JOURNAL OF GENETICS CAMBRIDGE UNIVERSITY PRESS iLonUDlt: FETTEE LANE, E.G. C. F. CLAY, Manager also H. K. Lewis, Gower Street and William Wesley and Son, 28, Essex Street, W.C. CEUinbursI) : 100, PRINCES STREET Berlin: A. ASHER AND CO. ILeipjiB: P. A. BROCKHAUS fitfoj Horft: G. P. PUTNAM'S SONS ffioOTbae anH Calnitta: MACMILLAN AND CO., Ltd. All rights reserved JOURNAL OF GENETICS EDITED BY W. BATESON, M.A., F.R.S. DIRECTOR OF THE JOHN INNES HORTICULTURAL INSTITUTION AND R. C. PUNNETT, MA., F.R.S. ARTHUR BALFOUR PROFESSOR OF GENETICS IX THE UNIVERSITY OF CAMBRIDGE Volume II. 1912— 1913 Cambridge : at the University Press 1913 PRINTED BY JOHN CLAY, M.A. AT THE UNIVERSITY PRESS m A-31 AIM >b CONTENTS. No. 1 (February, 1912) PAOB Arthur W. Hill. The History of Primula Obcoiiica, Hance, under Cultivation, with some remarks on the History of Primula sinensis, Sab. (With Plates I and II) ..... 1 H. Drinkwater. Account of a Family showing Minor-Bi-achydactyly. (With 15 Text-Figures) 21 A. H. Sturtevant, A Critical Examination of Recent Studies on Colour Inheritance in Horses . . . . . . . 41 R. H. CoMPTON. A Further Contribution to the Study of Right- and Left-Handedness. (With 4 Diagrams) .... 53 No. 2 (June, 1912) W. Neilson Jones. Species Hybrids of Digitalis. (With Plates III — V, one coloured, and 45 Text- Figures) . . . . 71 L. DoNCASTER. Notes on Inheritance of Colour and other Characters in Pigeons .......... 89 C. J. Bond. On Heterochromia Iridis in Man and Animals from the Genetic point of view. (With Plates VI — IX, and 7 Text- Figures and 1 Chart) 99 Richard Staples-Browne. Second Report on the Inheritance of Colour in Pigeons, together with an Account of some Experiments on the Crossing of certain Races of Doves, with special reference to Sex-limited Inheritance. (With Plate X, coloured) . , 131 Frederick Keeble. Gigantism in Primula sinensis. (With Plate XI, and 5 Text-Figures) 163 vi Contents No. 3 (November, 1912) PAGE L. DoNCASTER. The Chromosomes in the Oogenesis and Spermato- genesis of Pieris brassicae, and in the Oogenesis of Abraxas grossulariata. (With 15 Text-Figures) . . . . .189 Clifford Dobell. Some recent work on Mutation in Micro- organisms. Part I. (With 3 Text-Figures) .... 201 R. C. PuNNBTT. Inheritance of Coat-colour in Rabbits. (With Plates XII— XIV, two coloured) 221 A. H. Trow. On the Inheritance of Certain Characters in the Common Groundsel — Senecio vulgaris, Linn. — and its Segre- gates. (With Plates XV— XVIII, and 4 Text-Figures) . . 239 Frederick Keeble and E. Frankland Armstrong. The Role of Oxydases in the Formation of the Anthocyan Pigments of Plants. (With Plate XIX, coloured, and 5 Text-Figures) . . .277 No. 4 (February, 1913) A. H. Trow. Forms of Reduplication : — Primary and Secondary. (With 6 Text-Figures) 313 Clifford Dobell. Some recent work on Mutation in Micro- organisms. Part II. ....••• • 325 K. ToYAMA. Maternal Inheritance and Mendelism. (First Con- tribution.) (With Plate XX) 351 Volume II FEBRUARY, 1912 No. I THE HISTORY OF PRIMULA OBCONICA, HANCE, UNDER CULTIVATION, WITH SOME REMARKS ON THE HISTORY OF PRIMULA SINENSIS, SAB. By ARTHUR W. HILL, Assistant Director, Royal Botanic Gardens, Kew. Primula obconica, Hance', was introduced to England from China by Maries, one of Messrs Veitch's collectors who in 1879 sent home seeds from the Ichang gorges, where the great river Yangtse rushes out of the mountains. The plants raised from these seeds flowered in September, 1880 2. In the Botanical Magazine, t. 6582 (Sept. 1881), the plant is figured and described under the name P. pocidiformis, Hook, f, and in this figure the petals are shown with a simple notch or indentation similar to that of the common Primrose, P. acaulis. The wild specimens collected in China and preserved in the Kew Herbarium all show this character of the simply toothed perianth segments though they exhibit a considerable range of foliar variability. In the description the plant is said to have the habit and foliage of P. cortusoides (see t. 399 and t. 5528) and the calyx of the Himalayan species P.jUipes — a native of rocks at Chuka in Bhotan at an elevation of 6500 feet. P. obconica as introduced appears to have been a well-defined plant showing on the whole but little variation and, except for slight diver- gences in the colour of the flowers and character of the leaf margin, to have remained fairly true to type for about the first ten years after its introduction. The finding of the wild plant in China by Maries is best described in his own words : — " When I was travelling in Central China, I was » Joum. Bot. 1880, p. 234. » Horttu Veitehii, pp. 82 and 292. Joam. of Ckn. u 1 2 History of Primula much puzzled how to bring out living plants 1100 miles to the coast at Shanghai. I, of course, took plants of the things I thought were best for garden purposes, but Ferns and herbaceous plants were altogether out of the question. I thought, however, that many seeds would germinate if they were kept in soil, so I collected surface soil from Ferns and Primulas, and other plants. This was kept in an old wine box and eventually taken to Hong-Kong. I took this home twelve months afterwards and the soil was ' sown ' in a glass house. The first thing that came up was Primula obconica in large quantities, several shrubs, and a lot of ferns \" Sir Joseph Hooker, writing in the Botanical Magazine^, states that the plant is probably very variable and that the earliest flowering specimens sent by Mr Veitch were less hairy and had rounder and nearly entire leaves and very much smaller flowers than that figured in the plate. The flowers are of a pale lilac colour with a yellow eye, and the perianth segments, which are rather narrow, show a deep apical notch. Messrs Veitch =* speak of the colour in the virgin species as an "undecided lilac," while according to another writer* the flowers are said to be pinkish-white and it is suggested that white forms might be raised by careful selection. From a notice in The Garden^, we learn that "the flowers though somewhat small are of a pleasing mauve tint. The (almost entire) leaves are large and broad and they form a distinct tuft which lies almost flat on the soil." In the Botanical Magazine on the other hand the leaves are represented as upstanding with lobulate- dentate edges ^ With these preliminary remarks as to the incidence of variation, the detailed history of our modern forms may now be examined. In attempting to trace the history of Primula obconica under cultivation it will be convenient to arrange the facts under different headings such as colour of the flowers, size, shape, fimbriation of the corolla lobes and doubling and lastly questions connected with hybrid- isation. 1 The Garden, November 22, 1890, p. 479. 2 Bot. Mag. 1881, t. 6582. 3 Hort. Veitchii, p. 282. •* Gardeners' Chronicle, October 28, 1882, p. 565. 5 The Garden, June 25, 1881, p. 655. " See also The Garden for 1884, September 6, p. 236, the perianth segments are drawn with an apical notch. A. W. Hill Flower Colour. For the first few years after its introduction, as has been shown, the flower colour is always recorded as pale or undecided lilac (Plate I, figs. 1, 2) or pinkish-white but references to the variable character of the plant are frequent. In 1886 there is a record of plants from the Royal Botanic Garden, Edinburgh, with blooms " ranging from mauve to lilac and frequently pure whiter" "N. G." in the following year refers to the variability in flower colour and speaks of a few nearly white forms'. In the same note, it is mentioned that " we have observed in two or three of the plants this year a much deeper shade of rose round the eye than previously." This appears to be the earliest record of the appearance of the dark eye which was only found in a few cases among a batch of ordinary forms. The next record of the dark eye is in 1889 when "R. D." (Richard Dean) writes "The variety I have is of a very delicate mauve colour, with a slight purple ring round the eye'." In 1893 mention is again made of a distinct circle of a dark colour surrounding a lemon eye and the lemon eye itself is recorded as a novelty*. A "dark carmine shaded eye" is also mentioned in a note in 1895' and a "distinct eye" in the following year*. The "dark eye" is now a very common feature in many of the present day forms and the depth of the colour of the eye tends to in- crease as the flowers remain open (cf. Plate I, figs. 8 and 17, Plate II, figs. 27, 30, 32, 34). White variety. The white variety to which references have already been given does not receive further mention in horticultural journals until the year 1896. In that year Messrs Vilmorin Andrieux et Cie" of Paris exhibited "P. ohconica a grande fleur blanche" at the meeting of the French Horticultural Society, on Feb. 27th. This form had fimbriated petals. A pure white form was shown again, together \vith coloured varieties before the same Society by Messrs Vilmorin on May 2nd 1899', and was catalogued by Messrs Vilmorin in that year under the name "P. o. grande fleur blanche pur." » T. W. Sanders in Joum. Hort. May 6, 1886, pp. 358, 359, with fig. 66. * "N. G." in Joum. Hort. May 26, 1887. » "B. D." in Gard. Chron. November 2, 1889, p. 504. * "A. D." in The Garden, October 7, 1893, p. 327. * J. C. Tallack in The Garden, AprU 6, 1895, p. 240. « The Garden, December 12, 1896, p. 481. ' Rev. Hort. 1896, p. 238. » Bev. Hort. 1899, p. 169, "A cdte descoloris rose et blanc pur...". 1—2 4 History of Pi'imula I am informed by Mr A. W. Sutton that the pure white variety first appeared with his firm in 1899. The next reference to the white- flowered plant is in 1904, when a plant figured in the Gardeners Chronicle^ is described as having " white flowers or as nearly white as possible." Mr Gumbleton also in this year sent "almost white" flowers to the editor of The Garden'^ raised from seed obtained from Messrs Haage and Schmidt. Further a group of P. obconica alba was shown by Messrs Veitch at the Temple Show that same year, and it was remarked that ..."never before had been seen so near an approach to pure white''." P. obconica has been largely grown by Her Grace Adeline, Duchess of Bedford, and has been the subject of numerous experiments by the head gardener, Mr Dickson. I am indebted to Her Grace for affording me every facility for obtaining information about the culti- vation of the plant at Chenies. The first white P. obconica arose there in 1903, and its origin would appear to be independent from the white forms previously mentioned since it is alleged that no admixture from outside sources has taken place. In present day collections the white- flowering varieties can usually be easily distinguished by the darker tint of green in their leaves, and by their more delicate and paler stems. Messrs Veitch have a fine strain of this plant, very similar to those raised by Mr Dickson and others. The variety includes both toothed and fimbriated flowers and is found to come true from seed (Plate I, figs. 15, 16, 19, 20). Rose variety. A very distinct break in colour and one of the earliest to arise was started with the development of the rose colour from the original pale lilac and this yielded the remarkable rose-carmine series which includes some of the most striking of our modern varieties. Mr Sutton informs me that the first break from the original lilac was a good rose-pink, seed of which was offered by their house as P. obconica rosea* in 1895, and a variety under this name was exhibited at a meeting of the Royal Horticultural Society early in 1896 and was said to be an " undoubtedly most decided break in point of colour." Otlier references are to be found in The Garden about this time* (cf. Plate I, figs. 3—5). In 1897 Mr T. S. Ware« also showed a variety 1 Gard. Chron. 