203 321 + 3.5 + 13 DETERMINATION OE THE ELEMENTS OE THE SUN'S SPIN. 281 Table I —Com imted. 1884. Spot. a' L IT. Diff. Spot, a' I /it. Diff. Spot. a' Lat. Diff. o 0 o O o o 336 + 3. i + 14 ■".2 + 1 7.3 + 6 229 213 — 3.9 + 19 350 + 3.9 + 8 68 + 1 7.4 + 58 32'.» — 3.6 -11 ■ > ■ » + 4.0 80 +18.1 357 — 3.9 207/) 1 1 + 1 11,8 21 217 336 -22.«.» -12 223 290 + 1.7 + 6 25 j~ t.o + 13 2 -23.2 8 307 + 1,8 18 to - 1.7 0 15 -23.3 - 1 5 318 + 1,6 -IS 56 -r 1.7 2s _23.5 + l 8 331 _L I.I ii 207/1 i n + 5.0 - i + 5 to 23.2 •Vi — ■' ° 9 + 30 3 17 i + I.I i •> -1 1 i * * "r ■ i . _ 21 — i i I i ._ - 30 + I.1.' + 20 (in —23.0 + 21 16 j_ 1.3 1 1 to + 1 5.2 20 83 -22.7 58 _j_ '■>. i — 6 60 + 1,9 , - 219rt 352 + 1-1 + 25 i 5 + 3.6 71 + 1 5.0 12 +13.9 — 8 228n 329 - 8.0 + 25 285 4- 1.3 2 1 +13.8 1 1 349 i .5 + / + 32 20'.» : 1.7 -13 12 +13.0 17 17 6.6 o 31 I + 1,5 - 4 54 +12.8 + 7 31 - 0.0 + 12 359 + I,:! ii "'»8 +12.9 + 13 51 fi.-l 12 13 + 1,3 + 10 83 +13.1 + 1 1 63 6.9 15 39 + 1,8 + 11' 97 +13.3 9< • — 7.:! — < 55 + 5.1 + 13 222« 289 + 5.1 0 L0-5 — 7.1 71 + 5.3 ii 316 + 5.1 0 229 313 — 3.9 + 19 83 + 5 . : ! 332 + r>-l _ | 329 — 3.6 1 | 227/, 282 + Lii.2 + * n + 5.C II ■\Ôt - 3.9 1 1 296 _L 16.3 i 11 + 5.0 -14 231 1 307 — 12, s -13 31 1 + 6.3 + : 25 + t.8 + 7 322 — 13.0 — 2'.' 353 + L6.5 + o in + 4.9 — / 339 — 13.5 ii 10 + L6.6 + !•' 54 + 4.8 ii 351 — 13.5 -11 36 + 17.n + 1Î 68 i + 4.8 is — 13.8 0 282 S. HTRAYAMA. Table I. — Continued. 1884. Spot, a' Lat. Diff. 1 Spot, a' Lat. Diff. 1 Spot, a' Lat. Diff. 32° — 13Î8 J) 262« + * 33° + 7°.Q 0 94° + 3,3 . + - 10 -13.7 -i3{ 02 + 7.6 -17 î 08 + 3.4 — / 01 -13.9 - 9 92 + 7.1 - 2 123 + 3,3 + ? 105 -14.3 135 + 7.0 1 38 + :]A +30 2:31 a 340 -13.8 4- 7 262b 00 + 8.1 + <; 1 18 + 3.7 lu -13.6 + 13 91 + 8.3 + 2 20!» 19 2<; -13.4 0 133 + 8.1 18 - »•* + 7 11 -13.4 ,_ 265<( — / 92 -15.1 + 7 02 - ,i:1 o 56 -13.5 + 1 1 106 - 1 5.0 + -S i i - °-8 o 1 00 -12.9 119 - 1 4.9 + 6 91 - 6.3 238 a 4 + 1 1,8 0 1 30 -1 1,8 0 270 ,;; + 4,4 ... 18 + 1 l,s + 6 149 - 1 1,8 28 + 1,2 7 + / 49 + 1 5.0 - 4 200 13 + 15.1 -18 12 + |:! 0 77 + 14.9 ii 60 + 1 1,8 -47 5-1 + 4-3! + )ß ss + 1 1,9 75 + 1U +20 07 + 5.0 238/3 17 + 1 1,3 -lu 90 + M.4 ^ 27 1. 21 1-8 + 13.8 1 1 9 + 13.8 .)•) -12.1 , r + 1.S + * + 6 70 + 14.3 0 131 + 13.9 + 33 ■s 7 87 + 1 1,3 1 19 + 14.4 118 -11'7 +12 253« 1 + °9.<; * . + 13 200/, 41 + 1 3. 1 -53 135 -11.5 17 + 0.8 + S 56 +'2-"1 -50 2i»3 71 - 0.8 — i 30 + 9.9 70 + 1 1 .9 117 - 7.1 + 1 1 + i 0 14 + 10.1 85 + 1 1 + 12.0 + 10 |(i() - 7.1 58 + 10.3 11 1 + 12.3 2 77 50 - 7.0 + 3(5 + 12 + •> 72 + 10.8 + 13 128 + 12.9 -I:'. 113 " 6-7 + 8 102 + 1 1 .2 + 19 113 +12.7 181. - o.i 118 + 11.5 + S ,5° 1 130 + 11.0 208 05 + 3.1. o ) - à DETERMINATION OF THE ELEMENTS OF THE SUN'S SPIN. 28; Table II. gives, as final results, the total sums of the variations of latitude per 10 degrees of longitude and the corresponding total number in the year 1884. In like manner, similar sums were obtained for each year from L861 to 1884. The results of the whole are tabulated in Table III. In the last three rows of Table HI"., the results of 24 years are given, the last row giving the mean value of the variation of latitudes per 10 degrees of longitude. Supposing the weight of each column to be proportional to the total number in the corresponding column, multiply each of these mean values by its weight. Sum the whole. and divide by the sum of the weights. There results a mean excess of — 0°.00217, which would imply that on the whole there is an average tendency toward the equator of nearly 8" in the time during which the >uii rotates through 10 . But considering the approximate character of the method of reduction and the probable effects of the err» >rs of observations, we are. I think, justified in altogel her neglecl ing this result. To annihilate this equal departure on either side of zero, I have deducted this quantity 0°.002 from the mean values in each column. Su thai we have finally the following; data to lind A and Y. 0° 10° 20° 30° K.i° 50° 60° 70° 80° '.mid Differences + 1.4+2.4+3.7 Wci-ht 272 261 245 + 4.0 + L5+ 3.5 + :;.! + 4.6 238 23(i 223 208 201 + 2.8 190 Differences Wei -lit. . 90° luii0 1 lu° I2'iu 130° I I"' 150 160° 170° 180° + 2.2 + 2.1 - 0.2 + 3.1 + 3. 7 + 0.5 - 1.4 + 0.9 182 180 .70 is; 195 197 203 223 231 284 8. HIKAYAMA. 180' 190° 200° 210° 220° 230° 240* 250° 200° 270° Differences Weight. . - 1.2- 0.3- 1.0 256 27»; 2H0 - 1.1 294 0.2 308 - 1.4- 2.5 316 318 - 3.2 804 270° 280° 290° 300° 310° 320° 330° 340° 350° 360° Differences. Weight. .. 6.2 305 4.8 291 3.0 _ 2.1 - 0.9— 1.6 - 1.2 + 1.3,+ 2.0 ■2\)-2, 286 281 282 278 281 281 From the-se we get the following 36 equations of condition, (-015 X-174 F+ 14=0) x 2.72 (_04.5 X — 168 F + 24 = 0) x 2.61 (-074 X-15S F + 37 = 0)x2.4o (-100 X-143 F + 40 -<<; x2.38 (_123 X-123 F + 45 =0)x2.30 ( -1 13 X-100 1*+ 35 =0)x2.23 + 015 X + 174 F-12 =0) x2.56 + 0 15 X+168 Y— 3 =0)x2.76 r + 074 X+158 F— 10 = 0) x 2.80 ;+l(H) X+l 13 F-ll=0)x 2.9-1 + 123 X+l 23 F- 2 = 0) x 3.08 +_ i |:j .v+ 100 F-1 l-=0) x3.10 (_158 X — 074 F+ 31=0)x2.08 (+153 X + 074 F — 25 -0) x 3.18 (_168 X-045 F+ 13 :0)x2.01 ( + 168 X + 015 F-32 = 0) x3. 17 ;_171 X-015 F+ 28=0) x 1.90 (+174 X + 015 F-25=0)x3.04 (__ i 71. X + 015 F+ 22=0) x 1.82 (+17 I X-015 F-62=0) x 3.05 (_168 X + 015 F+ 21=0) x 1.80 ( + 138 X + 015 F-48 = 0)x2.91 ( _ 1 58 X + 07 1- F- 0 - - 0) x 1 .7(5 ( + 1 58 X + 07 I F— 32 = - 0) x 2.92 ^ _ i |.:j x + 1 00 F~ 2 = 0) x 1 .87 ( + 1 43 X + 100 F - 2 1 = 0) x 2.80 (-123 X+123 F+ 31 -0)xl.95 ( + 123 X + 123 F- !» 0. x 2.81 1 - 1 00 X + 1 13 F+ 37=0) x 1 .07 ( — 07 1 X+158 F+ 5= :0)x2.03 ( — 045 X + 168 F- 1 1 0)x2.23 (_015 X+l 71 F+ 9 =0)x2.31 ( + 10O X+l 13 F— 10 0) x2.82 1 + 071 X+158 F- 12=0) x 2.78 ( + o 45 X + 1 68 Y + 1 3 = 0) x 2.8 1 ( + 015 X+174 F+20 = 0)x3.81 The solution of these 36 equations by means of the least squares gives X= +0M8051 F= +o°.038l2 Hence Ô '- ô = + 0°. 185 cos (a + 1 2°.0) Table II. V" |6n , in 4., 3,1 .ii 7ii 3 m- 1: ''ii 13" h;, 17,1 13', 13:: 200 210 220 S30 240 KO »00 27: BW) 200 : 130 lin 350 :n;n T -'"' 4- 3.3 -"' - 0 - 1 - 3 1 : + 14+14 33 4- Il + 3 + 3 0 0 Û - 7 4 7 3 0 + 10 + 10 13 13 + S1 us 0 1 13 • 2 t 24- 2 + 2 +26 , 2i 35,. 2 3 2 ■l - :'d 37 -111 -16 1., 1<: (i 23 -83- 5 - 5 - .-. - 3 ir, u + 21 h 4 13 23 -27.+ 2 + 2 + 2 + 2 3 2 13 - 1-, 0 - Il 3 m + 10 11 4 14 -s 0 27 27 -27 + 4 4- 4 4 4 4 1 + 4 ,; i:i - 14 :i - 14 - Il H - n ■s 0 + 2< -211 + 1" 4 lu 21 - 21 - r - 0 3 - 7-13 -î: - Il - Il - 1] -13-13 10 - 1« - lu -14-14 19 3 -20 -21. 1 + 7 " 4- 1 -3,3 -3,3 3.3 32 32 + 13 + 13 50 5! 4 13 + 13 2 ; + 7 + 0 + 0 Il U -14—14 +48 y iS + (8 -43 10 - 13, - li; 0 ' : - : : : 03 56. + 5 4 7, + 5 - 2 - . 1.- R 1 ( > 2 s 7 11 - » 11 - 10 - Il 0 ! - H - 14 2:1 . 2: + 24 741 , , , _ . „ + r. - s + 10 + 10 7 11 u 11 1 - 1' H + 7 < 75° u ( + 7 - 12 2 13 + 7 + 13 - 13 n + 1.3 + 7 - 7+15 -Hl Till + 1" : 107 S7 -20 0 0 0 1 - 21 - 3( - 3G 24 - 24 14 14 - 14 + 1 + 1 0 + S + W 1! 01, 93 - 14 -14-7 15 + 43 »ni. 21 21 33 - 21 - 21 + 7+14 + 14 ( - 5 l - 55 + 2E 07 + 4n + Il + 11 - 43 - 22 1 ! - ï) : 15 31 + 7 lis« 4,3 4 13 4 21 21 T" ,! 0 ( 0 10 - 10 H 11 + 30 130 4- 7 + 11 + 10 2* + 14 + 14 ■ si - M -14 ,4 + 8 -11 + ,; \ ; - f + 7 + 7 1111/ 1LT-; h 13 0 127 + 18'+ H + 14 ie 130u - 14 1301 0 134 - 17 0 - 4 i u 14 -l 152 + 13 + 5 + 5 5 - E - 17 ! I ( 171 21 11 M 0 - s. + 1H 137 - 21 21 -2Ü 1: 14 + 13 150 - n "'. 14" 150 14 7 - r ! '; 1 ; 1 f 1 14 1 | 14 14 ; |: + 11 + 1. + 1. , NI 0 - i:i 14 3f t- 1 142n ■ 1 1 (-:! 13 « 14 - 1- 142/ + S 8 113 f ï J< 1 - S S ■_■■- 1. 1 0 4- 7 - 15 12 IL ( 144 d+sii+a 141' - Ufiif 6 2< L'' + 8 147 + ? 1 U 4 1 140 [ + 14 + 1-1 - u - i: 11 160 2i 17 t i: 0 0 141« G C + i + 4 ( - 7 1 l l. '""" 1 II I r. G 150 + ! ± 0 J7 M + + 13 ieoi H3 4 13 4- 7 + 7 180 21 13 - If) 1! 3: 11 ■s s - G 1031 +2 1 160 14 + 21 18 in 4.7: + 17 173 8 .- 1: - i: 3 1 1 1 1 74,: .3,. + 1! 1 15 1 51 174 I7i, + 17 " 1: 1 7, + .V 177 -13 4 2 1 31 2, 1 l- i- + 13 - 15 + 12 li 17«.t 1:1:1 - a 8 ■ s + 6 + 0 + ' + 7 I860 14 4- 13 13 t- 18 -, B -14 ','" 133, 188« 1 - 7 0 1: 2! 1331, ■ 4 II - 11 H ■i: 11.4, , - G 2! + 41 134 -H 1 ( -31 4 SI 4 31 4- 4 - 7 - 18 1« 100 . 7 M - : 197 + 1! f e + i 2113: 207 4 5 îï 13 i - 4 - < 227/ 5 + 7 0 + S + •' + 5 4 1 217 \ '[ ' Il ■ "• 2 H4 - 14 11 12 4 27 222- 0 1 22:1 , : + o + H fil ~v -14 -14 323 223, < 32 1 3 - 1 ' 15 .'- : 32 - Il 11 1 1 ■ - 1! i: ■21 21 11 230 - " ■ 1 1 r 7 2 13, : 1 4 1 ' 1 1 1 - 1 ,' ; 4 13 4 13 4 II + 10 - „ 2';:, + 8( + 2'.,, • 1 .' - 1 1 S - .:' +4" U 13 - 21,3 ; [00 B70 1 1 -," 4-1 Il 41 >1 t 1 : 1 ■ 4 6 4 1 268 n + i 5 . ■ it u 1:.-- 127 15 1 + 13 5+3 t.' 2 +: 4 : -4 - - 05 - 4 +60 -s ,4 ' im —157 H. 101 .37 7: +112 +82 + 6 -12 1 21 2 1 2 : 2 2 1 2 : 2 2 - ■ 3 2 ; ,; 3 30 ( 1 3 3 s * * -1 i. ' 3! 31 -41 31 2.7 36 32 31 DETERMINATION OF THE ELEMENTS OF THE SEX'S SPIN. 285 Also / :r+F=6°.968+0\038=6°.996= 7°.006) for 1861 N = N'+X cösec I = 74".523 + Ie. 480 = : 76°.003 ) Here we sec that the value of the node is larger than any of the values given at the beginning of the paper. The value of the inclination is nearly the same as thai given by Sporer. As a check on this result, I calculated the variations of latitudes at equal intervals of 30° degrees of longitude, by summing the three successive columns of the Table III. and dividing the results by the average number of spot occurrences. Also to see if there is a notice- able secular change in both node and inclination, I divided the interval of 24 years into three separate periods : first, from 1861 to 18G8 ; second, from INCH to 187!» : third, from 1880 to 1884. Separate reductions were made for these three distinct periods. The resulting- differences of latitude and the number of spots are shown in Table IV. Table IV. 0 30 60' 90" 120e 150° ISO" 210" 240° 270° 300° 330° 360 g (Diff. _i_ 2 is . 383 + 406 + 75 + 78 + 85- 326 - 151 - 485 t24 30 + 560 00 ~. 1 J Nos. 46 42 39 3] 2s 37 54 56 57 53 50 48 w — ' Mesins. CO 1 + 5,4 1 9,] + 10,4 + 2,4 + 2,8 + 2,3- 6,0 - 2,7 - 8,5 - 8,0 - 6,0 + 11,7 11 -L. i.e. 8,3 + 9,6 + 1,6 + 2,0 + 1,5- 6,8 - 3,5 - 9,3 - 8,8 1,4 + 10,9 g /I.itl + 97] + 1147 + 989 + 244 4- 17 r.21 2S7 - 560 -1336 2194 766 + 01 00 S -! ) Nos. 8! 76 63 61 67 78 95 in; 124 116 101 95 ci Menus CO + 10,£> + 15,1 + 15,7 + 4,0 4- 0,3- 6,7 - 3,0 - 4,8 - 10,8 1S + - 7,6 + 0,6 ii + L3,0 + 17,2 + L7,8 + 6,1 + 2.4 - 4,6 0,9 -'7 - 8,7 - 16,8 .">..-> + 2,7 30 Ditï. + 542 + 1115 + .V.(4 + 246 - 1082 307 - 232 295 - 913 -1747 - 677 - 183 30 Nos. 125 IK '.IS 88 1.7 104 124 134 131 127 133 138 "" © Means. + 4,3 + 9,8 + 6,1 + 2,8 + 11.2 , 3,0 - 1,9 — 2.2 - 7,0 - 13,8 5,1 - 1,3 oc 00 + 4,4 ■ 9,9 ■ 6,2 + 2,9 + 11,3 + 3,1 - 1,8 2.1 - 6,9 - 13,7 - 5,0 - 1,2 ôo (Diff- _L 176] 1 2642 4 1989 + 565 + 1177 - 129 - 845 - 1006 -2734 -4365 1473 T* 438 oc 1 _ -! ] Nos. 259 232 200 179 193 221 1 27:? 306 313 296 284 281 — _' Means. + 6,8 + 11,4 + 9,9 -- 3,2 + 6,1 0,6 - 3,1 - 3,3 - 8,7 - 14,7 5,2 + 1,6 oo + 7,5 12,1 _ 10,6 f :;.'. , 6,8 0,1 - 2,4 - 2,6 - 8,0 - 14,0 - 4,5 + 2,3 286 S. HIRAYAMA. Table IV. gives the following four sets of équations of conditions deduced respectively from A, B, 0, and D, of that fable. These four .sets of equations I shall call A', 1>\ C\ and D' respectively. A'. ( ". (-loi X— 500 F + 46 ii) 1.15 ( — 134 X-5O0 F+ 44 = 0) 1.25 (-066 X— 366 F + 83=0) 1.05 (—366 X— 366 F + 99=0) 1.14 (-500 A' -1:31. F + 98 = 0) 0.98 (-500 A'- 1:54. F + 62 = 0) .98 (-500 X + 134 F + Hi 0) 0.78 (-500 X+134 F + 29 = 0) .88 (—366 X+366 F + 2<) 0j 0.70 (-860 X+366 F+113=0) .97 (—134 A + 500 F+ 15 0) (1.98 (-181 A + 500 F + 81 0) 1.04 ( + 134 A+500 Y- 08= 0) 1.8:, (+181 X + 500 F- 18 =0) 1.24 (+366 X + 3■> ,•> j) >> 1 *-' y )) A + 0°. 102:5 1' + 0°.077 A + 0°.2134 1' + 0°.060 A - + 0M692 F + 0°.004 A + 0°.1792 F + 0°.038 DETERMINATION OP THE ELEMENT? OF THE SEX'S SPIN. -' i These give the following results : - from A' N -7ö°.602 / 7°.04o » B' .V 76°.379 I 7°.028 » v Ar 7EM8I- I 6°.964 » & X 75°.992 I 7°.0()6 It should be noticed thai the values of the ascending node obtained from J)' serves to check the previous determination. The difference amounts to almosl forty seconds of arc which 1 do nol consider too much for this kind of determination. À comparison of the values deduced from A'. B\ and C, seems to hint at a certain increase in the /allies of node in time. This was assumed by Spörer, but I do not insist on this point firmly. [ also here notice that in his reduction of spot observations, Spörer has since 1880 applied a correction for refraction due to the sum's atmosphere; so thai the heliographic latitudes deduced are somewhal different. I have, however, used the deduced values of longitudes and latitudes without alteration. throughout the long series of 24 years. The principal conclusion to be drawn from the present calculatio - is thnt the values of the ascending node of the sun's equator given by Carrington and Spörer are distinctly smaller than the values now deduced. VVilsing's values, however, resembles mine pretty closely. Finally I venture to propose the loll »wing elements for 1861. / 7°.006 ) 1861 Ar 76°.003 I In conclusion 1 desire to acknowledge my indebtedness to Prof. Terao, Director of the Tokyo Astronomical Observatory for his kind and valuable suggest i' ms. DO On the Fineness of the One Yen Silver Coin. by Yoshimasa Koga, Rigakushi, F. C. s., and Osamu Yamagala, Rigakushi, Assayers in the Imperial Mint at Osaka. Introduction. The experiments recorded in the following pages have for their object the ätudy of the nature of segregation of silver in the one yen coin, and of the mode of taking portions for assay from the coin in such a manner thai portion o taken shall represenl the exact com- position of the whole piece. The subject is of considerable importance in the technical opera- tions of the mint, and also with regard to the provisions of the law which governs the issue of national coinage. It is comparatively easy to determine with exact ness the average fineness of a mass of silver coins, for they may be melted together in a crucible, and a small portion dipped out of the melted mass after a thorough stirring may then be granulated in wafer. The granules thus obtained will exactly represent the average composition of the mass, provided that the coins have been melted in sufficient quantity and with certain precautions in order to prevent oxidation in the process of fusion. 290 T. KOGA & 0. YAMAGATA The exact determination of the fineness of any individual coin is not so easy, since the silver-copper alloy usually employed in coinage shows the property of segregation to a considerable degree, the silver beino; concentrated towards the centre of the ingot made from this alloy. A small portion cut from the coin in a haphazard manner will not therefore be the true representative sample of the coin taken as a whole. Although the experiments strictly refer to the silver yen coin, yet the conclusions drawn from them may be applied equally well to many other coins, when the conditions of manufacture are similar. In order to obtain a clear understanding of such conditions, it may be perhaps useful to give a brief account of the process of silver minting in the Imperial Mint. The process of minting begins with the melting of refined silver with the right proportion of copper in a crucible, the metal being covered with a layer of powdered charcoal. When in perfect fusion, the bath is stirred up thoroughly; a small portion is dipped out and immediately let fall into cold water. The small shots, or granules of the alloy thus obtained are dried and reserved for assay, which determines the exact proportion of silver in that melt. The metal is then poured into cast iron moulds and transformed into long flat strips or " bars," which are taken out of the moulds and cooled in water. The bars are next rolled out by a pair of rollers, until they are reduced to the thickness of the coin to be manufactured. In the course of lamination, these bars, now known as " fillets," usually require to be annealed, an operation which is performed in a sort of reverberatory furnace, where they are heated to a low redness ; they are then taken out and immediately cooled in water, so as to reduce the oxidation of the metal to a minimum. The finished fillets are taken to the cutting machine, by which ON TUE FINENESS OF THE ONE YEN SILVER COIN. 291 circular discs or i; blanks " are punched out of them. A single row of blanks are cut from each fillet in the case of large coins, such as the silver yen. The blanks are next adjusted in weight, usually by means of an automaton balance. After being "marked" or compressed on the edges, they are annealed in iron pans, introduced into a furnace similar to that just mentioned. When taken out they are cooled in water and washed in very dilute sulphuric acid, by which the superficial layer of copper oxide is dissolved out, leaving the blanks with a clean white surface. After washing and drying, the blanks are ready fa- the coining press, where they are transformed into coins. Finally the coins thus prepared must lie tested as to the sound, and the correctness of weight and fineness, before they are allowed to be issued to the public. The dimensions of the silver yen coinage bar and of the finished til let are : Bar. Fillet Length. iiit.00 cm Breadth. 4.30 „ 5.00 cm Thickness. 1.25 „ 0.23 „ The dimensions, &c, of blank and coin are : One Yen Blank. One Yen Coin. Diameter 38.0 mm 38 mm Thickness 2.3 mm — Legal weight -116 grains, (26.957 Gram.). Legal tolerance in weight 1.5 „ above or below legal weight. Legal fineness 900 per mille. Legal tolerance in fineness 2.0 above or below 900. It will be seen from the preceeding account, that many other 292 Y. KOGA & O. YAMAGATA. causes than mere segregation of silver have more or less influence on the actual fineness of the finished coin, for some stages of the process tend to lower the fineness by the oxidation of the metal, while other stages act in an opposite way by taking away a part of the oxide formed. II. Distribution of Silver on the Coin. The first series of experiments consisted in the endeavour to discover the average fineness of the One Yen coin produced from bars, of which the assay of granulation was known, and to examine extent the of segregation by which different portions of the same coin differed in composition. We selected four bars from four different melts named A, B, C, and D, the granulation assays of which were carefully made. These were rolled down to the required thickness as usual. Three blanks were cut out from each of the fillets, one from the middle and one from each end. These were marked respectively, A2J A15 A3 ; B,, B„ B,, &c. &c, the direction of lamination of the bar being carefully marked on each blank. These blanks were then treated exactly the same as the blanks for regular coinage. Care was taken in annealing to place each of them among separate lots of ordinary blanks, so as not to allow the effect of any accidental irregularity to fall simultaneously on all of them. The loss of their weight in grains in annealing and pickling is shown in Table I. ON THE FINENESS OF THE ONE YEN SILVER COIN. 293 Table I. Showing Loss in Annealin g- Mark of Blank. Weight before Annealing. Weight after Annenling and Pickling. Loss in the Operation. A. 415.7 1 H5.65 .09 'A,. 415.98 415.89 .09 4, 415.87 115.78 .09 Bv il»;.. m 416.39 .12 B,. 115.80 115.7»! .10 B* 415.83 415.73 .10 Ci. 1 15.» il 115.52 .09 Cr »6.06 415.97 .09 c3. »5.63 »5.58 .10 Dv »5.92 »5.84 .08 ]>,. »5.76 »5.69 .07 1>, 416.21 416.13 .08 Each blank was then marked with six lines parallel to the direction of rolling, and cut into seven strips of equal breadth which were named as in Figure 1. Fid. 1. middle line of original bar. 291 Y. KOGA & O. Y AM AG ATA. The narrow strips thus obtained were cut into smaller pieces and assayed. The results of the assays are incorporated in Table II. Assaying was conducted by the usual volumetric method. The proportion of silver was estimated to tV of 1 per mille. When it is given to the second place of decimal, the number is the mean of several experiments. r— I CO Eh d CO t— I -1-3 o CD en CG CD d ■H to ■H o 02 ejisodraoQ joobujoay ,—. o> e5 -r. d u Ö V o< V O < Ç5 -i 1^ CO — i 43 1 ö 3 O S ■*§ / uS 1 -W » c3 J3 ^ / 3 2. / a / S5 ü >> / o,a "^ CO / 1° CS / CO J cà 03 Ul (3 ec c: ^ M e3 =4-1 o a o a ci a> Ol es g H .2 d ON THE FINENESS OF THE ONE YEN SILVER COIN. 295 In order to show the distribution of silver on the coin, we produce below the details of assay of the blanks (7„ G2 and C33 the figures which denote the fineness indicating as nearly as possible the relative positions of the respective portions. Fig. 2. I. II. III. IV. V. VI. VII. Brank (' 8987 8988 8989 900-8 9000 8996 8995 899-8 ' 9010 901-2 9006 9009 9011 9018 902-2 901-9 902-2 9032 902-7 'J 02 6 903-1 9012 9012 9010 901 7 9004 9009 9017 899 C S99-1 899 4 S99 7 899-6 899-7 8989 8990 89S-0 I. II. III. IV. V. VI. VII. Fig. 3. Blank C. 898-8 893-1 8981 8987 8990 / / 9006 900-3 CC0S 9C05 900-8 9029 9018 902 7 C02 7 9031 9029 C02-5 903 0 9010 900-6 CC0-6 9011 S990 8994 898 9 296 T. KOGA & O. YAMAGATA. Fig. 4. Blank C,. I. II. III. IV. V. VI. VII. 8980 8988 80S 2 " 899-6 809-6 89D-6 901-1 900-9 901-0 900-8 9025 902 7 9020 9012 1)01 9 9020 S00-5 . 900-6 900-9 899'9 8997 899-5 S99 5 "\ 898-9 899-5 898 8 . Cx. C2. C3. Average of all assays. 900*617 900*787 900-331 Highest assay. 903*2 903-0 902-7 Lowest assay. 898*6 898-1 898*2 Difference between highest and lowest assays. 4*6 4*9 4*5 The preceeding results furnish fresh illustrations of the well known fact, that, in an ingot made of the alloy containing 900 per mille of silver and 100 of copper, there is a segregation or concentra- tion of silver towards the central line of the ingot. The middle portions in such an ingot contain the highest proportions of silver, while copper predominates along the edges. We observe in the blanks under .examination that this difference may be as much as 4*9 per mille, as in blank C,. It is also to be, observed that even adjacent parts lying in the same line may show a considerable difference in composition, which is probably due to several causes. We may, however, regard each of the strips which we have prepared as of one fineness by taking the mean of all the assays made from it. Taking 0NT THE FTXEXES3 OF THE OXE YEN SILVER fOI.W 297 tho strip IV as tlic middle, the strips I, II and III, lie on ore side of the central line at the same distances as VII, VI, and V, respectively on the other. Comparing- the numbers corresponding to each of these strips, we obtain the results shown in the following table. Table III. Showing the difference of fineness between two similar portions of blanks. Mark of I)ifl('rence in fineness between strips. Blank. i— vir. II— VI. 1 1 1 -V. A,. 0-20 0-40 1-02 A,. 0-15 0-57 (»•42 A3. 0-63 1-55 2-60 I'- 0-33 0-72 0-98 ll. 0-00 0*98 1-65 B3. 0-05 0-38 0-77 C, 0-03 0-43 0*05 C,. OC,7 I/o 231 (', 1-53 0-02 0-41 A- 0-02 0-19 0-13 D,. 0-50 0-78 1-26 D,. 0-40 0-3!) 0-40 AVe may conclude then that the metal lying in the central line of the original bar shows the highest average fineness, while on both sides of this line, the fineness gradually decreases towards the edges, 298 Y. KOGA & O. YAMAGATA. and the strips which lie parallel to and equidistant from the middle line have nearly the same fineness. The correspondence is not strictly exact as may be seen in Table III. The small divergences may be due, apart from slight irregularities occurring in the metal itself, to the fact that the strips, into which the blanks were cut, were not, from instrumental causes, mathematically corresponding divisions of the original bar. In order to eliminate, as far as possible, the effect of such irregularities, we have taken the mean of the several blanks as given in the last column of Table II. Thus we obtain, so to speak, an ideal composite blank or coin the fineness of the several paris of which is represented in Fig. 5. Fig. 5. The fineness of the entire blank— 90O3. We also see that on an average the coins produced from a melt are about 1*3 per mil higher in fineness than the assay of granulations ofthat melt, as shown in the following statement : — A. B. c. D. Mean difference. Assay of granulation of melt Average fineness of blanks 899-10 900-427 898-75 899-899 899-35 900-578 898-75 900-325 Difference 1-327 1-149 1-228 1-575 1-319 ON THE FINENESS I »F THE ONE YEN SILVER COIN. 299 III. Methods of Cutting Portions for Assay from the Coin. If the figures ) 2 Y. EOGA & O. YAMAGATA. melted together. But this agreement, it is needless to remark, does not sufficiently prove the correctness of the individual assays. The remark however applies equally well to any other method. We refer to this, because the said agreement has been put forward as an argu- ment in favour of this method. Method d. This method, due to M. Stas of Belgiern and in use in the mints at Brussels and at Paris, consists in punching out small pieces from certain portions of a coin by the single action of a special much inc. These pieces put together make up an assay. The number, positions and forms of the punching* differ in different denominations of coins and have been determined by experiment and calculation. In the case of the 5 Franc Silver coin, similar in size to the Silver Yen, they are as represented in Figure 7. Fig. / . The punching* taken from inner parts are distinguished by a square form. This method as applied to the Silver Yen coin appears to give re- markably exact results. We discovered however that the cutters beipg necessari y of a slender form frequently meet with serious ac- cidents, so that after trying the machine for a considerable time, we were obliged to give it up and substitute for it another and simpler form described in the next paragraph. ON THE FINENESS OF THE ONE YEN SILVER COIN. 303 New Method. A careful study of the distribution of silver in the coin has induced us to choose for our assay of the Silver Yen, three circular punching* of equal weight taken from the angles of an equilateral triangle, which may be supposed to he described on the coin, the distance of the centre of each punching from the circum- ierenceofthe coin being made equal to one-seventh of the diameter. Portions of metal taken from any set of such positions when put together, give almost exactly the average fineness of the entire coin, of which therefore they are the true representative sample. Thus in Figure 8, let c be centre of the eoio, ml be its diameter and the middle line of the bar. Let cc', diameter of the inner circle, be equal to *- ml, and let equilateral triangles be described in this circle as shown. Fig. 8. 304 Y. KOGA & O. Y AM AG ATA. Then from the nature of segregation in the coin, it will he readily seen that, c — c'; a=a ; b = b' = b" = b"' ; and d = d'=d" — d!". Therefore the small punched out circles at a,b," b'" = the same at a,'b,b'; and the circles at c ,d,d" = thc same at c'.d.d". In the same way it may be shown that the circles at e,g, /, have their equivalent set. Now by interpolation from a curve based on the numbers in Figure 5, we get the fineness of the parts marked a, b.c &c. on the ideal composite coin as follows : a = a = 899-0 b = /)' = b" = b'" = 90O8 c = c = 902-3 d = d' = d" =d'" == 899-3 e = 901-7 / = 899-9 g = 899-1 The three punchings may be any one of the sets above indicated or those lying between them; the resulting fineness of the sample will lie nearly identical with each other and with the exact fineness of the composite coin, as shown in Table IV. Table IV. Showing the Fineness of three portions in the ideal composite Coin of fineness 900-3. Position of Punching-. Fineness. Average Fineness of the three Portions. c or c d'" or d" d' or d 902-3 899-3 899 3 900-3 ' e (1 f 901-7 899- 1 899-9 900-23 1) or b" a or a b' or V" 900-8 899-0 90O8 900-2 OX THE FINENESS OF THE ONE YEN SILVER COIN. 305 Numerous double assays of individual coins by this method have been made. The results do not indeed show absolute coincidence, but the agreement has been in most cases quite close. Thus double assays made by this method on 105 coins gave the following; percentage occurrences of differences : Difference nil = 1 Ü" 1 * 0 Difference of between 0*1 and 0.5 per mille = 5!>0 ,, „ 0-6 and L-0 „ = 181 „ ., ., „ 1-1 and 1-5 „ = 2*8 ., 1-6 = 1-0 „ The correctness of the results obtained by this method has been verified by us in many instances by the direct assay of the winde coin, after taking the portions for assay by punching, the coin being cut up into small parts and divided into as many assays as there are pieces of metal. In other instance- the entire coin, after the removal of the punchings for assay, has been dissolved in nitric acid and made up with water to a certain known volume, an accurately calibrated bottle of about 2 j litres capacity being used for this purpose. An aliquot part of this liquid was taken up in a Stas pipette. The liquid was then transferred to a shaking bottle and assayed after having been evaporat- ed on the water-bath to the same degree of concentration as in ordinary assays. The results have shown in all cases a close agreement be- tween the assay on the punchings and the exact fineness of the coin ascertained in this way. 306 Y. KOGA & O. YAMAGATA. Table. V. Showing the results of assays by the new method on coins whose compositions have been accurately determined. No. of * Experiment. Assay on Coin. Exact known fineness of j coin. No. of Experiment. Assay on Coin. Exact known fineness of coin. 1 899-8 899-75 il- 899-7 899-82 9 900-8 900-07 ls 899-9 900-61 0 900.3 900-62 10 900-7 900-78 4 899-8 900.37 17 899-8 900-33 5 900-1 900-17 18 • 899-7 900-31 6 900-1 900-25 19 900-3 900-12 7 899-6 900-15 20 900-5 900-50 8 900-0 900-30 21 900-1 900-59 9 900-3 900-20 22 900-6 900-59 10 899-6 899-80 23 900-1 900-81 11 900-6 900-58 24 900-5 900-39 12 899-0 899-75 25 900-2 900-18 13 900-0 900-11 26 900-0 900-39 The assays of granulated dips taken from (he melted mass of a large number of coins, which had been individually assayed by this method, also «hew a close approach to the average of the individual assays. In melting, special care was taken to cover the metal with powdered charcoal. OX THE FINENESS OF THE ONE YEN SILYEB COIN. 307 Table VI. Showing Assays of Mass compared with Average of individual Assays. No. of Number Coins melted A.ssa; ( Iranulated Mean of in- dividual Assays Experiment. together. Dip. 899*90 of all the Coins. 