1904, p. 244, Fig. 103, p. 245. 2 The Garden, 1904, January 9, p. 18, see also I. c. April 7, p. 304. 3 ..T." in The Garden, July 2, 1904, p. 3. * The Garden, February 20, 1897, p. 143 "E. J." 5 The Garden, November 7, 1896, p. 383, see also August 7, 1897, p. 110. 6 The Garden, March 13, 1897, p. 197, March 20, 1897, p. 216. A. W. Hill 6 with flowers of a " warm rose tint " and the managing director of Ware's Nurseries has sent me the following account of the origin and development of their rose-coloured forms : "The first noticeable tendency (with us) of Primula obconica to produce other than the pale lilac flowers occurred about twenty years ago, when amongst a batch of many hundreds of seedlings (from seed obtained from a Continental source), five or six plants showed a deeper coloration beyond anything we had ever previously noticed. These plants were isolated, cross fertilized, and the seed saved separately from each plant. The resulting seedlings produced a fair amount of rose colours of varying shades. The best of these as regards depth of colouring, size of bloom, and good habit were retained, the remainder being destroyed. These were again cross fertilized and the selection carried on as before. Year after year seedlings were raised by this means (the colour becoming more intense each generation) until at last we reached the climax, that is, a deep self-rose of good habit and large flowers. In each successive batch of seedlings we always found one or two plants with extra deep and a greater number of serrations and of good colour. These we made the seed-bearing parent, as after repeated trials we found the serration was more pronounced in the oflfspring than when these plants were used as pollen bearers only, and so together with the development of the desired rose colour, the fimbriation or excess of serrations was proceeding at the same time in each successive generation " (cf. Plate I, fig. 3). Messrs Vilmorin of Paris had also at this time produced a rose form of P. obconica^, but it is not possible to determine whether the rose break originated in one place only or in several nurseries at about the same time, though the latter suggestion seems the more probable^ A form apparently very similar to these early rose or pink varieties is still to be seen planted out in the beds of the Temperate House at Kew where it has been grown undisturbed for many years (Plate I, figs. 3—5). A rose variety' formed the starting point of the experiments carried out in the gardens of the Duchess of Bedford. ^ Rev. Hort. 1897, p. 141, annoanced in their catalogue for 1898 as " P. obconica k grande fleur rose vif." 2 See also /. c. 1899, p. 548, with coloured plate. The variety "Bose Chamoise" was catalogued by Messrs Vilmorin in 1900. * P. obconica rosea, seed purchased from D. W. Thomson, Edinburgh, who obtained seeds from Stewart and Co., Covent Garden. The origin of the seed was very possibly continental. 6 History of Primula A further improvement in the production of a red ohconica " Vesuve," was due to Messrs Rivoire, pere et filsS and the variety "Vesuve" pro- duced in 1903 was said to surpass in colour all other reds and carmines, Messrs Rivoire, like Messrs Vilmorin, attribute all their improvements simply to selection. In the next year Messrs Barr and Sons exhibited their "Crimson KingV described as a "rich deep lilac-crimson, a decided advance in coloured obconicas," followed in 1905 by their "Crimson Queen'," a "deep crimson-rose." In the note on this plant it is remarked that the first pink and rose forms were thought to have been sent from a large Fifeshire garden ten years before. Rover* had also an assortment of reds from "Zartesten blassen Rosa bis zum diinkelsten Karmin." Messrs Sutton and Sons chronicle the production of a crimson form in 1906, and on Nov. 23, 1909 they exhibited a remarkable variety ^ " Sutton's Fire King," having terra-cotta crimson flowers with a yellow throat surrounded by a darker ring. The origin of this form is also attributed entirely to selection. Herr Georg Arends of Ronsdorf, who has kindly supplied information about his experiments with P. ohconica, obtained varieties which he has named "rosea" and "kermesiana" (Plate II, fig. 35). In a letter dated Nov. 24, 1909, he writes " The last quite new colour I gained was the var. 'Feuer Konigin' (Fire queen), that came out of the Kermesiana in four or five years' work. There is a kind of salmon-orange in the crimson of this variety and I think it will be possible to have a pure salmon-pink shade from it in a few years." So much interchange has gone on in recent years between nurserymen that it is highly probable that many of the varieties recorded by different houses have a common origin and are really the same plant. A dark claret form to which the name "Chenies excelsior"" has been given has the darkest coloured flower of the red series so far seen (Plate II, figs. 37, 38). Violet-blue varieties. One of the most recent and striking colour shades which have been evolved in P. ohconica under cultivation is a decided violet-blue shade which is now a well-marked and good colour. The improvement in this direction in England appears to be due very largely to the efforts of Mr Dickson. 1 am informed that a small 1 Rev. Hon. 1903, p. 442, see also Rev. Hort. 1906, p. 487. a Garden, 1904, March 24, p. 261. =* Garden, 1905, February 9, p. 116. * Gartenflora, 1905, 54, p. 82 ; see also idem 1903, p. 204. * Garden, 1910, April 9, p. 179, with plate. 6 Gard. Chron. 1911, April 29, p. 268. A. W. Hill 7 flowered blue variety was raised a Chenies in the autumn of 1904, and this was followed by the production of a large-flowered blue form in 1906 (Plate I, figs. 10 — 12). A variety named coerulea, remarkable for the blue colour of the flowers, was shown by M. Ferard* at the French Horticultural Society, October 1907, and the production of a similar variety in France is of interest as it seems almost certain that the blue colour has been developed quite independently in two different places. Herr Arends has also produced a blue form and he says in his letter " out of the white I raised the blue, beginning with plants which only showed a slight bluish hue in the bud. It took about ten years to bring this colour out clearly." Unfortunately, Herr Arends has not supplied exact dates of the origin of his varieties, but it seems clear that his blue has had no connection with the similar English variety and may possibly have been independent of Ferard's var. coerulea. Magenta and deep purple forms (Plate II, figs. 22 — 26) have also been produced and should perhaps be more properly regarded as belonging to the red than to the blue series. Size of flowers. The increase in the size of the flowers was one of the first of the changes noticed in P. obconica after it had been under cultivation for a few years. In 1887- a few plants were reported to have produced flowers nearly double the size of the original plants. In France the tendency to variation in this plant does not appear to have received much notice until 1892^ In this year M. Lille of Lyons brought out a variety grandifl^ra, and in the following year a variety with flowers larger than the type was produced and fixed by Messrs Vilmorin and sent out by them in the year 1894 with the designation "a grande fleur amelioree." In England in this same year plants grown at Gunnersbury Park are reported to have had flowers which approached very closely in size to those of the ordinary Chinese Primula*. In 1895, however, Mr Tallack, writing in The Garden', expressed the view that there had been little or no advance on the best flowers of former years, and that » Rev. Hon. 1907, p. 531. « Joum. Hon. 1887, May 26, p. 417 (N. G.). See also The Garden, 1886, March 13, p. 241, and Gard. Chroiu 1890, February 8, p. 175 (D.). 3 Rev. Hon. 1897, p. 374, 1899, p. 548, ihid. 1893, p. 123. * The Garden, 1893, March 25, p. 242 ; see also ibid. p. 327. ' The Garden, 1895, April 6, p. 240. 8 History of Primula it was not likely tbat further advance would take place owing to the difficulty in obtaining seed from the better varieties. Herr Arends who commenced experimenting with P. obconica in 1888, says that by careful selection and inter-crossing of the best strains he first succeeded in obtaining his var. grandiflora with larger pale- lilac flowers and that this was followed by various colour breaks in later years. At the present day the increase in size of the flowers in comparison with those of the plants originally introduced is very marked. The largest flowers have been noticed principally in pink and lilac-purple shades and have measured as much as 4 — 5 cm. in diameter^ (Plate I, figs. 13, 15; Plate II, fig. 30). Fimbriation. The commencement of the fimbriation of the corolla segments in P. obconica is a matter of some interest since this form of variation appears to occur, sooner or later, in most species of Primula under cultivation. Such variation of course is particularly noticeable in the cultivated Chinese Primula, and as far as can be ascertained from the earliest records this tendency to fimbriation had already been initiated when the plant was under cultivation in China^ In wild species of the genus, except in a few cases, the corolla segments show a simple notch at the apex^ but under cultivation fimbriation has developed in P. Forbesii, P. verticillata, P. kewensis, P. Sieboldii, P. japonica and also in the varieties of the common Primrose P. vulgaris. In the allied genus Cyclamen, fimbriation of the petals has also been developed as a result of cultivation*. Another point of interest in connection with the fimbriation of the corolla is that this variation has, without doubt, been developed quite independently in different places, and at different times. The earliest case of which a record has been traced occurred in the garden of Mr J. 1 See also Gartenflora, 1904, 53, p. 139, ibid. 1905, 54, p. 82, and Rev. Hart. 1906, p. 487. 2 The Botanical Register, 1821, torn. 539. 