1. 130 899-90 2. 50 900-10 900-04 3. 64 900-00 900-08 A small screw press worked by hand is used by us for cutting out the small circles from the coin (Fig. 9.), one circle being punched out nt n time. A steel punch of the required diameter is attached to the screw. The ruin is introduced by means of a brass sliding guide. The constant position which the guide is caused to take brings the coin to the exact relative position with regard to the punch. The latter descends by the action of the screw and cuts oui n circular piece from the required position. The coin is then turned round Through an arc of 120° in the guide by bringing the hole made by the first" puuching opposite n mark on the guide. A second action of the screw punches out another circle. The coin is turned round another 120° and the third circle is cut out. The process is simple and may be performed in less than half a minute The somewhat large size of the cutter, or punch, and the very simplicity of the apparatus make any accidents extremely rare. In this apparatus, the punch strikes out circular pi if almost exactly the same weight. In default of the punch, we recommend the 308 Y. KOGA & O. YAMAGATA. following method of cutting which may be performed with a pair of shears. It consists in cutting three equilateral triangles as shown in Figure 9. Fig, 9. Fig, 10. The centre of gravity of the triangles should be upon the cir- cumference of the inner circle whose diameter is equal to £■ of that of the coin. In order to facilitate the cutting of such triangles of equal weight, it is advisable to use a pattern of tin plate cut as shown in Figure 10. APPENDIX 309 APPENDIX. Note on the St as Pipette. The Stas pipette used by the writers in the assaying of silver delivers a nearly constant volume of liquid as shown in the following rtatement : — Weight of water at 70° F. delivered by the Stas Pipette. Time occupied in delivery = 13 seconds. Experiment 1 1000*140 grains TT 1000-146 III 1000-120 Mean 1000-13X3 „ In the arrangement of common forms of the pipette, a shallow bell-shaped glass is suspended over the top of it. Any excess of the liquid, which is introduced from below and is projected upwards, strikes against the glass and is caused to drop down into the glass dish adapted to the neck of the apparatus. There is in this arrange- ment a constant danger of the dish overflowing, and even of the liquid being splashed out on the table. The arrangement employed by us consists of an overflow tube instead of the bell-shaped glass, fr is said to have been first devised by the late Mr. W. E. Dubois, assayer in Philadelphia Mint, and has been adopted in that mint and also in the Royal Mint in London. In the form used in those institutions it is constructed of gutta- percha. But as made by us it is of glass, and with a little mani- pulation in the blow-pipe, may he easily constructed in the laboratory from a piece of bulb-tubing. 310 Y. KOGA & O. Y AM AG ATA. The apparatus, consists of a glass bulb a, Fig. 11, blown on one ond of a tube. The under side of the bulb is made into a cavity the top of which is cut open into a hole, thus forming a circular groove inside the bulb. The bead of the Stas pipette comes just below the cavity. Fig. 11. The liquid is fdled into the pipette from below as usual ; any excess of liquid is projected by its own force into the bulb above. APPENDIX. 311 and thence round the inner groove, quietly flowing away by the inclined tube to any suitable receiver placed at the end of the tube, without fear of spilling or splashing. The Imperial Mint, Osaka. April, 1890. On Cordierite as Contact Mineral by Yasushi Kikuchi. Witli Plate XXVIII. The occurrence of ( îordierite in some metamorphosed rocks along the margin of the granitic intrusion has often been recognized in many places in Europe. In this condition, however, the mineral seems to be only of secondary prominence, being usually found as formless grains associated with other minerals of contact origin, such as Andalusite, &c. The case which we have to describe is of especial interest, since so far as we know, ii is an instance in which this mineral has been solely developed as :i result of contaci metamorphism in a very peculiar modification, differing from the ordinary form of Cordierite by want of pleochroism, and the structural habit of its crystal which is hitherto unknown. The volcanic product of Asama-yama has yielded an interesting ( îordierite- rock, which was microscopically studied by Dr. E. Hussak.* This Cordierite, the characters of which I was also able to examine myself by the specimens kindly furnished to me by Dr. T. Harada, differs from that here to he described, as might have been expected from the different mode of its origin. * Ueber dea Cordierit in vulkanischen Auswürflingen — Sitzungsb. cl. k. k. Akatl. d. Wiss. in Wien. Mathem.-Naturwiss. Classe. 1883, Bd. LXXXVII. p. 332. -°)14 Y. KIKTJCHI. The following' article is an account of the results which have been obtained principally by the study of the specimens collected from the bordering; region of the Provinces Ivödsuke and Shimotsuke, along' the Watarase-gawa, where the mineral is developed in its most typical and fresh state. Previous to the discovery of this occurrence the attention of geologist was drawn to an interesting pseudomorphic mineral found imbedded within a dark slate developed at the granitic border, near the town of Kameoka in the Province of Tamba. It is always found in a prism of hexagonal form, apparently cleaving par- allel to the base, and consisting essentially of a greenish coloured mica. It is generali v known as Salcura-ishi (cherry-stone), as its hexagonal cross section presents a radial arrangement like that of a flower. With regard to the nature of this pseudomorph, nothing definite was known, for the original mineral which had given rise to it, was not discoverable at or near the locality mentioned. It was generally supposed to have been derived by alteration from some such mineral as Andalusite. Close examination showed, however, that there is a striking analogy of structure existing -between this pseudomorph and the fresh specimen developed at the contact zone of the Watarase-gawa region above referred to. Here also crystals occur in a dark slate. They were proved to have the essential characteristics of Cordierite, and on being altered gave rise to a pseudomorph analogous to the " cherrv- stone " of Tamba. Dr. Geerts* gives the description of Sakura-ishi under the variety of ' Chlorite hexagonale et lamellaire. ' It is mentioned as having a pale rose ( ?) colour, and consisting of elongated hexagonal prisms made up of tender and flexible lamina?. His specimen is said to have come from Prov. Mikawa. The occurrence of Cordierite in the Watarase-gawa region was * Les produits de la Nature, Japonaise et Chinoise, p. 433. ON CORD1EKITE AS CONTACT MINERAL. 315 first noticed by the writer in the summer of 1887. By occasional trips undertaken since then, the geological condition of the district has been made out. the summary of which may here be sketched as follows : — The Watarase- gawa, ;i tributary of Tone-gawa, takes its origin behind Lake Chiuzenji in Xikkô, near the ( !opper mine of Ashio in Shimo- tsuke, and flows south-westward for about 35 kilometres, along the strike-line of the Palaeozoic strata here develop- ed, to the town of Omama in Ködsuke. Here it leaves the mountainous region and makes a sudden turn to S. E., entering the plain of Kwantö. An elliptical mass of Granite makes its appearance along this river, beginning at the village of Karafuro near Ashio, and ending near the village of Gödo in Ködsuke, extending nearly 11 Worn, from X. to S, and 5 kilom. from E. to W. The southern extremity of this granitic mass gradually tapers and merges itself under the thick Palaeozoic layer, granitic exposures appearing here only in the river-bed; while the bordering ridges are composed of sedimentary rocks, which are Grunitt Cordicrib Sl.it, . Spatted Boundary qf th.- Statt ■ Provinces . 316 Y. KIKUCHI. often invaded by apophyses. As we proceed northward, the granite gradually gains over the higher level, attaining its maximum breadth just where the Watarase-gawa passes the boundary of the two Prov- inces already named. Further north of this point, it again loses itself under the sedimentär}' rock masses. The whole of these phenomena suggests that a huge granitic mass has been forced up from below, probably along a line of fissure running X. and S., partially represented by the valley of the Watarase-gawa, into the overlying stratified rocks. The accompanying «ketch map may serve to illustrate the brief description just given. The altered zones are marked at those places only where they have been actually observed. The Granite developed in this region is composed of Quartz, Orthoclase, and Biotite, with some accessory minerals ; Garnet, microscopic Zircons, &c. The structure is usually uniformly granular; sometimes porphyritie by the development of Orthoclase into large tabular crystals having zonal inclusion of mica flakes. The thick strata of the Palaeozoic date (probably the ; Kobotoke ' system of the Geological Survey) developed around this granitic mass consist of hard sandstone, — the ' Greywacke '■ — Hornstone, Slate, and some Limestone, all non-fossiliferous. Their direction of strike is N". E., dip generally toward S. E. Wherever these rocks come into contact with Granite they undergo peculiar alteration. The hornstone is converted into a harder Quartzite, the fresh fracture of which shows a fine granular saccharoidal appearance ; the sandstone becomes more compact in texture and much mica is developed in it. But that which concerns us here is the effect of metamorphism on the Slate. The first sign of alteration of this rock is manifested by the development in it of fine black specks, which examined under the microscope are seen to consist of aggregates of black mica. This ON CORDIERITE AS CONTACT MINERAL 317 kind of spotted slate is followed toward the Granite by a narrow zone of altered slate containing the characteristic crystals of Cordierite. The Cordierite- bearing slate is found close to the Granite and usually no particular indication of metamorphism can be detected other than by the presence of these crystals, or else by the compact texture, like that of "Ilornfels.' which it assumes. This inner zone is comparatively narrow in breadth, and exposures of some 10 mein* may be seen cropping out at the immediate vicinity of the Granite, while the outer zone is usually broader, being double or treble of the inner. Microscopically examined the ground mass of the Cordierite slair differs, in no essential respect, from the spotted slate already mentioned, consisting of minute quartz grains, ami a black mica, with irregular dots of opaque carbonaceous matter. [ am indebted to Mr. T. Xasa i'w the following chemical analyses of the slate, which were made in the laboratory of the Geological Survey. I. II. III. Si 0, 66-10% 67-50% 74*50 AL Oj 16-39 Fe <> 4-65 Fe, 0 -30 ' 'a 0 1-60 Mg 0 2-41 IC, 0 3-3(5 Na, <> 1-86 H,0 3-60 14-65 t; •46 trj nee !■ 10 1 ■90 3' 76 ■>. •79 2' nn 1 t-20 4-45 •30 •50 1-62 2-08 •!ll 1-33 100-^7 100-16 99-92 I. is the slate with the Cordierite crystals, which showed a slight sign of alteration, assuming a greenish tinge. The Ijarge olS Y. KIKUCHI. amount of water is no doubt due to this alteration. Loc. Söri in Ködsuke. IL, is the same as I., but with all the visible crystals of Cordierite removed. III. is the ordinary unaltered slate. Loc. near Grödo in Ködsuke. From these few analyses nothing quite certain can be asserted. It may be pointed out. however, that the relative amount of silica is less in the altered slate with Cordierite than in the unaltered one without any Cordierite, allowing that there was no very great difference in the original character of the two slates taken for the analysis. On comparing II. and III., we see that the altered slate approaches, on the whole, to the normal slate by the partial removal of the Cordierite crystals. It is probable that by the process of contact alteration, the slate was made basic by the addition of the Cordierite-silicate. The Cordierite-bearing slate presents a very characteristic ap- pearance in virtue of the elongated fusiform sections of the crystals irregularly scattered in the gênerai ground-mass. Further, on closer observation, rounded six-sided sections may often be detected. In weathered pebbles which have long been exposed to atmospheric action in river-beds, &c., the erystals are more readily recognizable than in the freshly fractured surface, on account of the fact that natural corrosion has been more active on them than on the surrounding mass, so that they are often found depressed at the exposed surface. They are also stained reddish-yellow by the iron hydroxide, which has probably partly resulted from the oxidation of the Iron-pyrites which they contain (Fig. 7). In those specimens, however, which have undergone more or less metasomatic alteration, a greenish tinge is always observable, resembling the pseudomorph from Tatnba. Sometimes stellate aggregates, which are probably due to tiwn -groups, are observed (Fig. 8). As the form of such aggregates OX CORDIERITE AS CONTACT MINERAL. ?>10 has n cherry-blossom like structure, the cordierite-slate was known by the name of Salura-islii, as in the case of the Tamba specimen. The size of the crystal is variable ; usually about 2 cm. in length and 5 cm. in breadth; but in some cases it attains the much larger dimensions of some 5 cm. in length and 1*5 cm. in breadth. In specimens in which the alteration has not much advanced, the fresh portion is colourless and transparent, and the fracture presents a highly vitreous lustre like that of quartz. The transverse section of the Cordierite crystal always presents an hexagonal figure divided up into sectors (Fig. 1), the boundaries of which are formed by the filling up of the black carbonaceous matter. There is usually in the interior a central nucleus or core, also hexa- gonal and parallel to the outer boundaries, but apparently devoid of any radial divisions or sectors as in the outer. Under polarized light the different sectors are found to differ in optical orientation ; the opposite [»airs, however, behaving alike. Thus we see that the crystal must be made up of the union of three individuals according to a twin law, forming a Trilling, which, as is well known, often assumes a pseudo-hexagonal symmetry, when the twinning plane is a prismatic face having an angle near to \20°. This twin, which was first observed in the specimen from Bretagne by J)f.s Cloizeaüx,* is very characteristic of Cordierite, and has also been des- cribed from the Asama and the Laach crystals. j- But there are in the case which we are now describing some points of peculiarity, which have not been recognized in the others. These are, firstly, the separation of the cross section into the outer and the inner portions, * Mantie] de Mineralogie. I. p. :C>.">. t HnssxK, (I.e.) and v. Las.vulx — Ueber Cordierifc-zwillinge in einetn Auswürflinge dos Laaeher Sees— Zeitschft. f. Krystall. 