3 Primula Winteri {Gard. Chron. 1911, March 4, p. 130, Gard. Mag. 1911, March 4, p. 163, with figs.), is of interest in this connection as the corolla segments are strikingly fringed. P. Stuartii, P. petioloris and some other species also have fringed corolla segments in the wild condition. * See Thiselton-Dyer, "The cultural Evolution of Cyclamen latifolium," in Proc. Roy. Soc. Lxi. (1897), p. 143. A. W. Hill 9 Crook of Forde Abbey', blooms being said to show "distinct evidence of fringing of the edges." It should be added that these plants were supposed to be the result of crosses with P. sinensis but it will be seen later that the evidence in support of this view is very slight. The next record comes from France, about two and a half years later, a plant with fimbriated white flowers being exhibited by Messre Vilraorin in February 1896^ before the French Horticultural Society. M. Mottet writes in reply to queries addressed to M. Ph. L. de Vilmorin about this var. firnbriata : " The variety first appeared here in 1896 in a large-flowering white strain, in the form of slight denticu- lations on the edge of the petals. This was selected carefully and the variety announced in our catalogue in the spring of 1897," he adds that its origin had nothing to do with any seed or plants outside their own stock. In the early autumn of 1897 a var. fimhriata was being grown at Kew^ from seed obtained from Messrs Heinemann of Erfurt (cf Plate I, fig. 3). The fimbriated variety was also produced independently at Kew and the fringed character was attributed to the results of high cultiva- tion. As the variety was not considered an improvement its cultivation and selection at Kew were discontinued. In 1899 Messrs Vilmorin* were exhibiting this variety with rose-coloured flowers, and from thence onwards a var. fimhriata was ofiered in most seedsmen's catalogues (c£ Plate I, figs. 1.5, 17; Plate II, fig. 30). Mr A. W. Sutton writes that the fimbriated form appeared at Reading in 1901 and he adds " we do not., however, catalogue this, as the fimbriation has the effect of making the flowers look smaller." Double Flowers. One other striking floral variation in Pnmula obcanica has yet to be described, namely the occurrence of double flowers (Plate I, fig. 21). This appears to have originated with Messrs Vilmorin and as far as can be ascertained nowhere else. M. Mottet writing in Le Jardhv early in 1901 mentions that the double variety was put in commerce in that ' The Garden, 1893, Oct. 7, p. 327. 2 Rev. Hart. 1896, p. 238 ; ihid. 1897, p. 141. ' The Garden, 1897, September 18, p. 227 ; Gartenflora, 1897, 46, p. 143, text fig. 23. * Rev. Hon. 1899, p. 169; see also The Garden, 1904, April 7, p. 304; Gard. Ckrm. 1904, April 16, p. 244. » Le Jardin, 1901, p. 89, fig. 52, see also Rev. Hort. 1901, p. 238, figs. 100, 101 ; Jmtm. Hort. 1901, p. 14. In 1902 Sir Trevor Lawrence received an award of merit, R.H.S., for a' mauve-purple coloured semi-double var. of P. obcanica, see Joum. Hort. 1902, p. 548. 10 History of Primula year, and in a recent letter he informs me that the double variety may have appeared some years before it was sent out, Messrs Vilmorin do not remember exactly, but they state that at first the flowers were rather small and only half double but that since then the variety has been greatly improved in size and duplication of the flowers and that it now reproduces as well as any other variety. The colour is stated to have been pale rose like the type. Herr Arends informs me that he obtained seeds of the double form from Messrs Vilmorin and attempted to improve the variety by crossing with his own plants. In the course of several years he has obtained " larger flowers, stronger growth and rose-coloured flowers as well as the old lilac ones." Reference to modern catalogues shows that there are now several double varieties offered by nurserymen. Doubling of the flowers in the genus Primula as is probably well known, may take place in one of two ways. In the case of P. sinensis the modern varieties such as Crimson King,, etc., double the flowers by corona-like outgrowths from the corolla lobes at the back of each anther and the coloured surface of the outgrowth faces that of its corolla lobe or in other words the added lobes show the colours reversed. In the old double white variety of P. sinensis, known as P. sinensis var. jlore pleno, which appears to have been produced about the year 1839^ and which can only be propagated by cuttings, the doubling is of a "hose-in-hose" character, the colours of the added lobes not being reversed but corresponding in arrangement to those of the original corolla. In P. obconica the doubling is also "hose-in-hose" and there is no reversal of arrangement and colour as in the modern P. sinensis doubles. P. obconica, however, sometimes shows a tendency to doubling in yet another manner owing to the splitting of upgrowths of the corolla between adjoining corolla lobes, these ridges split towards the centre and so tend to produce a kind of corona with the colours reversed, but with this difference that the "corona lobes" alternate with the corolla lobes while in P. sinensis they are opposite to them. This method can at present be hardly considered as more than a tendency towards doubling, for it may only be shown by some of the flowers in an inflorescence ; it is however of considerable interest and it seems possible that by careful selection a new type of double P. obconica might be developed (Plate II, figs. 27, 29). 1 See Paston's Magazine of Botany, vi. 1839, p. 262. This sport appears to have arisen at Henderson's Nursery, Pine Apple Place, London. A. W. Hill 11 Variation or Hybridisation, That Primula obconica as cultivated to-day is a totally different plant from the wild species introduced from China in 1879-80 is a perfectly obvious fact, but as to the cause or causes which may underlie the changes which have taken place there is a considerable difference of opinion. Some of those who have worked for many years on this plant and to whom several of our modem improvements are due contend that the amelioration of this species is due in the main to hybridisation with other species of the genus, while others who have been equally successful in raising new varieties protest against this view and claim that the improvements are entirely due to selection and cross- fertilization of the best forms within the species. It is unfortunate that experiments with this plant have been undertaken from the horticultural rather than from the purely scientific point of view and that the results which tend to be matters of conjecture and a&sumption have been accepted in many cases as proved fact. For there has been a tendency in some quarters to assume that because a given variation appeared to fit in with a preconceived idea or with expectation, therefore such a variation was due to a definite experiment or series of experiments. In order to arrive at a conclusion as to the right explanation of the course of events displayed by the history of P. chconica it will be necessary to examine the evidence somewhat in detail. In the foregoing pages it has been shown that for some ten to fifteen years after its introduction P. obconica displayed but little tendency to vary. A slight increase in the size and slight changes in the colour of the flowers were recorded such as would be expected under artificial cultivation, but the bulk of the seedlings tended to come more or less true to the original form. According to Messrs Vilmorin* the first variations in P. obconica were noticed in 1892. In England variation both in size and colour is recorded on several occasions between 1886 and 1893^ but compared with present-day forms the improvement of the plant does not appear to have been very striking until about fifteen years after its introduction. Messrs Ware name about 1890 as their starting point, while Messrs Veitch consider the year 1898, in which their Feltham nurseries were opened, as the date from ' S. Mottet in Le Jardin, 1901, p. 8. 2 See especially The Garden, 1888, p. 550, 1890, p. 354, and 1893, p. 242 ; Joum, Hort. 1887, p. 417 ; Gard. Chron. 1890, p. 175. 12 History of Primula which their improvements should be reckoned. Even as late as 1897 "R. D." (R. Dean) writes in the Gardeners' Chronicle (1897, p. 65) that "so far comparatively little variation has appeared among seedlings. The blossoms of some are larger, rounder and stouter than others and there is a tendency to deepen the tints of some individuals to something approaching mauve." As with all new introductions gardeners very soon attempted to " improve " P. ohconica by crossing with other species of the genus, and P. sineTisis appears to have been tried in most cases as the pollen parent. As early as 1887 it was suggested^ that the variations noticed might be due to attempts to cross P. ohconica with Alpine auriculas and Primroses though it was considered as unlikely. That P. sinensis pollen was answerable for the improvement in the flowers is put forward in the case of the first occurrence of fimbriation recorded in 1893^ and several references to the action of P. sinensis are to be found from that dlate onwards. In 1896* Dr Masters exhibited a "hybrid" at the scientific committee of the Royal Horticultural Society supposed to be the result of crossing P. ohconica with the wild form of P. sinensis, but there was apparently very little to distinguish it from the female parent, and in the spring of 1898'' Mr Shea exhibited the result of a similar cross with (?) cultivated P. sinensis before that committee in which the influence of the Chinese Primula appears to have been accepted, though the predominance of the female is recorded. The possibility of hybridising P. ohconica with P. sinensis was accepted definitely in Germany and Herr Arends informs me that a fine batch of hybrids was raised at Fursten Walde near Berlin in 1893 "with the growth and leaves of ohconica and size and colour of sinensis flowers." He adds in a further letter that the plants had the " large brilliant flowers of sinensis. They represented in perfection that which we had tried to get for so many years." These plants all died without having been described or figured arid it is not now possible to say whether they may or may not have been hybrids, but in the light of our present knowledge it would appear to be a matter of considerable doubt. Herr Arends states that he has made this cross again and again but without result and Messrs Sutton, Veitch and Vilmorin^ all express the 1 Journ. Hort. 1887, p. 417. ^ The Garden, 1893, p. 327. 3 Gard. Chron. 1896, pp. 600, 790. * Gard. Chron. 