1884. VIII, p. 76. 320 Y. KIKUCHI. which* exhibit some difference in structure; and, secondly, the inclusion of foreign matter in the boundary between the outer and the inner portions, and between the twinning sutures in the outer por- tion. The inner portion or (lie core is different in character from the outer, as mav be easily revealed under the microscope by the use of oolarized li^ht. The twin-structure is soon evident under crossed Niçois, but the three systems of the twinned individuals often inter- penetrate in a very complex and irregular manner. Figure 3. is a mag- nified representation of the structure of the inner portion, in which the different patches having an analogous optical orientation are re- presented by the same shading. Tt will he observed that these differ- ent patches are reducible to three systems which are continuous with those of the outer portion. The directions of extinction, therefore, in the outer and the inner portions, are always parallel and at r'12'ht angles to the outer sides, and differ by nearly GO0 from each other. tinder a convergent polarized ray the different sectors of the more uniformly built outer port ion show the Axial-image (Interference-figure), the position of which in each sector is represented in Figure 1. The Optical-plane is situated at right angles to the outer side of the crystal, and therefore it is in the Macropinacoidal face (100). as in this twin the outer side corresponds to the Brachypinacoid (010). The Axial image examined with mica plate reacts negative, with a weak disper- sion, p-. * 1. c. p. 355. Tafel. II. fig. 28. 326 Y. KIKUCHI The composition of the mineral may probably be represented thus : — Oxygen-i'atio. Si02 4S-95% 26-10 5-2 Al,03 32-71...... 15-24 3 FeO(MnO) 0-98 2-21) 1 \ 5'55 ...1*1 MqO(CaO) 8-36 3*34 100-00 Thus we have ti MgO • Î FeO • 2. ALO, • 5 810,=$ MgO ' 4 FeO ' 10 AW, • 25 SiO,, which requires the following percentage composition : — s;n2 49-05% Ah03. 33-68 FeO 9-42 MgO 7-85 100-00 Hardnesss=7. Specific gravity=2-642, a mean of several ob- servations as determined by the Thoulet's solution. A thin splinter fuses at the edge to a transparent globule, and gives a blue colour on being" moistened with cobalt-solution and then ignited. Alteration — It is a significant fact that Cordierite is a mineral which is easily subject to alteration at the surface of the earth ; so that we know more of the altered products in various modifications from different localities than of the unaltered mineral. The heterogeneous outward appearances presented by these products have created many ill-defined names, such as Praseolite, Aspasiolite, C.hlorophyllite, Fahl unite, Gigantolite, Finite tvc. The final product of alteration is that usually known as Finite, a pseudomorph consisting of ON COEDIERITE AS CONTACT MINERAL. 62 1 mica. The peculiar pseudomorph from Tamba already mentioned be- longs no doubt to this category. In the specimens from the \Vatarase- gawa region, we can trace under the microscope the progress of the grad- ual transmutation from the transparent fresh mineral to its final stacre. The alteration first sets in from the irregular cracks traversing the crystal, somewhat like the formation of Serpentine from Olivine. In this first stage all that we can observe is, that there is formed along these cracks a greenish or white impellucid material in the form of a double layer (Fig. 10). This layer is found on close examination to consist of crystalline aggregates of a very fine fibrous mineral, exhibit- ing under crossed Niçois, an irregular aggregate-polarization. In most cases, however, we observe that the alteration proceeds along a series of very fine linear fissures, which are differently arranged in the different portions of the crystal, analogous to the arrangement of the enclosures already described. Figure 10 shows these fissures in a sec- tion parallel to the vertical axis, the irregular cracks, from which the alteration started, running in a transverse direction, while the linear fissures run at right angles to it. The alteration gradually makes its progress, until there is left no trace of the original substance, and the final product is an impellucid crystalline substance which usually bears a greenish tinge. It is usually mixed up with some greenish coloured mica, which is sometimes only very sparingly developed, or often fills up almost the whole space. But it is in the specimens from Tamba that this mica is most typically developed as the sole product of alteration. In its outer aspect the altered product of the Watarase-gawa region generally looks like Aspasiolite from Kragaroé, Finnland, while that of Tamba resembles Ghloraghyllite from Haddam, Connecticut,- U. 8. A. The pseudomorph from Tamba is of a greenish colour, imbedd- o'28 Y KIKUCHI ed in a «lightly reddish coloured slate looking like ' Hornfels ', and has a marked tendency to separate along the basal face. The hardness is somewhat low ( H = 2*5). Before the blowpipe, it fuses slightly at the edge. The crystal is built up exactly in the same manner as the unaltered crystals from the Watarase-gawa region, and therefore the forms of the transverse and longitudinal sections are exactly alike to those already described. The separation along the basal face is caused by the regular arrangement of mica-flakes parallel to the Base (001) of the crystal. This arrangement may be well observed in the longitud- inal section. Figure 11 is a magnified view of this section, in which the line lines are the cleavage traces of the mica-flakes. The black dots arranged in the form of rows or wedires, are the inclusions of carbonaceous matter. Examined under crossed Niçois, the layers of mica extinguish the light simultaneously, and therefore form optically a con- tinuous individual, but they are usually intermixed with irregul- arly arranged aggregates of the same mineral. The green mica is faintly but distinctly dichroic : when the ray of light vibrates parallel to the direction of the basal cleavage traces, it is of a light greenish colour, but is of a light yellowish tinge when the ray vibrates at right angles to this direction. A. Wichmann* distinguishes two phases in the alteration of Cordierite. For example in case of Cldoropliyllite, the first stage is characterized by the formation along the fissures of a certain crystal- line fibrous substance, while in the second stage the alteration occurs essentially in the mica flakes. The two modes of alteration which we have distinguished above evidently correspond to these two stages. Wichmann observes that the transition of the crystalline substance • Die Pseudomorphosen des Cordierite — Zeitschft. d. deut. geol. Gesellft. 1874, Bd. XXVI p. 079. ON CORDIERITE AS CONTACT MINERAL. 329 into mica is rather abrupt, there being no evident sign of gradual passage. We are rather inclined to think that the formation of mica may directly follow as a result of alteration. For, in some cases, we rind, that the tissures of the crystal are lined abruptly with the mica flakes, as if a thin clear cleavage-piece of this mineral were insert- ed between the fresh fissures. In such eases, which we could observe very distinctly in a specimen from the Watarase-gawa, the mica begins either to be formed very regularly disposed at right angles ti- the vertical axis, or from the sides of the crystal, sometimes parallel, sometimes very irregularly disposed, to the outer wall. It would be of interest to know the chemical alteration which the altered Cordierite has undergone. A sample of the Tamba pseudo- morph free from admixture, was analyzed, with care, by Air. Tamura of the laboratory of the Geological Survey. The result was as follows : — Si02 4092% AW, 31-06 FeO 7-99 CaO trace. MgO 6-71 KJ) 8-60 Na20 -72 Loss on ignition 3*22 99-22 G = 2-77. By comparing this result with that already given, we see there is a considerable addition of alkalies (potash) and of water, while the amount of silica is less. This pseudomorph, therefore, belongs to that class of the alteration-products of Cordierite, which has taken up not only water but also potash, and is essentially analogous to that 330 Y. KIKTJCHI. usually known as Finite. The addition of potash was explained hy Gr. Bischof* as due to the action of circulating water containing in solution, the alkaline silicates which had been derived from the decomposition of granitic rocks. With this addition of potash there is usually a diminution of magnesia. But in the present case this loss is slight compared with the addition of potash. The amount of water is also a little less than is obtained for other Pinitcs. From the foregoing descriptions it will be seen that Cordierite, as a product of contact metamorphism, presents many peculiarities which have been hitherto unobserved. These anomalous characteris- tics may be accounted for by the fact that the same mineral may assume different aspects according to peculiar conditions in which it was formed. This is well exemplified in case of Andalusite and Chiastolite. The latter is always found in the slate near the granitic contact, and is distinguished from the former by the well-known cross-inclusion of foreign matter, which seems to have been taken up during its growth, and by the absence or weakness of pleochroism. As this diagnostic difference has an important bearing upon the genesis of the crystal, it would be appropriate to emphasize the name of Chiastolite, so as to — , Wiedemann's Annale», Bd. 38. p. GS, 1889. OX TRANSIENT ELECTRIC CURRENTS. 337 The twisting' apparatus consisted of a graduated circle provided with a small twisting arm (a), which can he clamped in any position. (See Fig. 1). A stout brass axis passed through the centre of the circle, and the twisting arm was firmly fixed to it. Round the rim of the graduated circle, two grooves were cut. In one of these, two clamps (b) (6) were made to slide. These clamps were capable of being fixed in any desired position by means of screws, and served to stop the motion of the twisting arm at the desired angle of twist when the wire was suddenly twisted. The clamp and the grooves are shewn in Fig. 2, which represents the lateral view of the twisting apparatus. The magnetized wire could not of course be near the jnüvano- meter; and being under the necessity of working alone, I was com- pelled to use some device for effecting the twist at a distance. For this purpose, two strings were attached to the twisting arm, and made to slide in opposite directions. The strings after passing the groove were made to slide on two pulleys fixed on the table, on which the whole apparatus rested. The axes of these two pulleys were parallel and their planes perpendicular to the face of the circle. The distance between them was ecuial to the diameter of the circle, and they were fixed in such a position that the strings on leaving the groove and passing down to the pulleys were both vertical. The two clamps (/>) (b) being adjusted so as to stop the twisting arm at the desired posi- tions, the wire was twisted in either direction by simply pulling the proper string. In fact, the mechanism was similar to that of the rudder of a ship. The wire under examination had its one end clamped to the ex- tremity (c) of the axis of the graduated circle, being held in position by pressure sustained by a small nut (V), as sufficiently indicated in Fig. '2. Between the clamp and the graduated circle a projecting rim 338 H. XAGAOKA. of circular .shape was attached. This dippsd in a mercury cup placed underneath, and was in connection with one of the terminals of the galvanometer coil. In order to avoid any induction current that might arise from the sudden motion of the twisting arm and its appendages, that extremity of the axis which served to clamp the wire and lead the current to the galvanometer was insulated, from the rest of the axis and the graduated circle, by a piece of wood (e). The other end of the wire was clamped somewhat similarly on a piece of thick brass plate of rectangular shape, of which the front view is shown in Fig. 3. It was provided with two V shaped projec- Fig. 3. Fig. 4. ti p w b lions (b) (/>), which could slide smoothly on V grooves cut in a brass stand as shown in Fig. 4. The metal plate was prevented from turning over when the wire was twisted, by means of two hooks ft) (c), which slid in two grooves (<:') (c) cut on the lateral sides of the stand. This groove arrangement was necessary when the effect of longitudinal stress was to be studied. The longitudinal stress was applied by hanging a pan of weights to the end of a flexible string which was fastened to the clamping plate and passed over a pulley fixed to the table. The plate was connected with the other terminal of the galvanometer. The wire to be examined was always Avell straightened and carefully annealed. AVhen the wire was to be renewed, precaution was taken to cut it always from the same bundle for differences of material necessarily produces difference in the transient current. The wire ON TRANSIENT ELECTRIC CURRENTS. 