1898, p. 119. « See Le Jardin, 1901, also The Garden, 1897, pp. 193, 197, 213, 216, 227, 394 ; 1899, pp. 144, 366 ; 1910, lxxiv. p. 179 : Rev. Hort. 1899, p. 548 ; 1906, p. 487. A. W. Hill 13 opinion that no hybridisation has ever taken place between P. obconica and P. sinensis. Mr Valentine, managing director of Messrs Ware's Nurseries, also says that all their improvements are regarded as being due to selection and cultivation. M. Mottet points out that the failure to produce artificial hybrids in the genus Primula is all the more curious since many natural hybrids in the genus are known as for instance among Alpine species and with P. ofwinalis, P. acaulis, and P. elatior. It has been suggested once or twice that P. obconica and P. sinensis may have hybridised naturally in the same way that P. kewensis arose in the first instances but this view is opposed by M, Mottet', who even goes so far as to say that even P. kewensis cannot be considered to be a hybrid. This latter case has, however, been proved more than once by artificial crosses carefully made at Kew'. Besides P. sinensis various other species of Primula have been used in the attempt to produce hybrids such as Alpine auriculas and prim- roses*; P.fiorihunda, P. vertidllata, P. japonica, P. farinosa, P. cortu- soides, P. sikkimensis^, etc., but with regard to all these it is stated that though seedlings were often obtained there was no evidence of hybridi- sation. In an account of an attempt to cross P. obconica with a well-coloured form of P. Sieboldii cortusoides "J. H. W." writes that the latter plant was used as the seed parent and every care was taken to prevent self- fertilisation. Seed was duly formed but the seedlings were nothing but P. Sieboldii cortusoides^. Several interesting varieties have been exhibited in recent years by the Duchess of Bedford; Mr Dickson, the head gardener, started the experiments in 1901 with a fimbriated variety of P. obconica and pollen of Polyanthi, Primroses and P. Sieboldii in varieties was used, later the pollen of P. cortusoides, P. sinensis and P. rosea. Mr Dickson claims that his results are due to the use of P. sinensis pollen and that one plant shows distinct evidences of the effect of P. rosea. In April, 1911, Mr Dickson showed a plant at a meeting of the Royal Horticultural Society^ under the name of " Chenies excelsior" (Plate II, figs. 37, 38) 1 See Rev. Hort. 1906, pp. 448, 449, fig. 176, where M. Grignan puts forward this suggestion to explain the origin of P. obconica tuperba raised by M. Nonin. 2 Rev. Hort. 1906, pp. 498, 499. » See Keic Bulletin, 1910, p. 325. * Joum. Hort. 1887, p. 417. ' The Garden, 1897, p. 193. • Gard. Chron. 1897, p. 128. ' Gard. Chron. April 29. 1911, p. 268. 14 History of Primula and he informs me that the pollen of P. japonica was used to fertilise the deep red-flowered plant of obconica mentioned above which was thought to show evidence of the influence of P. rosea. In general habit however the plant shows no trace of P. japonica either in leaves or flowers. The flowers are of a dark claret-magenta colour not more intense but not unlike that which has been produced by other growers. The inflorescences tend to become whorled as in P. japonica but this occurs not uncommonly with robust plants of P. obconica towards the end of their flowering season. The results achieved in the short space of about eight years are certainly very remarkable, but except for colour changes in the flowers there appears to be no evidence to support the view that the plants should be considered as hybrids especially as similar series of forms are known to have been produced elsewhere by selective methods alone without any recent attempts at hybridisation. Yet another species of Primula, namely P. megasaefolia from the Caucasus, is claimed to have been successfully hybridised with P. ob- conica. Herr Arends of Ronsdorf writes that he started working with P. megasaefolia, then recently introduced, about the year 1902 and produced in course of time the strain to which he gave the name P. obconica gigantea. Professor Pax to whom some of the plants were sent accepted them as hybrids between P. obcoriica and P. megasaefolia and in his account of the genus in the Pflanzenreich^ has given the name P. Arendsii, to this supposed hybrid. In habit this gigantea strain undoubtedly shows some differences from the ordinary grandiflora type in its stouter leaves and pedicels, but the flowers both as to calyx and corolla are those of P. obconica and the plants do not appear to show any character which can be definitely attributed to the influence of the pollen of P. megasaefolia. Forms of P. obconica closely resembling the gigantea of Arends appear to have been raised in different nurseries and by other means about the same time. P. obconica robusta raised at Lyon, for instance, by M. Choulet, is stated to be the same thing as P. obconica gigantea, but to have been produced entirely as a result of selection, Messrs Rivoire of Lyon write as follows on this subject : " Vous avez raison de mettre en doute I'origine supposee de cette Primevere et de vous refuser a croire a une hybridation ; c'est d'ailleurs, la aussi, I'avis de I'obtenteur. Nous ne pouvons que vous confirmer 1 Das Pftanzenreich, iv. 237, Primulaceae, p. 346. There is also a note in Gartenjlora, 1908, 57, p. 632 on P. obconica gigantea rubra, "the first true dark red hybrid of the new gigantea race." A. W. Hill 16 daus votre opinion en vous disant que nous avons, depuis I'apparition du Primula obconica, tent^ des hybridations avec un grand nombre de Primeveres. Nous n'avons jamais reussi, et les varieties que nous avons mises au commerce, telles que V^suve (a fleurs rouge carmin) et robtista, ont ete obtenues uniquement par voie de selection. " Nous connaissons d'autres horticulteurs qui ont tente de leur cote des hybridations, mais egalement sans succes. Aussi ne croyons-nous nullement a I'origine hybride signalee par I'horticulteur allemand qui annonce le Primula obcanica gigantea. "A ce propos, nous vous serious obliges de rappeler que le Primula obconica robusta, que nous avons annonce I'an dernier et qui a et^ obtenu par M. Choulet, chef des cultures florales du Pare de la Tete- d'Or, presente absolument les memes caracteres que ceux qui sont signales pour la variete gigantea, c'est-a-dire feuilles de consistance ferme, fleurs de dimensions tres grandes (les plus grandes connues 5 centimetres de diametre) de couleur blanc lilace, ombelles ^norraes et surtout tiges rigides, qui lui ont fait donner ce nom de robusta^." Whatever may be the explanation of some of the forms of P. ob- conica which have been obtained, it is evident that numerous attempts have been made to eflfect hybridisation with other species and that a great deal of work has also been done on the lines of selection and cross fertilisation of the best varieties. In the cases of the assumed hybrids it is remarkable that the results, whatever species may have been the pollen parent, are all strikingly similar and only a better form of undoubted P. obcanica has been obtained. Further the forms alleged to have been produced by hybridisation can hardly be distinguished from those produced by selection. In this connection also it is worthy of note that in the only case on record where the pollen of P. obconica was used the seedlings raised were purely of the type of the female parent- (P. Sieboldii cortusoides). The experiments in hybridi- sation appear to have been made with proper care in many cases and the conclusion seems to be suggested that the pollen may in some way stimulate the development of the ovule without efiecting hybridisation . The case of the orchid Zygopetalum Mackayi^ crossed with the pollen of other genera but always yielding seedlings closely resembling the female parent may perhaps be considered as a somewhat parallel case. It is true that Arend's gigantea strain shows a stoutness in the leaves which is more marked than in the ordinary forms and there is » Bev. Hart. 1906, p. 487. ' See p. 15, and Gard. Ckron. 1897, p. 12a » See Joum. R, Hort. Soc. xxi. 1897, pp. 476, 477; Orchid Review, vi. 1898, p. 19. 16 History of Primula also the case of a peculiar claret-coloured, small-flowered form produced by Mr Dickson, which is unlike any other variety I have met with. This latter plant was the only one of its kind produced by a cross alleged to have been made by P. obconica and P. rosea splendens. The leaf of this plant also differs somewhat from the normal though it is quite like that of some of the wild specimens preserved in the Herbarium at Kew. The whorled character of the inflorescence, also, which has been developed in the variety "Chenies excelsior," cannot be accepted as an indication of hybridisation with P. japonica since it may occur in uncrossed plants. The evidence for hybridisation in P. obconica cannot therefore be regarded as convincing. A careful series of experiments have been conducted at the John Innes Horticultural Institution at Merton in which the pollen of eight species of Primula has been tried. The results so far obtained tend to show that good seed has been produced only with the pollen of P. obconica itself all other crosses being failures, and this corresponds with the results of similar experiments made at Kew. It is just possible however in view of the conflicting evidence that further careful experiment might demonstrate some form of hybridisation for it may be, as Doncaster^ suggests in dealing with the question of crossing between species, that the multiplicity of characters concerned makes analysis very difficult and thus the evidence of hybridisation may not be apparent. It has been pointed out above that the difficulty of producing artificial hybrids in the genus Primula is somewhat remarkable in view of the fact that natural hybrids are not uncommon between certain species. Whether any natural hybrid between P. obconica and any other species exists is not certainly known but a specimen preserved in the Kew Herbarium collected by Wilson (no. 4052) in Western China suggests such a possibility. Mr J. F. Duthie who has kindly examined the plant is of the opinion that it may be a natural hybrid between P. obconica and P. cortusoides ; and in this connection it is of considerable interest to find that a plant (or plants ?) of P. cor- tusoides came up with the seed of P. obconica collected by Maries in Ichang, the specimen being preserved at Kew. Mr Duthie says of Wilson's plant — "It agrees with the former (P. cortusoides) in the shape, texture and pubescence of the leaves, but the calyx is that of P. obconica." In its native home P. obconica appears to show a considerable range 1 L. Doncaster, Heredity, Cambridge University Press, 1910, Chap. viii. p. 109. A. W. Hill 17 of variation, though the seed sent to England would appear to have belonged to a fairly uniform type. The plants lately collected by Forrest and referred by him to P. Listeri^, King, are now considered to be the variety glahrescens of Franchet. Many of the specimens, how- ever, show marked differences from the typical P. obconica and are also no doubt quite distinct from P. ListeH, but it seems open to doubt whether all of Forrest's specimens from Yang pi in Western Yunnan are rightly included under P. obconica and whether some should not rather be considered as belonging to a distinct though only slightly differing species, intermediate perhaps between P. obconica and P. Listeri. P. obconica and P. sinensis. The history of P. obconica and of the changes in form and colour which it has undergone in the comparatively few years of cultivation suggests that it may afford a parallel to the case of the long-cultivated Chinese Primula whose origin is still a matter of dispute and con- troversy. Primula sinensis unlike P. obconica was not introduced to this country as a wild plant but as a species which it is believed had long been cultivated in Chinese gardens^ P. sinensis, Lindl., as described and figured in the Botanical Register, 1821, t. 539, as P. prae- nitens, Ker.