339 was clamped while the pointer of the twisting arm was at zero of the graduated circle. Two clamps were previously fixed on both sides of the twisting arm such that when the wire was twisted in either direction, the arm should stop at the required angle of twist on either side of the initial position of no torsion. The wire was twisted be- tween two extreme limits of twist by pulling the strings attached to the twisting arm in the way before mentioned. When the Avire was first twisted, the deflection of the galvanometer magnet was generally large, but after repeated twistings and untwistings, the current settled to its ultimate value, and the reading was then noted. The coil of the low resistance ballistic galvanometer was wound in 20 layers of thick copper wire. It had a resistance of 0.28 ohm. In the core of the coil, a heavy magnet was suspended and the first swing was read by the deflection of a spot of light reflected from a mirror placed outside the coil, but rigidly connected with the magnet. The suspension was similar to that of the ballistic galvanometer made by Siemens and Halske. Since the logarithmic decrement of the vibrating1 magnet was very small, no correction arising from it was thought necessary, so that the amount of current in the arbitrary scale was taken to be equal to the reading of the first swing. As the moment of the galvanometer magnet is liable to changes especially when the momentary currents are passed through the galvanometer coil, the galvanometer was gauged from time to time by mean- of an earth inductor. The magnetizing coil was wound in six layers. The coil was placed on a table due magnetic east and west, and at a distance of nearly 5 metres from the ballistic galvanometer. It was 25 cm. long, and had a resistance of 1.3 ohm. The magnetizing field due to a current of one ampere was 35.5 C. G. S. units. The current was derived from 16 Daniell cells, and its strength was measured by means i-10 H. NAGAOKA. of a Thomson Graded Galvanometer. For obtaining strong mag- netizing forces, a large coil wound in 12 layers was used with a sufficient number of Bunsen cells. The arrangement of the coil and the twisting apparatus is shewn in Fig. 5. Fig. J. il M Transient Current in Iron and Steel. The transient current produced by twisting an iron wire placed in a magnetizing field has already been studied to a small extent. The results of this earlier investigation have been given in the paper referred to above. The method used in the experiments now to be discussed was different from the earlier method ; and in addition, wires of various thickness were examined. The present results are, however, essentially the same, although there are many features which passed unnoticed in the earlier and less detailed experiments. I shall first of all describe the experiment in the varying magnet- izing held, the wire being, in each successive field, twisted to and fro through a given constant angle. Most of the experiments were made on a soft iron wire, 1.24 mm. thick and 27 cm. long. The wire was carefully annealed, and deprived of residual magnetism by heating it red hot as it lay in a position perpendicular to the magnetic meridian. This precaution is always necessary, for if there remain but a small quantity of residual OX TRANSIENT ELECTRIC CURRENTS. 341 magnetism, a transient electric current will be produced, when it is twisted even in zero magnetic field. In fact, one can test whether the wire is completely deprived of magnetism or not. simply by twisting it with its length placed due magnetic cast and west, and O Ol K-> ' observing if there exist any transient current or not. The wire thus prepared was placed within the magnetizing solenoid, and suddenly twisted through an angle of 1.")°. It was then twisted back through zero to an equal angle of twist on the opposite side of zero. The total twisting was therefore through an angle of 30°, namely, + 15° — ( — 15e). This twisting I denote as on previous occasions, by ±15°; and generally the twisting ±0 means a twisting to and fro through an angular range 20 about the original position of no twist. Generally the transient current did not settle to a constant value until the wire was twisted several times between the two extreme limits of twist. For each of a series of gradually increasing values of the magnetizing force, the transient current produced by one sudd.cn twisting from limit to limit was measured on the ballistic galvanometer after the system had through repeated to and fro twistings reached a steady cyclic condition. The following are the values in successive fields : — Mao-ncti/àny force. Transient Current.* 0.7 12 1.7 2<; 3.2 34 4.5 36 5.8 35 7.5 :;i 9.8 32 14.6 27 22.0 21 20.0 17 36.7 15 * One scale division corresponds to 1.1 x 10-° Coulomb. 342 H. NAGAOKA. Taking for abscissae the strengths of the magnetizing force, and for ordinates the corresponding transient currents, we get the curve (Plate XXIX, Fig I, Curve I) representing the relation between the two. An examination of the curve will show the changes in the transient current as the strengths of the field is increased. The current at first increases nearly proportional to the increase of the magnetizing force. On passing the field of about 3 units, the increase takes place very slowly and ultimately reaches a maximum for iQ= 4.5 nearly. The current then gradually decreases, though at a slower rate than that at which it first increased. This rate of decrease becomes still slower as the field is further increased. The curve representing the relation between the transient current and the strength of the magnetizing force has thus two inflexional points, the one just before reaching the maximum, and the other after passing the maximum. From its similarity to the analogous feature in the curve of magnetization, I shall hereafter call the point at which the ratio <>f the transient current to the magnetizing force is greatest the 'Wendepunkt' of the transient current curve. On increasing the angle of twisting to ±30°, and experimenting on a new piece of wire taken from the same bundle and treated in the same way as before, Curve II. Fig. I. was obtained. The examina- tion of this curve shows the same characteristics as before. But the transient current has greatly increased for all strengths of the mag- netizing field, while, at the same time, tin; maximum current occurs in a higher magnetizing field than in the former case. This takes place for ft — 6 nearly. By increasing the twisting to ±60°, Curve III. was obtained. The changes wrought bv this second doubling of the twisting were not nearly so great as in the first doubling from 15° to 30°. The general feature remaining the same as before, we notice a peculiar ON TRANSIENT ELECTRIC CURRENTS. 343 difference between the curves II. und III. At first the transient current increases for r=±60° as the field is increased, hut until the magnetizing force is about 5 C. G. S. units, the current is always smaller than that for the twisting of ±30°. When 0=5, the tran- sient current attains the same value for these two different twists, and henceforth the current for the larger twisting acquires an ascen- dency over that for the smaller. The rate of decrease after passing the maximum is slower for the larger twist than for the smaller. The maximum point occurs in a still higher field, namely, &=8 nearly. With the twist increased to ±90°, the transient current curve (IV.) becomes still further transformed. What was noticed on com- paring II. with III. applies equally well when we compare III. with IV. The curve in weak fields lies below that for r = ±60°, but when § exceeds 13, the former lies above the latter. In this case, however, the maximum current is less for the larger twisting. From the comparison of these four curves, we may reasonably infer what would be the course of the curve when the angle of twisting is further increased. Evidently the transient current in weak magnetizing fields as well as the amount of maximum current will be smaller for high than for moderate angles of twist. After passing the maximum, the slope of the curve will become less steep, as the angle of twist is increased, so that the curve will necessarily cut all those having a greater maximum current, and ultimately lie above the corresponding branch of the curves for smaller twistings. At the same time, the strength of the magnetizing field correspond- ing to the maximum transient current will be increased as the twist is taken greater. The effect of longitudinal stress on the transient current was also tried. It simply produced decrease of the current as I have already mentioned in my former paper. 344 H. SAGAOKA. Professor Ewing has remarked that the production of a transient current by twisting an iron wire in a magnetizing held is a natural consequence of Sir William Thomson's discovery that aeolotropic stress gives rise to aeolotropic magnetic susceptibility in iron. The stress on twisting the wire is equivalent to compression and stretch- ing along lines perpendicular to the radius, and inclined at 45° to the normal plane section. The susceptibility along the lines of com- pression is different from that in the direction of stretching. The consequence is that the lines of induction originally parallel to the axis of the wire are changed into helices which are inclined toward the direction of the relatively increased magnetic susceptibility. These considerations lead to the conclusion that in nickel, the direc- tion of the current is determined by that of compression while in iron it is determined by that of stretching. Thus the direction of the transient current in iron and nickel should be in opposite directions, and in fact this was found to be the case. It ma}', however, be doubted whether the effect of twisting is even approximately equivalent to compression in one direction and extension in the other. It is not at all improbable that twist gives rise to changes in the molecular configuration of the wire. As we are absolutely ignorant of the ultimate structure and arrangement of the molecules, nothing definite can be stated. But if there is any change produced among the polarized molecules of the magnetized wire, it is quite probable that the displacement of the molecules will give rise to changes in the transient current, which will be quite inexplicable in terms of aeolotropic susceptibility as generally as- sumed. Although such aeolotropy ma)' be one of the chief causes of the transient current, yet other causes ma)7 exist whose effects can not be neglected. In fact, I have reason to believe that aeolotropy does not sufficiently explain all the phenomena observed in TE W CENT ELECTRIC CUREES r . 345 iron ana nickel, as will be described afterwards. Considering the results of the preceding experiments, one naturally asks, why does the transient current flow in the direction above specified and why does it reach a maximum? Supposing that aeolotropy is the only cause of the transient current, the increase of the current would mean a greater magnetic induction in the direction of stretching than in the direction of compression. The susceptibility of the stretched iron is greater and that of the com- pressed iron is less than that of iron in an unstrained state, up to a certain critical value of the field. Thus the helices representing the lines of induction would be inclined toward the direction of stretch- ing, and the current must be constantly on the increase until at a certain field the difference of magnetic induction in these two direc- tions begins to diminish. Reasoning in this way, we may explain the general feature of the curve in the following manner. The amount of compression as well as that of stretching remaining constant throughout the course of experiment, incren the magnetizing field produces a greater increase of the circular component of the lines of induction in the direction of stretching, than in the direction of compression. The difference in the circular components of magnetization at last reaches a maximum fora certain strength of the magnetizing force. This critical strength of the magnetizing force depends upon the amount of compression and of stretching. In the wire tried above, the position of the maximum is gradually shifted into higher field as the compression as well as î retching is increased, but at the same time, the current does not increase proportionally. There is a certain maximum twisting for which the difference of circular lines of induction in the direction of compression and of stretching reaches the maximum. This seems to occur for the twisting of about ±60° in the wire discussed, 346 TT. XAGAOTCA. that is, a twist per cm. length of ±2.°2. When once the maximum difference in the circular lines of induction is reached, the suscep- tibility in the direction of compression and of stretching gradually tend to equality, so that in strong magnetizing field, the current he- comes very small. Subjecting the wire to higher degrees of com- pression and of stretching makes the difference of susceptibility in strong fields greater, although the maximum difference of suscep- tibility becomes less after a certain amount of stress is exceeded. Thus it will be seen that the existence of a maximum transient current bears a close relation to the Villari critical point and an analogous point in compressed iron. Another set of experiments was tried by the following method. Instead of twisting the wire always through the same angle while the magnetizing field was varied the latter was kept constant, while the wire was twisted through angles of twist of gradually in- creasing amount. For each successive value of twist, the transient Curve T. Curve II. Curve III. Curve IV. r .s> 2.4 § = 5.3 0 = 20.7 .§ = 58.2 8 •> — — K° 10 12 5 la 40 31 2 1 — 20° 51 46 1 I. :]()D 55 56 ! I. 19 40° 54 58 53 24 50° 52 59 55 27 GO0 51 58 57 20 Tu" 51 57 58 30 80° 50 57 58 3] 00° 50 50 • > i 31 OX TRANSIENT ELECTRIC CURRENTS. o47 current was obtained and measured exactly ns in the previous set of experiments. The wires used in these experiments were in :ill cases taken from the same bundle, and had a diameter of 1.24 mm. The accompanying" table gives the currents in the arbitrary scale unit. These readings are plotted in Fig. II. (Plate XXIX.), in which the abscissa gives the angle of twisting and the ordinate the corres- ponding transient current. In general characteristics all these curves strongly resemble each other. The transient current increases at first very rapidly. Ultimately the current attains its maximum value, and thereafter it begins to decrease very slowly. The general feature of the curves obtained by varying the amount of twisting does not differ much from those obtained by varying the magnetizing force. Each of the curves obtained in both sets of experiments has a maximum point ; namely, that corresponding to a particular twist in a constant mag- netizing field, or that corresponding to a particular field when the amount of twisting is kept the same. In the latter case, the mag- netizing field that gives the maximum current is greater for greater twistings, while in the former the twist giving the maximum current is greater in the stronger field. Although it is difficult to find the o o o particular angle of twisting for which the current attains its maximum value, ir can lie approximately estimated from the curves plotted in Fig. II. Thus the angles of torsion giving maximum transient cur- rents in different strengths of the magnetizing field come out as follows : — 348 II. NAGAOK "A. v> r 2.3 ±30° 5 . : > + 50° 20.7 ±7(3° ■".s. 2 + 90° (about) The increase of the angle of torsion giving maximum transient current with the strength : of the magnetizing force is thus apparent. We see a strong analogy between the magnetizing force giving maximum current when the twisting is kept constant, and the angle of twisting producing maximum current when the magnetizing force is kept constant ; the increase in the one produces increase in the other. When the wire is subjected to longitudinal stress, the transient current is greatly diminished, while the maximum point occurs for larger twists than in the case of the unstretched wire. In § = 2.4, the maximum current occurs for the twist of ±30° when the wire is unleaded : when it is loaded with 4 kg. weight, it occurs at «about ±36°. and when the load is increased to 8 kg., it takes place at about ±42° as will be seen from the curves V. and VI. For SW ">..';, the angle of torsion giving' the maximum transient current is ±.