-Gawl., and figured in the Botanical Magazine, 1825, t. 2564, is not a wild form but a domesticated plant. The first figure published depicts a flower with the corolla segments fimbriated and it is of interest to notice that the later illustration in the Botanical Magazine shows the corolla segments with the notched apex and the plant is in general characters very similar to P. sinensis stellata of to-day. The two pictures are of interest in connection with the history of P. ob- conica, P. Forbesii, P. japonica, P. cortusoides, P. Sieboldii, P. kewensis, etc. The amelioration of P. sinensis both as to flower colour and shape ^ See Gard. Chron. 1909, November 20, p. 544, with figure. ^ The following account of Primula praenitens is given in the Bot. Reg. vii. 1821, t. 539. " It had been brought by Captain Bawes from the gardens at Canton, where it had probably found its way from some far more northern quarter of the Chinese Empire. Samples in a dried state had been previously submitted by Mr Reeves, a gentleman in the employment of the East Indian Co. at Canton." In the figare the corolla segments are fimbriated and the calyx has many lobes ; the account continues — "The plant not having been known in its wild state, can we be sure that the multiplication of the segments of the calyx does not arise from loxoriance induced by exotic cultivation?...." Joum. of Gen. ii - 2 18 History of Primula and the remarkable leaf development have proceeded steadily since its introduction so that now many of the cultivated races are very distinct from the plants introduced about 1821. What we may ask was the earlier history of the plant under the hands of the Chinese ? Is it too great a step to take to consider that the plant found by Henry in the limestone gorges of Ichang is really the original wild type of this species ? I for one, in the light of the history of P. ohconica, am inclined to think that it is not too great, and that we have in this little plant with its lilac flowers the true wild type of the species. Some corroboration seems to me to be given to this view by the \ Bxiety fiore pleno of P. sinensis. This plant somewhat closely resembles Henry's wild type in foliage and may be considered as offering a parallel to the old-fashioned double white and double lilac primroses which in the dim past must have been derived from P. acaulis. P. sinensis also offers another interesting parallel to P. ohconica in respect of the old double white variety since it appears that this arose as a sport about the year 1839 after P. sinensis had been in cultivation about eighteen years. P. ohconica has also yielded a similarly con- stituted double variety as a result of cultivation about twenty years after its introduction. ConclvMon. The conclusion to which one is led from the investigation of the history of P. ohconica under cultivation would therefore appear to be that the amelioration and development in form and colour of the flowers, etc. which have taken place during the past thirty years must be attributed to selective processes. The evidence which has been adduced in support of theories of hybridisation with other species is not sufficiently confirmed by facts to justify its acceptance. In view, however, of certain doubtful points and of some interesting questions as to the influence of foreign pollen in effecting fertilisation it would seem desirable to suspend full judgment until the results of further careful experiments in the fertilisation of P. ohconica with foreign pollen have been obtained. A. W. Hill 19 CHRONOLOGY OF /'. nliCONICA. (1 (2 (3 (4: (5 (6; (7 (8 (9 (lo; (11 (12 (13 (14 (is; (16 (17 (18 (19 (2o; (21 (22 Collection of seed in China . Plants flowered in England . White flowers first recorded. Dark eye first noticed . Increase in size of flowers . Var. grandiflora, M. Lille . Fimbriation first recorded, Mr J. Crook Rose-flowered variety {P. obconica rosea) Mt White variety second record Fimbriation second record, Messrs Vilmori Rose-flowered variety, Messrs Vilmorin Pure white flowers, Messrs Sutton Double flowers first put on the market Red-flowered var. "Vesuve," Messrs Rivoire Pure white flowers, Duchess of Bedford Pure white flowers, ilessrs Veitch " Crimson King," Herr Arends and Messrs Rirr Small Violet Blue variety, Duchess of Bedford . ■L'^'^6 ,, „ „ „ ,, „ „ Var. coerulea, M. Ferard ..... "Fire King," Sutton " Chenies excelsior " deep claret, Duchess of Bedford 1879 1880 1886 1887 1887 1892 lS9:i 1 s;*.") 1896 1896 1897 1899 1901 1903 1903 1904 1904? 1904 1906 1907 1909 1911 DESCRIPTION OF PLATES. PLATE I. 1. Pale lilac-flowered form with toothed corolla segments very near the type originally introduced from China. (3) 2. The same to show the calyx. (3) 3. Fimbriated pink variety similar to the first fimbriated form? produced. (1) 4. Side view of toothed pink variety. (1) 5. The same face view. (1) Figs. 1—5 from plants grown undisturbed for many years in the Temperate House, Royal Botanic Gardens, Kew. Figs. 6 — 12 illustrate the development of the blue strains of P. obconica. 6. Lilac-blue variety raised by selection, Messrs Veitch and Son, Feltbam. (37) 7. Calyx of the same. (37) 8. Lilac-blue variety with red eye. (27) 9. Calyx of the same. (27) 10. Blue variety raised by Mr Dickson, Head Gardener to Her Grace Adeline, Duchess of Bedford. (38) 11. Calyx of the same. (38) 12. Violet-blue variety with darker stellate eye, raised by Mr Dickson. (28) 2—2 20 History of Primula 13. Large soft pink flowered variety, raised by Messrs Veitch. (35) 14. Calyx of the same. (35) 15. Large pure white fimbriated variety, raised by Messrs Veitch. (33) 16. Calyx of this variety showing tendency to fimbriation. (33) 17. Eose-pink variety with crimson eye, fimbriated, Messrs Veitch. (29) 18. Calyx of this variety. (29) 19. Toothed white variety with conspicuous yellow eye, Herr Arends. (20) 20. Side view of the same flower showing the conspicuously lobed calyx. (20) 21. Double pink variety, hose-in-hose type, raised by Messrs Vilmorin, Paris, drawn from a plant sent by Herr Arends. (24) PLATE II. 22. Purple variety with conspicuous yellow eye, Herr Arends. (21) 23. Side view of the same flower. (21) 24. Calyx showing its hemispherical shape with unconspicuous teeth. (21) 25. Large-flowered deep purple variety with reddish eye, Messrs Veitch. (30) 26. Fimbriated calyx of the same. (30) 27. Deep rose-pink flowered variety with red eye, showing splitting of the corolla at the sutures between the corolla segments represented by colourless lines, cf. fig. 29, Herr Arends. (10). 28. Calyx of this variety with acuminate segments. (10) 29. Flower enlarged to show the splitting of the raised sutures between the corolla segments producing the commencement of a form of "reversed " doubling, cf. fig. 27. (9) 30. Very large pale pink fimbriated variety, raised by Messrs Veitch. (26) 31. Calyx of the same. (26) 32. Violet-pink variety with large crimson eye, Messrs Veitch. (28) 33. Calyx of the same. (28) 34. Deep rose-flowered variety with crimson eye, Herr Arends. (7) 35. Crimson variety with yellow eye, Herr Arends. (22). 36. Calyx of the same. (22) 37. "Chenies excelsior" (see Gard. Chron. April 29, 1911, p. 268), raised by Mr Dickson. (39) 38. Calyx of the same.. (39) Specimens of all these and other varieties are preserved in the Herbarium of the Boyal Botanic Garden, Kew, and the numbers in brackets at the end of each description refer to the numbers in the Herbarium. JOURNAL OF GENETICS, VOL. II. NO. 1 PUTE I ^ ^^ .t" >;^ ,1 12 ..1 r tL* > 9 1! •Ife '^ J ^^ r i6 W| 1. '7 20 I 4 21 r JOURNAL OF GENETICS, VOL II. NO. 1 PUTE II ACCOUNT OF A FAMILY SHOWING MINOR- BRACHYDACTYLY. By H. DRINKWATER, M.D., F.R.S. (Edin.), F.L.S. In the autumn of last year (1910) a medical friend resident in Liverpool informed me that a relative of his, whilst making his official medical inspection of school children, had seen a boy whose hands appeared to be of the same Brachydactylous type which I had described in a communication to the Royal Society of Edinburgh in November 1907. It naturally occurred to me that some femily whom I had already examined had removed to Lancashire ; but as soon as this boy's name was communicated to me I knew that he did not belong to any of the families already described. In December I wrote to the Headmistress of the school which the boy had attended and received the following reply : — Dec. 29. 10. Deab Sir, Your letter has just reached me having been forwarded from '. The boy, whom you refer to, has now left school. His parents live at . He is a rather peculiar boy, and dull by nature. At school we used to attribute his stupidity to the fact that his parents are related — (first cousins). His short fingers did not seem to hinder his manual work, but they are remark- ably short. I remember being told that the grandfather, on the father's side, had also very short fingers. The boy is now 13 years of age and is apprenticed to a joiner. Dr very kindly offered to have the boy's fingers examined at a Liverpool Hospital, but the parents refused their consent. The father is a rather intelligent man and by occupation a salesman at . If I can supply you with any further information I shall be very pleased to do so. I am. Yours Cuthfiilly, W. . ^ Names and addresses omitted for obTioos reasons. 