*>f) when the wire is unstrained, but becomes greater than ±90° when the wire is loaded with 8 kg. If we suppose the development of the transient current to be entirely due to the aeolotropic magnetic susceptibility caused by twisting the wire, the transient current in the present set of experi- ments must depend upon the difference of susceptibilities along the OX TRANSIENT ELECTRIC CURRENTS. 3 L9 lines of compression, and the lines of stretching. Taking this as the basis of our reasoning, let us examine what these curves giving the relation between the current and the amount of twisting indicate. The difference of magnetic susceptibilities along lines perpendicular to the radius and inclined at 45° to the normal plane section of the wire increases at first as t lie wire is subjected to greater twistings. This différence, however, attains a maximum value. Hereafter the suscep- tibilities along these two lines gradually tend to equality though very slowly as the twisting is taken greater. Now let us analyze the susceptibilities along the directions of stretching and of compression separately. We know that the suscep- tibility increases with loading, but when the wire is stretched beyond a certain limit, the susceptibility does not increase any more, and begins to decrease. Although the stretching and compression caused by twisting the wire varies at different distances from the axis of the wire, yet on the whole the increase of twisting necessarily gives rise to increased susceptibility in the direction of stretching until the critical value of stress is reached. It is well known that this critical point occurs with smaller amounts of pulling stress as the magnetiz- ing force is increased. The consequence is that by twisting the wire in strong magnetizing fields, the wire acquires the maximum susceptibility in the direction of stretching for smaller amounts of twisting. Accordingly if stretching was the only stress acting in producing the circular lines of induction which produces the transient current, we should expect to obtain the maximum current at smaller twistings as the magnetizing force is taken greater. This is contrary to the observed fact, which is that the twisting corresponding to the maximum current is increased with the strength of the magnetizing force. Thus the effect of compression must play an important part in changing the amount of the transient current. 350 H. NAGAOKA. Unfortunately wc have not many experimental data on this subject, so full and detailed as those we have on the Villari critical point. According to the experiments of Ewing and Low*, the effect of compression in altering the susceptibility of iron has a certain similarity to the effect of stretching ; and there is a similar reversal corresponding to that of Aillari. Under constant compressional stress, the curve of magnetization for compressed iron lies below that for the unstrained, in weak magnetizing; fields. In strong inagnetiz- ing fields, the former acquires greater susceptibility, and the curve lies above the latter. I do not know of any experiment establishing' the relation between susceptibility and the compressional stress, when either the magnetizing force or the amount of compression is made to vary. Without these data, it is difficult to examine the effect on the transient current either by varying the magnetizing force or by varying the amount of twisting. The fact that compression reduces the susceptibility of iron in weak fields shows why the transient current increases with increase of twist. So long as the magnetizing force does not exceed a certain limit, the susceptibility diminishes in the direction of compression, while it increases in the direction of stretching provided the Villari critical point is not exceeded. Thus the circular component of the lines of magnetic induction is constantly increasing as the twisting is increased and will therefore give rise to an increasing transient current. The increase of current will take place so long as the difference of susceptibilities in these two directions do not diminish. Observation shows that the maximum transient current occurs for larger twist- ings as the magnetizing force is taken greater. Since the stretching stress giving the maximum susceptibility becomes smaller as the magnetization of the wire is increased, it seems that compression * Phil. Mag. 18S8. OX TRANSIENT ELECTRIC CURRENTS. 351 produces greater diminution of susceptibility as the magnetizing field is taken greater. But consistently with the existence of a point similar to the Villari reversal, and the diminution of magnetization in weak fields, we should find increase of susceptibility in the direction of compression when the magnetizing field is taken sufficiently great. Of course, the increase of susceptibility in a given field will take place only within a certain limit of compressional stress, beyond which there is decrease. If this is really the case, the increase of susceptibility in the direction of stretching must always be greater than the increase in the direction of compression however greal the magnetizing field is taken, for otherwise there would be reversal of the direction of current. Xow examining the results of Professor Ewing on the mag- netization of soft iron wire, I find that the increase of susceptibility by loading is very small, when the field strength is over 10 unit-. It is not thus improbable that when the field is taken sufficiently great, the increase of susceptibility in the direction of compression exceeds that in the direction of stretching, so that the current is reversed. But this is quite contrary to the results of the present ex- periments. On this account, it seems that twisting the wire does not give rise to simple compression and extension along lines inclined at 45° to the normal plane section of the wire and perpendicular to the radius. The molecular arrangement of the wire must also be affected and give rise to additional changes in the transient current, which cannot be ascribed to aeolotropy alone. In the experiments so far described, the wire was always twisted in the magnetizing field between two extreme limits of twist. In the experiments now to be described, the wire after being twisted repeatedly to and fro, was kept in position at either extremity of the range of twist, and the external magnetizing field was then suddenly 352 H. XAGAOKA. reversed in direction. This was done by means of a rocker com- mutator placed in circuit with the magnetizing solenoid. With the wire kept steadily twisted either on the right or left hand side of the initial position of no torsion after repeated to and fro twistings, a magnetizing force was applied. On reversing the direction of the magnetizing force a momentary current passed through the galvano- meter coil. In general the deflection of the galvanometer magnet set- tled to a constant value after numerous reversals, upon which the swing of the galvanometer magnet corresponding to a reversal was noted. The wire was then twisted in the opposite direction, and after numer- ous reversals of the magnetizing force, the deflection of the galvano- meter magnet was again noted. The mean of these Uvo readings was taken as a measure of the transient current. It should he noted that the direction of the transient current is reversed when either the twrist is reversed or the direction of current reversal is changed. The first set of experiments made in this way was by varying the magnetizing force, the amount of twist remaining constant. The iron wire used was 1.24 mm. thick and 27 cm. long. Instead of giving the observed numbers, I refer to curves plotted in Fig. III. (Plate XXIX.) These show at a glance the relation between the current and the magnetizing force for different amounts of twist. These curves were obtained for the following different angles of torsion. Curve I for the twist of ± 5°, 5? ^ ,, ,, ,, >>i^o, » HI ,, „ ,, „ ± 30°, 5j IV ,, ., „ „ ± 60°, V +120° The general feature of all these curves are similar to those obtained by twisting the wire through a given angle in different mag- netizing fields. The transient current increases on the application of OX TRANSIENT ELECTRIC CURRENTS. 35 3 an increasing magnetizing force, at first rapidly ; but ultimately the current attains a maximum value. Thereafter ir U-irins to decrease very slowly ;is the magnetizing force is increased. The transient current in weak field is smaller for large than foi1 moderate twists, hut after reaching the maximum, the rate of decrease becomes slower as the twist is taken greater. The consequence is that the curves for larger twists cut the curves for the smaller in a particular magnetizing held. This is well exemplified in the three curves [If, IV, and V. The magnetizing field corresponding to the maximum transient current becomes greater as the amount of twist is increased. The following numbers give the strengths of the magnetizing force cor- responding to the maximum transient current. A ogle of torsi« >n. § -i~ ■ > G ± 15° 9 ± 30 12 ± 00° 16 ±12i)° 17 The magnetizing field corresponding to the maximum transient current is however, nearlv twice as great when the field is reversed with constant twist than when the twist is suddenly reversed with constant held. The accompanving figure gives the relation between the magnetizing force corresponding to the maximum transient 354 H. NAGAOKA. 10 J 1 1 JJJ-J — c s** -— -'v 1 M. / | + U4- ^ fc-"" '1 M •K i Fia, ß. current and the angle of torsion. F. is for the me- thod of sudden twisting", while II. is for the method of sudden reversal. Tn the preliminary in- vestigation on this subject, r I have already noticed the 0 *'0 III 50 80 100 120 n ,i ,i ,• • fact that the magnetizing field giving the maximum transient current by sudden twisting is less than that obtained by Professor Ewing by the method of reversal. I then thought the discrepancy to be probabty due to difference of procedure. On experimenting in these two ways with the same wire. T now find that the difference in the position of the maximum is certainly due to the différence in the method. But it seems inexplicable why this should be so, unless the ultimate causes of the transient current are known. Corresponding to the experiments made by twisting the wire through different ranges in constant magnetizing fields, another set of experiments was tried by reversing the direction of the magnetizing force for different amounts of steady twist. Curves showing the relation between the transient current and the amount of twist are given in Fig. TV. These curves were obtained for the following different strengths of the field. Curve I for £>= 1.6 II „ ©= 5.2 III „ §= 8.3 IV „ § = 13.0 V „ \v 51.8 VI „ §= 5.1 (loaded 4 kg.) ON TRANSIENT ELECTRIC CURRENTS. ,100 Curve VII £=9.6 (loaded 4 kg.) , VIII § = 5.5 ( „ 8 „ ). These curves, in their chief characteristics, all resemble those obtained by the application of sudden twisting. As the angle of twist; is increased, the transient current due to reversal increases at first very rapidly, but ultimately it reaches a maximum whence it begins to decrease slowly. For strong values of the reversing Held, the current is small for small twists, but as the twist increases, it ultimately cuts the curves for smaller magnetizing force. Thus the rate of decrease of the current for greater twists becomes smaller as the field is increased. So far everything is similar to the results obtained with sudden twisting. In the curves given in Fig. TV., we again notice the shiftino- of maximum point to places of greater twist as the magnetizing force is taken Greater, as will be seen from the following table. ±.\> T wist corresponding t< i maximum current. 5.2 27° 8.3 31° 13.0 to° 51.8 60s The twist corresponding to the maximum transient current, whether that is due to reversal of held or to sudden twisting, has a similar relation to the magnetizing force. As the field is increased, the angle of torsion corresponding to the maximum current becomes correspondingly great; but. for the same strength of the magnetizing field, the twist corresponding to the maximum current when the field is reversed is smaller than the twisting of the same name, which for steady field «fives the maximum current. Of course the anode of 356 TT. NAGAOKA. torsion in these two cases is not strictly comparable, the strain caused by twist or by twisting being somewhat different. The difference between the two will bè seem from the accompany- ing; figure, in which the Fig. iooc soc noc -in 20c 1 ^ \i y < / -yC / ^x^ TT / X ^ | I Jt s f- 10 40 ÔO abscissa gives the mag- netizing force and the ordinate represents the twist or twisting corres- ponding to the maximum transient current. Curve I. is for the method of sodden twisting while II. is for the method of re- versal. If we compare Figs. 6 and 7, we notice that in the latter I. is above II. while in the former II. is above I. This at first sight seems paradoxical, but we see that one is a natural consequence of the other. We saw that for the same angle of torsion, a greater magnetizing force must be applied to make the current attain its maximum strength by the method of reversal than by sudden twisting. Thus it would follow that to make the current reach its maximum value in a given magnetizing field by the former method the amount of twist must be less than the corresponding angle given by the method of sudden twisting. On this account II. lies below I. in Fig. 7. The strain at any point by twisting the wire depends on the distance of the point from the axis of the wire. All parts of the normal section which are at equal distances from the axis of the. wire will he equally strained. Evidently the outer part of the wire must suffer the greater strain. With the same angle of torsion, the strain must be greater in the thick wire than in the thin. Had the ON TRANSIENT ELECTRIC CURRENT«. 357 wire been in the form of a thin tube, it would be easy to see how the change in the transient current takes place when the diameter is made to vary. Inside the cylindrical wire, the strain is different in dif- ferent places, and it is not at all easy to see how the transient current will be altered with the thickness of the wire. To find how this alteration takes place, wires of different thick- ness were examined by the method of reversal. These wires were 0.88 mm.. 1.24 mm., 1.54 mm., and 2.00 mm. thick respectively. In each case the wire was twisted through 60° in either direction, and the transient current produced in these different specimens by reversing the direction of the magnetizing force was measured. The following table gives the readings.* r=0.44 (Fig. V, Curve I.) r=0.62 (Fi g. Y, Curve IL) s> Transient Current. S> Transient Current. 3.4 . J 1.4 •) 6.6 18 3.4 17 11.0 27 6.3 13 10.4- 29 9.4 •:» i 21.0 28 13.5 61 24.3 27 16.4 OU 28.9 26 20.9 (ill 23 29.0 56 37.1 50 In subsequent expérimenta, one scale division corresponds to 2.1 x 10 u Coulomb. 