22 Minor-Brachydactyhj This letter contains three statements which are of interest from the biological standpoint. (1) As to the brachydactylous condition of the boy's hands. (2) The presence of the same condition in a grand-parent. (3) The blood-relationship of the parents. Clearly it was a case for further inquiry. I endeavoured to get some more particulars by correspondence with some other people who knew the boy's family but without success. All the information obtain- able was that the boy's parents and all his brothers and sisters were normal (Fig. 1), and the only known brachydactylous member, besides i I (? X ? I ■ — — 1 I I ^ ' some brothers and sisters all normal Fig. 1. Erroneous Chart. himself, was his paternal grandfather^ Moreover the parents would not consent to have either a radiograph or photograph of the boy's hands taken. I mention these facts in order to point out the un-reliability of second-hand information, for it will be seen in the sequel how erroneous the statements were from all sources. Accurate details can generally only be acquired by personal investigations. I have paid two visits to this boy's family and their relatives and though I have not succeeded in persuading them to do all I wished, I have been able to gather together sufficient particulars to make it possible to describe the essen- tial feature of the abnormality, and so indicate its hereditary bearings. I have interviewed most of the boy's relatives including the great- grandmother, grandmother, uncles, aunts, and cousins, and have made several measurements and obtained some radiographs and a couple of photographs. This is most satisfactory considering the great reluctance of these people to do anything which can possibly lead to identification. I am greatly indebted to Mr Thurston Holland for the excellent radiographs which are amongst the best I have ever seen. The abnormality can be traced through five generations. The oldest surviving members of the family amongst the abnormals are the man No. 5 in the chart (Fig. 2) and his sister No. 7. No information 1 Three correspondents declared that the boy's parents were normal. H. Drinkwatkr s o -o+ -♦o > -♦o ^ "&. -♦o -♦o o ♦o -o -O- -o- -o ■^ -*•- 24 Minor-Brachydactyly could be obtained from No. 5 as he is bordering on senile dementia. No. 7 was most reluctant to give me any information and tried to mislead me. (As an example, I may mention that on asking about her brother No. 5, and mentioning him by name, she said she had no such brother.) The facts about No. 1 and his five children were however correctly given by her, and were confirmed by her mother, the widow of No. 3, and by other members of the family from tradition. The abnormals are represented by black circles and the normals by white. Where there has been the least doubt as to the individual being normal or abnormal the circle is included in brackets; these must not be counted in reckoning the ratio of abnormals to normals. The abnormals only are numbered. The information I first obtained to the effect that the schoolboy (No. 21) was the only affected member of his family was not correct, for it is seen that not only is a younger sister (No. 22) of the same type, but the mother also (No. 14): and it was not "the grandfather on the father's side" but the maternal grandmother (No. 7) who had transmitted the abnormality. There were 16 abnormal members of the family alive at the time of ,my visits and I interviewed each of them, but I regret to say that in some cases I was not allowed to take any measurements. What is the condition of the hands ? The abnormality resembles in many respects that described in my paper "An Account of a Brachydactylous Family^": whilst there are other features in which it differs. The fingers are not so short, and for this reason 1 propose to term the condition "Minor-Brachydactyly." The former family will be referred to as "No. 1 family." Fig. 3 shows the hand of the boy already referred to (No. 21). The upper hand is that of a normal brother who is two years his junior (No. 21a). The Brachydactylous condition is sufficiently evident, but that it is not so marked as in No. 1 family is shown by comparing Fig. 4 which is taken from my former paper. As the bones of this boy's hands are not yet fully ossified, it will be well, before describing them from the radiograph, to point out the ^ Proceedings of the Roy. Soc. of Edin. Vol. xxviii. Part i. H. Drfnkwatkr •J.) Fig. 3. Hands of abnormal and normal brothers. Fig. 4. Members of No. 1 Family. The lower hand is " Brachydactylous. " 26 Minor -Br achy dactyhj peculiarity as seen in the hands of an adult, and for this purpose I shall select the radiograph of a woman, No. 9 (Fig. 5). What chiefly strikes one is the shortness of the middle phalanx in each finger. It is to this peculiarity that the shortening of the hand is principally due. Fig. 5. Hand of woman ( x |). The middle phalanx is very short. In the normal hand, the middle phalanx is intermediate in length between the first and the third. All the phalanges are seen to be distinct and separate : there is no union (ankylosis) of the second to the terminal phalanx, such as occurred in No. 1 family in every case H. Drinkwater 27 in the first and little fingers and frequently in the middle and ring fingers also. (See Fig. 6.) Fig. 6. Hand of adolt woman (slightly redaced). Belonging to No. 1 Family. 28 Minor -Br achy dactyly The variations of the bones from the normal type are shown, in out- line, in Fig. 7 where A represents those of a normal finger ; B, those of a minor-brachydactylous finger, and C, a brachydactylous finger from No. 1 family. The phalanges are numbered 1, 2 and 3 in each case (3 is the terminal one which supports the finger nail). In G it will be observed that the second phalanx (2) and the third (3) have become united into one bone : whilst in B the second phalanx is short, but remains as a separate bone. ABC Fig. 7. Normal and Brachydactylous Phalanges. (Natural size.) A. Normal. B. Minor-Brachydactylous. C. Brachydactylous. What is the cause of this shortening ? Each phalanx during childhood shows, normally, a thin bony plate at its base. This plate is called the epiphysis; it is attached to the larger portion — the shaft of the bone — by an intervening layer of gristle, which, being transparent to the X rays, shows in a radiograph as a blank space. (Figs. 8,. 9.) H. Drinkwatkr 29 In adult life this piece of gristle becomes ossified, and with the shaft and epiphysis forms one bone. Fig. 9 shows the epiphysis at the base of the first phalanx and one at the base of the third but the second phalanx is seen to be without any epiphysis except in the case of the middle finger. This absence of epiphysis accounts for a good deal of the shortening of the finger, but not for the whole of it ; for it is obvious that the shaft of the bone is of less than normal length and the first and third phalanges are also slightly shorter than they should be. CT) Fig. 8. Outline of bones (not fully ossified) of brachydactyloos finger of a youth. 1. First phalanx with epiphysis (ep.) at base. 2. Second phalanx without epiphysis. 3. Terminal phalanx with epiphysis (ep.). There are three factors producing the shortening of the second phalanx : (1) The slight shortness of the shaft of the bone. (2) The absence of the epiphysis. These have already been referred to, but (3) There is still another factor and to this is perhaps attribu- table the chief share in the production of the shortening. I have already drawn attention to the cartilage or gristle between the shaft and the epiphysis. So long as this cartilage remains the bone can and does increase in length with the growth of the individual but when it has become ossified, then no further growth, in length, of the 30 Minor-Brachydactyly Fig. 9. Hand of boy showing absence of epiphysis at base of second phalanges except in the middle finger and thumb (natural size). H. Drinkwater 31 bone can take place at this point. Now this layer of cartilage does not become ossified in the average individual until about the twentieth year, so that until that age the phalanx can and does increase in length. If however ossification occurs prematurely, then the growth of the bone will be arrested and a permanent shortening will be the result. This is exactly what has happened in this family especially in the first and little fingers. In Fig. 5 the second phalanx is seen to be short in all the four fingers. This is the hand of a woman (No. 9). Fig. 10 shows the hand of her son (aet. 14). Here there is no sign of the epiphysis of the second phalanx except in the middle finger where it has already united to the shaft. In the other fingers it has never been present or has united during infancy. The abnormality in this boy will therefore be an almost exact repetition of the mother's. These cases represent the extreme type of deformity in this family. Fig. 11 shows a modification of this type in the adult. It is the hand of No. 12. The second phalanx is seen to be much more shortened in the index and little fingers than in the middle and ring fingers. Why is this? I think the explanation is furnished by the radiograph (Fig. 12), which shows the hand of her dauorhter aged 8. In this girl's hand there is an apparent absence of the epiphysis in the fourth finger, and in the index finger ankylosis has already occurred, whilst in the middle and ring fingers the epiphysis is still separated by cartilage. When growth is complete this hand will be like the mother's Avith the second phalanx much shorter in the first and fourth fingers than in the second and third. In another child of No. 12 aged 2 years the ossification of the cartilage has already occurred in the first finger (Fig. 13). The third factor concerned in the production of shortening of the middle phalanx is, therefore, premature ossification of the cartilage intervening between the shaft of the bone and its epiphysis. The essential feature of the abnormality in No. 1 family was stated to be "an absence of the epiphysis at the base of the second phalanx" with subsequent ankylosis of the second to the first phalanx, so that we have in the present family an abnormality which is essentially the same developmentally but stopping short of ankylosis. The second phalanx in the middle finger is less affected than in the others and this was also a characteristic of No. 1 family. 