158 H. NAGAOKA. »■=0.77 (Fi, y. Y, Curve III.) »•=1.00 (Fig. V, Curve IV.) •V> Transient Current. 0o. The twist thus applied amounted to 13° per cm., and must correspond to great com- pressional and stretching stresses. On reversing the magnetizing current and observing the transient current, it was found that the direction was always the same and determined by the direction of stretching ; nor did the current due to this large twist show any notable decrease as compared with currents due to smaller twists as the comparison of the following readings with the same wire for smaller twists will show. OX TRANSIENT ELECTRIC CURRENTS. .ill.) s Trans. Cur. 2.0 o O 5.7 39 8.9 ■'»2 1 3.0 57 17.3 57.5 10.7 57.5 23.0 57 27.2 56 20.7 54 To .study the effect of longitudinal tension upcn the current, loads of 4 and 8 kg. were successively applied, and the reading of the galvanometer magnet measured by reversing the magnetizing force. The effect of loading was simply to reduce the cunvni by a small amount, and the representative curves were not greatly altered. The above result leads me to suspect that possibly Matteueci made a mistake in determining the direction of the current in relation to the twist and field, always in such experiments somewhat of a confusing matter. The main difference between the experiments of Matteueci and those of later investigators lay in the thickness of the bar or wire. Matteueci used mostly an iron bar 6.5 mm. thick, while in the present investigation, the thickest wire experimented on had a diameter of only 2 mm. Accordingly, I determined to repeat the experiment of Matteueci with a soft iron bar 5.6 mm. thick and 2S cm. long. This was tested both by sudden twisting and by reversing the direction of the magnetizing force. The arrangement for twisting the bar was different from the one described above. It was twisted by means of a lever, into the details of which E need nor here enter. 364 II. XAOAOTCA. The sudden twisting of the bar produced a large amount of transient current, but the direction of the current was always the same as with wires of smaller diameter. The only difference noticed by twisting the wire was the quantity of the current developed. Noth- ing differed from the results hitherto described. Professor Ewing lias suggested that the twist applied by Matteucci was already beyond the critical value, since the bar was very thick, and that consequently the current most have changed its direction. In the above experiment the bar was twisted through 90° (3° per cm.). This greatly exceeds the twist applied by Matteucci, his limit being 20° in a bar GO cm. long, or only 1° per cm. We cannot therefore explain the supposed discrepancy in the direction of the transient current as the result of a large twist. T<> shew the similarity of the effect obtained with the bar to that obtained with the thin wire. I give the following readings taken in varions magnetizing fields by the method of reversal. dr.Sj) Trans. ' !ur. for r 30 ±r£ Trans. Cur. for r = G0° 3.0 20 2.9 20 G.] 72 5.6 06 9.8 138 8.6 116 12. -j 177 11.8 160 16.4 209 15.3 194 21.0 233 1-8.9 219 24.3 238 ■22.7 236 30.9 240 27.1 239.5 38.1 23 L 32.6 239.8 45.4 216 38.8 2.;:» 69. 1. 169 61.4 195 78.9 148 74.8 174 ON TRANSIENT ELECTRIC CURRENTS. 365 The observations recorded by Matteucci are very few in number. He gives only four galvanometer readings for each experiment. In spite of the scantiness of his data, we can see in a general way whal form his transient current curves will have. In all cases, the tran- sient current increases at first with the number of cells used, but always decreases when 10 cells are used. This shows that the tran- sient carrent curve will have a maximum point, and must be similar in form to these obtained in the experiments described above. From this we may also estimate roughly the strengths of the magnetizing field used by Matteucci. These facts indicate that the direction of the current as stated by Matteucci, if not an accidental mistake, must be due to some other cause than that suggested by Professor Ewing. The transient current developed in twisted steel wires was examined by the method of reversal, as was done for iron. The wire was twisted through ±60°. and the strength of the magnetizing force gradually increased. The direction of the magnetizing current was reversed, and the reading of the galvanometer magnet was noted. Two specimens of steel wires were examined. The one had a diameter of 1.28 mm., and the other 1.50 mm. Plotting the read- ings on the galvanometer scale, Fig. VII. was obtained. The current increases as the magnetizing force is increased. Ultimately it reaches a maximum, and then gradually begins to diminish. The transient current flows in the same directum as in iron. The only differences, which will be apparent at a glance, are the comparative small ness of the current and the higher magnetizing force corresponding to the max- imum transient current. For wires of nearly the same thickness, the transient current in steel does not amount to more than a fourth part of that in iron. On the other hand, the maximum current in steel occurs in a stronger magnetizing field than in iron, as the comparison of the curves represented in Figs. V. and VII. will show. see H. NAGAOKA. With varying twists in a constant field of '20 C. G. S. units, Curves I., IL, III. in Fig. VIII. were obtained with three specimens of steel wires of different gauge. Their diameters were 1.26 mm., 1.50 mm., and 1.82 mm. respectively. The readings are as follows: — 7- = 0.63 (Curve I.) r=0.75 (Curve II.) r = 0.91 (Curve III.) T Trans. Cur. Trans. Cur. Trans. Cur. - o 0 1 — — 10° 3 5 I» 20° 7 11 1 1 30° 10 1 1 19.5 40° 15 L7.8 21.3 50° 16.5 18.3 20 60° 16.8 17.0 L8.5 70° 15.8 16 18.0 80° 15.0 15 On0 1 L3 In these three curves, we notice the same characteristics as have been already noticed in similar experiments with iron. There is a certain angle of twist for which the transient current is a maximum. This angle of twist varies with the thickness of the wire, and is smaller for the thick than for the thin wire. Thus the currents in iron and steel are in every respect similar, except in the amount of the transient current developed and the position of the maximum current. When soft iron or nickel wire is twisted it immediately acquires a large permanent set. and would in no case return to the initial position of no torsion. In steel, the limit of elasticity is very great compared with soft iron or nickel. Whenever the steel wire is twisted and afterwards released, the wire almost always returns to its former position although the amount of twist is considerable. The ON TRANSIENT ELECTRIC CURRENTS. 367 smallness of the current in steel indicates that the stress produced by twist has a smaller effect in steel than in iron. The principal point of difference in these two substances is at the same time attended with ;i difference in the permanent set acquired by the twisted wires. These facts suggest that there exist certain intimate relations between the transient current and the limit of elasticity. Transient Current in Nickel. As I have already remarked, the present investigation originated in searching if the transient current does not show any peculiarity in the magnetized nickel wire, which has undergone the reversal of polarity by the combined action of torsion and longitudinal stress. With this object in view, the transient current produced by twisting the nickel wire was examined under different longitudinal stresses, the strength of the magnetizing field being made to vary. No peculiarity occurring simultaneously with the reversal of polarity was observed. A more minute study was then instituted, in which the nickel wires were subjected to the same treatment as the iron and steel wires. The arrangement for twisting the wire, and the method of measuring the transient current, were exactly the same as in the experiment with iron, so that the further description of the process will be unnecessary here. Two specimens of nickel wires were specially examined. The one was 0.5 mm., and the other 0.43 mm. in radius. The wire was carefully annealed, and deprived of residual magnetism by heating- it red hot. This precaution was especially necessary with the nickel wire, for if there remained but a small quantity of the residual mag- netism, quite a large amount of transient current was produced. As in the case of iron, the current did not attain its ultimate value 368 H. NAGAOKA. until the wire was .subjected to many back and fore twistings. The following table gives the reading of the swing of the ballistic galvano- meter magnet, obtained by twisting the wire 1 mm. thick and 27 cm. long, through different angles, while the magnetizing force was made to vary. r -±30° r = ±45° r = +60° r -±90° £ Trans. Cur. £> Trans. Cur. ,s> Trans. Cur. § Trans. Cur. 8.9 67.0 2.3 69.6 1.2 63.3 0.6 60.0 0.7 70.8 4.4 74.6 3.8 75.8 2.0 73,6 10.1 71.7 8.5 75.9 7.1 78.6 4.5 7 the next reversal was taken. This was done with a series of gradually increasing values of the magnetizing force. Out of numerous experiments, \ give only five of the curves (Fig. XI) taken with the thick wire and four (Fig. X 1 1) obtained with the thin. For I he thick wire : — Curve I. for r = ± 30° If. „ r = ± 90° III. .. r = ±18l>° IV. „ r = ± 00° (loaded I kg.) ,, V. ,, r = ± 60° ( ., 8 kg.) For the thin wire : — ON TRANSIENT ELECTRIC CURRENTS. 373 ( lurve I. for r OCT II. r = ±360c „ III. „ r = ± 60° (loaded 4 kg.) „ IV. „ r = ± 60° ( ., 8 kg.) The principal features of all these carves are nearly the same in all cases, and resemble those for iron. The first application of the magnetizing force makes the current increase very rapidly. This increase, however, takes place only within a small range of the mag- netizing force. The curve soon reaches the Wendepunkt, and there- after goes on rising; very slowlv. In the first two curves in Fig. XT. the magnetizing force was not sufficient to make the current reach its maximum value. In the third experiment for r = ±180°, stronger magnetizing forces were applied, and then the existence of the max- imum was demonstrated. The following are the readings. Ô Trans. Cur. 3.7 3.8 5.4 1 5.3 7.6 38.0 9.5 48.4 12.6 56.1 1(3.5 50.9 20.7 64.3 34.1 67.4 50.6 67.4 78.0 00.9 As was remarked before for iron wires, the maximum transient current produced by reversing the direction of the magnetizing force occurs for higher vaines of the magnetizing force than is the case when the wire is suddenly twisted in a steady field. The comparison 374 II. NAGAOKA. of curve III. Fig. XL with any one given in Fig. IX. will .show that this must also be the case with nickel. Another point of difference between these two methods is the difference in the position of the Wendepunkt, and the initial form of the curve before reaching the Wendepunkt. In the experiments made by Midden twisting, the Wendepunkt occurs in weaker magnetizing fields as the twist is increased; but it is otherwise when the mag- netizing force is reversed. In this ease, the Wendepunkt gradually shifts into higher regions of the magnetizing field. The first increase of the twisting transient current, as the magnetizing force is increased, takes place more rapidly for greater twist ings ; but when the direc- tion of the magnetizing force is reversed, the same increase takes place more rapidly for the smaller twist. This fact is evident from the examination of the curves L, IL, III. in Fig. IX. and I., II. in Fig. XI. The curve for r=±90° lies at first below the curve for r = ±30°, while the curve for r =±180° lies, at the outset, below the curves for these other angles of twist. When the wire is subjected to longitudinal stress, the curve undergoes a slight alteration. The initial part of the curve lies below the curve for the unloaded wire, but the current increases steadily, and finally reaches the Wendepunkt, which occurs at a hjo-her field than in the case of wires under no longitudinal stress. As the longitudinal stress is increased, the current becomes smaller in weak magnetizing fields, but it increases steadily witli the increase of the magnetizing force so that the curve ultimately cuts the curve.' for no longitudinal stress. At the same time the Wendepunkt of the curve is shifted into stronger fields. Similar results have been obtained for the twisting transient current. In the experiment made by suddenly twisting the nickel wire, it was seen that with the angle of twist then applied, the current was ON TRANSIENT ELECTRIC CURRENTS. 375 always increasing as the twist was increased. This is in harmony with the view of aeolotropic magnetic susceptibility. In the experi- ments now to be described, the same series of experiments were per- formed by subjecting the twisted wire to reversals of the magnetizing forces. The results thus obtained are somewhat striking, and seem incapable of explanation in terms of aeolotropic magnetic susceptibility. The experiment was tried in exactly the same way as for iron. Out of a number of experiments made in this way, I give the follow- ing, as showing the typical relation between the amount of twist and the transient current. The following are the readings made with the thick wire; and the corresponding curves are shown in Fig. XIII. Curve I. Curve IT. Curve III. r £ = 10.6, w = 0. £=4.5, w= 5 kg. £=9.8,w=5kg. 10° — b 16 20° 21 13 50 30° 44 27 68 40° 63 38 74 50° 65 44 77 00° 64 43 75 70° 62 37 72 80° 59 23 71 90° 57 2.3 69 ] 00° 52 1 63 120u 44 0 56 140° 38 0 (150=)47 180° . 18 41 270° 7 360° 3 450° 2 540° 1.5 The following give the observations made with the thin wire. 376 H. NAG A OKA. Curve IV. Curve V. Curve VI. r £=9.8,W=0. $ = 5.0, W= 5. £=25.1, W=5. 10° 25 2 30 20° 44 — 59 30° 52 19 (if, 40' 56 2U.5 154 50° 50 6.5 63 00° 57 4.0 62 70° 56 — ci 80° 55 2.8 61 90° 54 — 60 120° 52 0.5 — 150° 50 58 180° 47 56 360° 30 50 On the first application of twist, the transient current increases with the angle of twist, but it soon reaches the Wendepunkt and very quickly thereafter a maximum. The curve then dips, and the decrease of current with the increase of twist takes place nearly proportional to each other, so that the curve appears to be nearly straight. But there is another point of inflexion (not shown in the figure, but evident from the readings). Thence the curve dips very slowdy, and ultimately becomes asymptotic to the line of no transient current. The current is very small when the twist is large, but from the course of the curve it is evident that further twistings will not reverse the direction of the current. The curve undergoes great deformation when the wire is sub- jected to longitudinal stress, especially when the magnetizing force is weak. The initial rise of the current takes place in the same way as represented by curves obtained with unstrained wires. The curve ox TRANSIENT ELECTRIC CURRENTS. 