32 Minor-Brachydactyly The second phalanx of the thumb differs in the two families : in No. 1 the basal epiphysis was absent but in this family it is present. The abnormality in the toes appears to be practically identical in the two families for in both there is an absence of the middle phalanx- Fig. 10. Hand of boy ( x f ). No. 18 in chart. epiphysis and in the adults there is ankylosis of the second and terminal phalanx in the small toes. In each abnormal individual both hands and both feet are symmetrically affected, so that, in this respect, the peculiarity is as extensive as in No. 1 family. H. Drinkwater 33 Fig. 11. Hand of Woman (slightly redaoed). No. 12 in chart. Joom. of Gen. n 34 Mi nor- Br achy dactyl f/ Fig. 12. Hand of girl, aged 8. (Natural size. H. Drixkwatkr 35 .SP S— 2 36 Mino7'-Brachydactyly Fig. 14 is taken from an adult member of this minor-brachydactylous family. The measurements so far as they have been made are given in the following table. Measurements of Abnormcds. in Chart Age Middle Finger Hand Height 5 81 2f inches 6i: inches 62 inches 7 — 2| „ 6 »» 59 ,, 8 33 n >> H 60i ,, 9 64 — — — 10 51 H >» 7 ,, 62 »» 12 40 H >> H >> 60 »» 13 — n ,, 6J ,, 57 ,, 14 — 2f ,, 6| ,, 58| ,, 16 33 — — — 17 17 — — 57i ,, 18 15 — — 54| ,, 19 8 — — — 20 2 — — — 21 14 H )> 5 >> 54^ >» 22 5 2i ,, H »> 47i >> Scanty as the figures are they indicate very decidedly two charac- teristics, viz. : (1) The shortness of the fingers. (2) The shortness of stature. (1) The average length of the middle finger of the seven adults (male and female) is 2*57 inches which is fully three-quarters of an inch less than the normal. The shortness of the hand of the boy (No. 21) is shown not only by the photograph (Fig. 2) but by comparison with the hands of two of his younger brothers as shown in the following table : Age Finger Hand No, 21 (abnormal) 14 Brother (normal) 12 7 2^ inches 2| „ 2| „ 5 inches 6 „ (2) The shortness of stature is well shown by the photograph of No. 21 (aged 14) and his younger brother (aged 12) who is normal and the taller of the two. (Fig. 15.) The average stature of the four women Nos. 7, 12, 13 and 14 is 58'6 inches and of the three men Nos. 5, 8 and 10, 61"5 inches. H. Drinkwater 37 Fig. 14. Foot of adalt showing abortive middle phalanx. (Natural size.) 38 Minor -Br achy dactyly iig. 15. Photograph of two brothers. The shorter boy is aged 14, Brachydactylous. The taller is aged 12 and is normal. J H. Drinkwatkr 39 Now these figures are very remarkable from their close approximation to the same measurements in No. 1 family where they were respectively 58J and 61 inches. The women are therefore about 4 J inches and the men about 8 inches below the normal height. It is the general opinion that the abnormals have better health than their normal relatives. The abnormals are slightly more prolific than the normals though the numbers are too small to enable one to draw conclusions on this point. Increased fecundity was a marked feature of No. 1 family. In both families a much larger proportion of normals have remained unmarried. The schoolmistress stated that the parents of the boy (No. 21) are cousins but such is not the case and I could not hear of any intermarrying in the family. Mendeli^m. This family illustrates certain Mendelian rules : (1) There is perfect segregation. The abnormality is either not transmitted or it is transmitted fully : i.e. so as to involve the digits of both hands and both feet, (2) The abnormality is transmitted only by the abnormals and never by the normals, so that all the descendants of normals are normal. (3) The offspring of parents, one of whom is abnormal (= dominant) the other normal (= recessive), should theoretically show 50 % of each type. The descendants, in this family, counting only those whose type is known for certain, amount to 47, and of these 21 are abnormal : i.e. 21 abnormals to 26 normals giving 446 instead of the theoretical But this percentage is not to be regarded as positively correct. It is certain that it is as high as this, but not at all certain that it does not more closely approximate to the theoretical number. When I first interviewed the woman No. 9 in the chart she informed me that of her ten children only one had short fingers like her own, viz. her eldest daughter (No. 16) and a casual inspection would have confirmed her statement. The shortening is so incon- spicuous that in some of the children it is only detected by fiexing the finger, and then the shortened middle phalanx is uoticciible but 40 Minor-Brachydactyly not otherwise. By this means I was able to show her that her two youngest children's hands were brachydactylqus and this is confirmed by the radiographs. In adult life the shortness of the hand is conspicuous enough and cannot be overlooked, but this is not so during childhood, so that it is possible that of the few children whom I could not see and who were declared to be normal, one or two may be of the abnormal type, and if so would make the percentage of abnormals still more closely approxi- mate to the theoretical figure. I have been able to obtain radiographs of the hands of Nos. 9, 12, 14, 17, 18, 19, 20, 21 and 22 and of the feet of all the same except No. 9. [Figure 4 is reproduced by kind permission of the Royal Society of Edinburgh.] The expenses in connection with this investigation have been defrayed by a grant made by the University of Edinburgh from the Earl of Moray Endowment for the promotion of original research. A CRITICAL EXAMINATION OF RECENT STUDIES ON COLOUR INHERITANCE IN HORSES. By a. H. STURTEVANT, Columbia UnivertUy. About a year ago I published a paper on the iuheritance of colour in the American Harness Horsed I concluded that the colour of these horses is, in general, determined by five factors : C (chestnut or yellow), hypostatic to the others and always present ; H (Hurst's factor or black) ; B (bay), epistatic to H ; R (roan), and G (gray). There were at that time two other papers on colour inheritance in horses with which I was not acquainted. The first, by Dr E. H. Harper^ was not written from a Mendelian standpoint, but contained confinnatory evidence for my views about black and gray. The second, by Prof. James Wilson', had already covered most of the points which I brought out, and some others as well, although Wilson does not seem to have had a very clear idea of the factors concerned, as is shown by his attempt to represent the gametic constitution of his stallions by only two symbols each. There is one point on which we reached quite diflferent conclusions — namely, the position of brown, which I shall discuss later. These three papers deal with five diflferent breeds, the English Thoroughbreds and Shires and Scotch Clydesdales being treated by Wilson, the French Percherons by Harper, and the American Harness Horses by myself. It is my purpose in the present paper to compare and combine the contents of these three contributions. Wilson and I agree that chestnut stands at the bottom of the scale, and neither of us found any horse lacking the factor for it » Biol. Bull. XIX. No. 3, August 1910, p. 204. « Biol. Bull. IX. No. 5, October 1905, p. 265. 5 Seient. Proc. Royal Dublin Soc. 12 (N. S.), No. 28, 1910. 42 Colour Inheritance in Horses The next factor, that for black, H, was discovered and correctly interpreted by Hursts Wilson's statistics show that this factor is carried by bays, browns, and blacks, though he gives no factorial interpretation of his results. The figures for matings of chestnuts, i.e., of h h individuals, are as follows: — Colour of Foals Breed Chestnut Black Brown Bay Authority Thoroughbred 1095 9 (bay or brown) Hurst Shire 44 1 1 5 Wilson Trotter 69 0 0 0 Sturtevant Total 1208 16 not chestnuts Hurst also gives five other authorities for the statement that chestnut breeds true, in various other breeds of horses. Hurst, Wilson, and I all found a good many sires producing no chestnut foals, i.e., homozygous for H. These are blacks, bays, browns, grays, and roans. The total foals from heterozygous sires and chestnut mares is as follows : — Breed Thoroughbred Clydesdale . . . Trotter Total Chestnut Not Chestnut Authority 347 355 Hurst 3 2 Wilson 65 56 Sturtevant 41c 413 My next factor was B, or bay, and I considered brown as being usually a heterozygous form (GHBb), although I also realised that horses with the above constitution are often bays. I found what I thought was a brown stallion with the formula GHBB, but this horse (Prodigal) as I have since found, is recorded in the later volumes of Wallaces Year Book (my principal authority) as a ba}^ The change was made after he had become famous as a sire, and is therefore probably correct. I was inclined to explain the four browns recorded as produced by two black parents by supposing that some GHhh horses are browns. I now believe these four cases to be errors in the record, this opinion being based on Wilson's and Harper's figures for matings of blacks to blacks (see below). Wilson reached a quite different conclusion. He regards bay and brown as dominant to black. His idea as to the relation between bay and brown is shown by the following passage. " The relative positions 1 Proc. Royal Soc. 77 B, 1906, p. 388. A. H. Sturtevant 43 Black X Black Colour of Foals Braed BUck Bay Brown Chedtnut AnUMrity Percheron . . . 49 (2 not black) Harper Shire 39 3 0 2 Wilson Clydesdale ... 36 0 2 0 Wilson Trotter 34 2 4 2 Stnrterank Total 158 Black X Bay Thoroughbred 1 33 27 14 Wilson Shire 39 125 43 19 Wilson Trotter 16 48 31 7 SturtCTant Clydesdale 40 104 67 7 Wilson Total 96 310 168 47 Black X Brown Thoroughbred 8 12 20 0 Wilson Shire 39 19 36 4 Wilson Clydesdale 61 34 106 1 Wilson Trotter 11 5 9 1 Sturtevant Total 119 Bay 70 X Bay 171 6 Thoroughbred 1 1295 125 270 WUson Shire 13 287 18 28 Wilson Clydesdale 6 243 59 5 Wilson Trotter 1 46 3 9 Stnrtevant Total 21 1871 205 312 Total Bay X Brown Thoroughbred 10 744 365 123 Wilson Shire 23 133 56 5 Wilson Clydesdale 25 206- 254 5 Wilson Trotter 9 81 31 8 StorteTant 67 1164 706 141 Brown x Brown Thoroughbred 6 78 114 11 Wilson Shire 7 20 27 2 Wilson Clydesdale 32 34 165 0 Wilson Trotter 5 7 7 0 Stortevant ToUl 50 139 313 13 44 Colour Inheritance in Horses of bay and brown remain to be settled ; and although there is evidence in favour of brown being dominant to bay, this conclusion is not clearly established. It must be remembered these are the colours breeders have the greatest difficulty in discriminating; and errors affect sires and dams and foals. In regard to sires it has been possible to correct the registered colours in several cases ; and while every correction has increased the evidence in favour of brown being dominant, it is still possible that there may be other explanations, as, for instance, that bay is a diluted brown." I shall first give the figures bearing on this question, and then discuss their significance in connection with the two views given above. My hypothesis that brown is a heterozygous colour was based upon two facts. In the first place, I am unable to see any very sharp line between bay and brown. "Wilson evidently thinks the two colours are distinct, but I can find no definite statement as to what the difference is, although he discusses the distinction between the various colours at some length. Secondly, all the browns from which I could find any fair number of foals produced some blacks, with the exception of the stallion Prodigal, which, as I have explained above, now turns out to be a bay. I have found 15 such brown sires producing black foals. In Wilson's tables appear five brown and one doubtful bay or brown Clydesdale sires, all of which have a fair number of black foals. There are also five brown Thoroughbreds. Two of these sired no blacks among 93 and 95 foals, respectively, though each has a foal recorded as "brown or black." Ladas has one black and one black or brown among 97 foals, Desmond three black and two brown or black among 48 foals, and Wolf's Crag seven black and one brown or black among 95 foals. Here we meet an