377 after passing the maximum dips very rapidly describing a curve which is more curved than that for the unloaded wire. The current decreases very rapidly, so that it is soon reduced to a very small amount. Thus after the angle of torsion exceeds 90° or thereabouts, the deflection of the galvanometer magnet produced by reversing the direction of the magnetizing force is scarcely appreciable. On greatly increasing the twist, the transient current remains in the same state, so that it is quite probable that the current ultimately becomes vanishingly small when the twist is very great, but the direction of the current will never become reversed. Curves II. and Y. show these features very distinctly. It must not be supposed that a similar curious change takes place in the transient current whenever the wire is loaded. It is only in weak magnetizing fields that the changes above described take place. On examining the transient current in .$ = 10.6 with the thick wire subjected to a longitudinal stress of 5 kg. weight, Curve III. was obtained. The course of the curve is slightly different from that obtained with the unloaded wire in § = 9.8 (see Curve I.). The maximum point is present for nearly the same angle of twist, both for the loaded and for the unloaded wire. The curve dips after passing the maximum, but the course is nearly straight and not curved as in II. and V*., so that the transient current will be appre- ciable unless the twist becomes very great. According to the researches of Professor Ewing, longitudinal stress always decreases the magnetic susceptibility of nickel, while compression always increases it. If we resolve the torsional stress, as has already been done into extension and compression at right angles to the radius of the wire, and inclined at 45° to the plane section normal to the axis, the circular component of the lines of induction will be constantly on the increase as the twist is taken greater. If, 378 ÏT. XAGAOKA. then, we are to explain tlie transient carrent in terms of known changes of susceptibility in the directions of elongation and compres- sion, we should expect the circular components of the lines of induc- tion in the wire, and probably the transient current, to increase as the twist is increased. But in the experiments just described, there is a maximum which conies with a tolerably small amount of twist, and there is always decrease of current when the angle of twist becomes sufficiently great. These results on the transient current, and the change of susceptibility in the directions of compression and of stretching lead to contradictory conclusions. Thus it seems that the peculiarities of the transient current cannot he explained in terms of aeolotropic susceptibility, but some other causes must be acting in producing the peculiarity mentioned above. In order to rind, how the transient current is affected by the thickness of the wire, four specimens of nickel wire were examined by the method of reversal. The angle of twist was in all cases equal to ±60°. The following table gives the readings;* the curves are shown in Fiff. XIV. r=0.5 (Curve I.) r 0.65 (< lurve II.) Trans. Cnr. 7.2 76 7.5 50 6.7 71 13.1 106 12.2 65 9.0 su 17.7 108 15.5 67 11.4 83 22.«; 11 1: 22." 71 14.7 87 31.8 1 i<; 2'. i.2 71.8 18.1 88 39. 1- 121 37.6 72.3 20.4 88 50.3 118 L9.8 71 23.4 89 83.0 11 ! 7<;.7 69. 2 7. s 89 125.1 108 34.0 90 to. 6 88 These curves for wires of differenl thickness, have all the same characteristics as already noticed. The current, after reaching the Wendepunkt, increases so slowly that it is difficult to tell where the curve reaches n maximum. Nevertheless, as will be seen from the readings, the transient current for each of these- wires does reach a max- imum, which seems to occur ar lower fields in The Thinner wire. The curves for nickel are similar To Those obtained wirb iron wires, the only points of difference between the two being The indistinctness of the maximum point in nickel and its occurence in higher fields. On stretching The wire longitudinally, the same effect as already described in similar experiment by The method of sudden twisting was obtained. In weak magrietizing fields, The current produced in The loaded wire is less Than in The unloaded. But near The Wendepunkt, the curve for The loaded wire passes above the unloaded, und beyond this point the current for The former becomes always greater than 380 TT. NAGAOKA. that for the latter. Moreover the maximum for the loaded wire takes place in stronger field than for the unloaded. Corresponding- to the above, another set of experiments was performed by the method of reversal, the strength of the field being kept erpial to 19.7 throughout the course of experiment and the twists being varied. The following are the readings; the curves being shown in Fig. X V. ?'=0.5(CurveI.) j-=0 65(Cim-eII.) r = 0.84(CnrveIII.) r = 1.00(CurvoIV.) r Trans. Cur. Trans. Cur. Trans. Cur. Trans. Cur. 5° 9 18 34 34 10° 10 34 51 62 20° 20 17 65 82 :3(YJ 80 52 71 85.0 40° • >■) 54 71 84.8 50° .) . > • >•> 54 69 83.8 60° 3 1- 53 67.3 83 70 34 52.5 07 82 80" 34 52.3 00.8 81 90° 34 — 100° 33 51.8 120° 00 51. The examination of these curves shows the increase of the tran- sient current when the diameter of the wire becomes greater. As 1 have already noticed before, the transient current always reaches its maximum value for a certain angle of twist. The angle giving max- imum current varies with the thickness of the wire. The thicker the wire, the smaller the amount of twist which is to lie applied to make the current arrive at a maximum, as will be seen from the following: OX TRANSIENT ELECTRIC CURRENTS. ;>S1 for r= 0.50 r-70°, „ r=0.65 r-45°, „ r=0.84 r-3G°, „ r-1.00 r=30°. The curves also resemble those obtained in similar experiments with iron. When the nickel wire is loaded and then subjected to varying twists, the curves of transient current present a peculiar aspect, especially when the field is weak. To test this for thick wires, experiments were performed with wires 1.68 mm. and 2.00 mm. thick respectively. The curves (see Fig. XVI.) thus obtained show the same characteristics as before noticed. After pass- ing the maximum point, the decrease of the transient current takes place very rapidly. The current, a little while after, becomes insisrniticantly small, so that for large twists, it is insensible to the galvanometer. Thus the peculiarity seems not to be due to the thick- ness of the wire. In niv former paper. I had occasion to remark that the transient current curve resembles the curve of the Wiedemann effect. In the latter, the twist due to superposed circular and longitudinal mag- netization of the wire increases at first with increase of the magnetiz- ing force, but ultimately reaches a maximum. According to the researches of Professor Knott, the twist in nickel takes place in the opposite sense to that in iron, and the maximum twist in the former occurs in much higher magnetizing fields than in the latter. These facts are in every way similar to those observed in the transient cur- rent. Following the close analogy between the Wiedemann effect and the transient current, experiments were made in order to find if the similarity can be pushed even to very strong magnetizing fields. Ac- cording to Mr. Bidwell, the Wiedemann effect in iron is reversed when 382 H. NAGAOKA. the magnetizing force is sufficiently great. If the similarity exists, there should be a reversal in the direction of the transient current. With the strongest magnetizing current at my disposal (§ = 500) no reversal of the transient current was noticed. The analogy thus seems to fail in very strong1 magnetizing" fields. The following gives the summary of the results obtained in the present investigation. 1. — The transient current produced by suddenly twisting the iron wire, or by reversing the direction of the magnetizing force while the wire is held twisted, through a constant angle, increases as the field is increased, but after reaching a Wendepunkt, it soon arrives at a maximum point, whence it begins to decrease slowly as the field is further increased. 2. — The rise of the transient current in gradually increasing fields is smaller for large than for moderate angles of twist, but in strong magnetizing fields, the current becomes greater as the twist is taken larger. o. — The maximum point varies with the amount of twist, and occurs in higher fields as the twist is taken larger. 4. — The transient current in stretched iron wire is less than that in the unstretched. 5. — The transient current produced in a constant magnetizing field by varving the amount of twist increases at fir.st with the increase of twist. The current, however, arrives at a maximum point, whence it begins to decrease slowly as the twist is increased. 6. — The initial rise of the transient current as the twist is increased is greater in weak than in strong fields. 7. — The maximum point varies with the strength of the magnetizing force, and occurs for larger twists as the field is stronger. The twist corresponding to the maximum current is smaller by the method ON TRANSIENT ELECTRIC CURRENTS. 383 of reversal than by sudden twisting. 8. — When the thickness of the wire is altered, the initial rise of the current with the increase of the magnetizing force is greater as the wire becomes thinner, but the maximum current and the magnetizing force giving the maximum current both increase with I lie thickness of the wire. 9. — When the twist is varied in a constant magnetizing1 held, the maximum point occurs for smaller twists as the thickness of the wire is increased. 10. — Everything with regard to the transient current in steel is similar to that in iron, with the exception, that the current is small- er, and-the magnetizing force corresponding to the maximum current is stronger, than those observed in iron wires of the same thickness. 11. — The transient current produced by twisting nickel wire, or by reversing the direction of the magnetizing force is opposite in direc- tion to that in iron. The wire being twisted into ;i right-handed screw, the current in nickel flows from the south to the north pole. 12. — The transient current produced by twisting, or by reversing the direction of the magnetizing force while the wire is held twisted, soon arrives at a Wendepunkt, beyond which the increase takes place very slowly, and ultimately attains a maximum. The decrease with further increase of the mao-netizino- force is very slow. The Wendepunkt, for the same angle of torsion, conies sooner by sudden twisting than by the method of reversal. 13. — The current in weak fields is smaller when the wire is loaded, but in strong fields the curve of transient current passes above that for the unloaded. 14. — With the magnetizing force constant, and the amount of twist variable, the transient current produced by sudden twisting always increases as the twisting is increased. The increase, however, takes 384 H. XAGAOKA. place very slowly beyond a certain angle of twist. 15. — Everything being the same as in (14), the transient current produced by reversing the direction of the magnetizing force in- creases at first with the increase of twist, but it soon reaches a maximum. After this a slow decrease of the current takes place, and when the twist is very large, the current becomes very small. 16. — Loading the wire and proceeding in the same way as in (15), the current decreases rapidly after passing the maximum, and soon after becomes very small. 17. — When the thickness of the wire is different, the transient current is in general greater as the wire becomes thicker. The current in weak fields is smaller with the thick than with the thin. The maximum current occurs in higher magnetizing field as the thick- ness is increased. 18. — With the magnetizing force constant, and the amount of twist variable, the maximum current occurs for smaller twists as the wire becomes thicker. PLATE XXIX. Plate XXIX. Fig. Fig- Fig. J.— Obtained by twisting iron wire (r = 0.62) in varying mag- netizing fields. Curve I. for r — ±15° ; Curve II. for r = ±30° ; „ III. „ r = ±60'; „ IV. „ r = ±90\ 77. — Obtained by twisting iron wire (r= 0.62), r being variable. Curve I. for£ = 2.3; Curve II. for 0= 5.3; „ III. „ 0 = 20.7; „ IV; „ 0 = 58.2; „ V. „ ö= 2.4,.w=4kg.; „ VI. „ 0 = 2.4,w=8kg.; „ VII. „ 6 = 5.3, w =8 kg. III. — Obtained by reversing the direction of the magnetizing force with iron wire (r = 0.62), fè being variable. Curve I. for r = ± 5° ; Curve II. for r = ± 1 5° ; III. V. r = ± 30°; r = + 120°. IV. r= ±(30°; Fig. IV — Obtained by reversal with iron wire (r = 0.62), r being variable. Fig. Curve I. for£=± 1.6 „ III. „ 6 = ± 8.2 „ V. „ $ = ±51.8 „ VII. „ S=± 9.(3 V. — Obtained by reversal for being variable. Curve I. for r=0.44; „ III. „ r=0.77; Curve- II. for£ = ± 5.2; „ IV. „ »=±13.0; „ VI. „ ö = ± 5.1, to ==4 kg. „ VIII. „ S = ± 5.5, w =8 kg. = ±60° with different wires. .s> Curve II. for r = 0.(32 ; „ IV. „ r = 1.00; Fig. Fig. Fig. Curve II. tor r=0.62; „ IV. for r= 1.00. u V. „ r = 0.91, (Swedish iron). VI. — Obtained by reversal in M =±19.7 with different wires, r being variable. Curve I. for r = 0.44 ; „ III. „ r=0.77; VII. — Obtained by reversal with steel wires, .£ being variable. Curve I. for r= 0.63, r = ±60°'; Curve II. for r=0.91, r = ±60°. VIII. — Obtained by reversal with steel wires, r being variable. Curve I. for r=0.63,£ = 19.7 ; Curve II. for r=0.75,£ = 19.7 ; ., III. ,. r=0.91. 0 = 19.7. Jour. Sc. Coll. Vol III. PI. XXIX 1 ~- *---* / &1 /iy.v!7 m yp s,» "*-"»" ^ 'J20 TT \ ""« *-*% J '\ 100 .^- p— -i^r — 100 £ '"-- ./I , m ' ,•■•-"■'' — - -va HO % ^M * V So y ..' 1 f / — — i /!/- 60 '<-+- -' -^ 60 / 4 "• VI 40 \\ / IE il 40 7 j 20 / t î E T 0 à 10 15 20 ~~ itf 30 35 '/ ?'/ •#C 60- SO' lui. FY-a /^ lu; IV ' J20 /r/T : l\ r .+■"' --*— — *■ #_ 1/1/ 1 1/1/ ! *■* 1 100 ^ r —f.. --* 1 h^~~^~ ' A ! /// 1 ^ m Èr"\"" ^< ^. So h #' ' s^—1 \lf V llh / "ï--. h---_ -it— 60 j // m 1 cil / 1 J 40 -* / (| 20 1 ~*~ j a '! r 0 it 10 15 20 , itf «?e 35 0 ar •W \G0' SO- KO 100 Fujj V N Fi y. VI / *-"' '"" / ■^ - 80 / / // -\n ni i * -x m // \ 00 j // n / je // t ; t= -> î 4" z / >i . 4 " I / 1 20 y*y k \ r- 0 S 10 15 ,Sfl '•5 .50 ^ 35 0 -o ./ .30 40 .50 ! 6V> 0 ?0* [ */■ 6Y/ Mr too- PLATE XXX. Plate XXX. Fig. IX.— Obtained by twisting nickel wires (?•= 0.50) in varying magnet izing fields. Curve J. for r ±30°, r 0.5 -, Curve II. for r = ±45°, r=0.5; „ III. ,, r = ±60°, r=0.5 ; „ IV. ior r = ±90°, r=0.5 ; V. „ r = ±60°, r=0.43; „ VI. for r = ±90°, r=0J48. j. •/,_,. X — Obtained by twisting nickel wire, r being variable. Curve I. for 6 = -5.0, r=0.5 ; Curve II. for 0 = 9.8, r=0.5, „ III. „ 6=10.5, r=0. 5, u> = 5; „ IV. „ 6 = 9.8, r=0.5,w=8. V. „ 6=32.8, r=0.43; „ VI. „ 6 = 9.9,r=0.43,«>=4. Fig. XI. — Obtained by reversing the direction of the magnetizing force, with nickel wire (r = 0.5), m T 50 7 ' / /;/ / ^-- ::o if /?: i' / / : r 10 30\ 1 * / Of / À K^ 40 /i / 1 w° ,W° 60J 70- SO' 90 / 30 //i ^ 0 ~°\\ ;i0 '/ hi 30 1 1 t" kfj Jt ih 20 0 lo- 20 1 w\ 1/ X /; % \ \x 1 20 Il j 1 / 20 12