interesting fact — the extreme scarcity of blacks among English Thoroughbreds. Wilson believes that most, if not all, the recorded blacks are really browns, and was not able to find a genuine black stallion. The Desmond mentioned above is recorded as a black, but Wilson ascertained that he is really a dark brown, and the same was found to be true of all the mares recorded as black of which he could get definite and reliable informa- tion. Only about 1 "/„ of these Thoroughbreds are recorded as blacks. Of course by my hypothesis it would be hard to explain how there could be about ] 4 °l^ browns and only 1 °/„ blacks. However, I shall not try to explain this, as the hypothesis is pretty well disproved by the result of mating browns together. If it were correct such matings should produce 25 °/^ blacks, 25 °/^ bays, and 50 "/^ bays and browns. A. H. Sturtevant 46 As is shown above they do produce a trifle less than 10 "/^ blacks. I am still convinced, however, that there is something in my original hypothesis. It could scarcely be a coincidence that twenty brown sires should be heterozygous and none homozygous, if we except the Thoroughbreds. Among the latter I believe that Wolfs Crag, Desmond, and perhaps Ladas are heterozygous, for it is evident that in a population containing only 1 °l^ recessives there would not be a large proportion of heterozygous mares, and a heterozygous stallion would of course not produce recessives when mated to pure dominant mares. Again, it will be noticed in the tables above that, excluding chestnut foals, black mated to bay gives only 16% black foals, while black mated to brown gives 33 7o> or twice as many. And while brown to brown gives only 10°/^ blacks, bay to bay gives still fewer — only about 1 %, or 3 7o> i^ w® omit the Thoroughbreds, of which the table illustrating this class has a relatively large proportion. According to Wilson's hypothesis that brown is dominant to bay, bay to bay should produce no browns. This would require the further hypothesis that the 205 browns recorded from such matings are errors in description, which certainly does not seem to me to be probable. Again, there should be some brown sires producing no bay foals, but as a matter of fact all of the 25 brown sires found produce a large proportion of bays. I have, in fact, never yet found a sire which did not produce bays. Finally, as stated above, I am unable to agree with Wilson that bay and brown can be satisfactorily separated. I base this upon my own observation, upon the frequent changes from bay to brown and vice versa which he mentions finding in the Clydesdale records, and the similar changes which I have observed among Harness Horse records, and upon the frequent recording of English Thorough- breds as "bay or brown." My conclusions, then, are that brown and bay are not distinct, brown being merely a dark bay, and that brown is more often CHBb than CHBB, and never CHbb. It would be interesting to know whether or not the heterozygous bays are darker than the homozygous ones. In regard to gray there is no great difficulty. I thought when I published my first contribution that I had a non-conformable case, where two brothers not gray were siring about 50 % gray foals. One of these, Dispute, I now know to have been wrongly recorded, as his owner, Mr John Taylor, and another breeder, Mr W, B. Gill, both write me that he is a gray. The supposition therefore is that his brother was also a gray, though I have been unable to verify this. But if he 46 Colour hilieritayice in Horses was, then the whole family works out exactly according to expectation. The produce from one gray parent and one not gray is as follows : Breed Gray Foals Not Gray Foals Authority Percheron 31 29- ■? Harper Thoroughbred . . 73 56 Wilson Shire 146 186 Wilson Clydesdale 9 15 Wilson Trotter ... 141 142 Sturtevant Total 400 428-? Of the 428 — foals neither gray nor chestnut 38, plus an unknown number of the Percherons, are blacks. This gives a clue as to whether or not the bay factor is necessary before 0 can cause gray. If it is, then these 38 + blacks should be eliminated from the table, and the result would be 400 grays to 390 — bays and browns, or, leaving out all Percherons, 369 grays to 361 bays and browns. Now gray is an unpopular colour, and it seems very unlikely that as large a proportion of the gray foals would be recorded as of the other colours, and I should therefore expect fewer grays than the calculation calls for, rather than more. For this reason, and because no black has yet been found to carry the G factor, I believe that G can cause gray in the absence of B, though this is by no means proven as yet. Grays to grays give the expected 3 to 1 ratio. Colour of Foals Breed Bay Black Brown Gray Authority Shire 3 1 0 12 Wilson Thoroughbred ... 0 0 0 1 Wilson Percheron 10 not gray 31 Harper Trotter... 1 0 0 1 Sturtevant Total 15 not gray 45 Just what this G factor is is a rather difficult question. Gray differs from the colours hypostatic to it both in the possession of white hairs, and in the mottled or dappled pattern. Now this dappled pattern, or one very similar to it, is to be seen also on chestnuts, bays, browns, duns, and sometimes even blacks. In the case of browns, the spot inside may be lighter than the ring around it, that is it may have more brown hairs, or this condition may be just reversed. If there is a separate dappling factor, then gray would not act like an ordinary colour due to a single factor, but would be produced by non-gray parents, unless the dappling factor is present in most horses. Many A. H. Sturtkyant 47 hoi-si> M ni Mil the .^tKt ts do not show dappling, but it is possible that it would show on most of them at the time they are shedding their coats. This whole question is one which can be settled only by further investigation. In my former paper I suggested (p. 215) that perhaps all horses carrying the roan factor are roan, the type of roan, or the ground colour, depending upon the colour the horse would have been if it had not had that factor. Wilson had already made the same suggestion, as I now find, but he has given little more evidence on the subject than I did. I have written to a good many breeders of Trottere, in order to try and get information which would help out on this point, but have not much to offer as yet. What little I have, however, is not encouraging. In my tables of sires homozygous for the bay factor (see former paper), and therefore producing no black foals, were two roans. Jay Bird and Margrave. I am informed that both these stallions were bay roans, and that Jay Bird's roan foals are also bay or red roans. Margrave likewise has some bay roan foals and a full brother of the same colour. This is according to expectation, but I am also informed that Margrave has at least one blue roan colt. Moreover, his former owner, who now owns the full brother mentioned above, writes me that their dam, Spanish Maiden, was a blue roan. Not only did this mare produce a supposed BB colt, but she was also the daughter of a BB stallion, Happy Medium. Margrave has 65 non-black foals and Happy Medium has 69, neither being credited with a single black, though eleven of Happy Medium's bays and browns are from black mares, so that it is very improbable that both should in reality be Bb animals. Either Spanish Maiden and probably the Margrave colt are not blue roans, or blue roans can carry the B factor. Wilson's tables give some information not brought out elsewhere, so I shall take the liberty of reproducing them, in part. He considers iron gray as a kind of roan. Of course his records may use that term in a different sense' from that in which American horsemen do, but unless that is true iron gray is not a roan at all, but merely a young gray. We, in America, mean by that term a very dark gray, which the casual observer might mistake for a roan, but which later develops into the ordinary dapple gray, and, still later, into white with black hairs in the mane and tail, and on the feet It is, moreover, the invariable colour of a young gray, so far as I know. Wilson classifies these iron grays separately but has only a few of them, and I shall omit them because of my doubt as to their real position — a doubt which his data 48 Colour Inheritmice in Horses do not help to clear up. The tables below were drawn entirely from the Shire records. Just what roan means is not made clear. Does it include only red or bay and chestnut roans, or may it include some blue roans whose colour is not more definitely stated in the records ? One Parent Eoan Colour of Foals dnlnnr orAfTiPT _, Parent Chestnut Black Bay Brown Gray Blue Roan Roan Chestnut . . . 7 2 7 2 1 0 8 Black 0 5 0 2 0 3 9 Bay 4 2 30 11 1 1 39 Brown 1 2 14 7 1 1 19 Gray 0 0 0 1 5 1 6 BlueEoan... 0 0 2 0 2 0 2 Eoan 0 0 1 0 0 0 5 One Parent Blue Roan Colour of Foals Colour of other Parent K Chestnut Black Bay Brown Gray Blue Roan Roan Chestnut ... 0 0 1 0 0 1 0 Black 0 1 0 0 0 0 0 Bay 1 1 4 1 0 1 0 Brown 0 2 1 2 0 5 2 Gray 0 0 0 2 0 0 0 Blue Bean... 0 0 0 0 0 0 1 Boan 0 0 0. 0 0 0 1 There is no evidence in this table against the hypothesis reached independently by Wilson and myself, but neither is there any of very much significance in favour of it. That black to blue roan gives one black, and blue roan to blue roan gives one roan is all very well, provided the one roan is not a red roan, but it does not take us very far. The table is of value, however, as showing that roan to chestnut black, brown, or bay, gives 80 roans to 118 not roans. Adding my figures for the same cross gives 225 roans to 284 not roans. I believe that this deficiency in roans is at least partly due to the fact that roans vary in the amount of white they show, and the breeders are apt to avoid calling a roan such if they can help doing so, as the colour is unpopular. Still, if this practice were very general it should lead to many roans being produced with neither parent recorded as roan, which is not the case, such records being rather rare. It is quite possible that there are other complications in this case. Neither Wilson nor I had enough evidence to enable us to make even a safe guess as to the relation between gray and roan. I have A. H. Sturtbvant 49 since obtained more information however, mating grays to roans. Below are the results of Colour of FokU Breed Shire ... Trotter.. Black 0 1 Brown 3 0 (;r»y 5 2 Roan 7 4 Anthority Wilson Startevank Total 1 11 The expectation is 50 % of the epistatic colour, 25 7o of the other, and 25 °/^ of colours hypostatic to both, i.e. brown, bay, etc. There are 50 % roans, but the numbers are small, and the grays are dangerously close. Wilson's tables show five grays produced among Shires by mating roans to chestnuts, browns, bays, or roans, but he says a similar number of cases of gray apparently carrying roan could be found. I have found four cases where gray has seemed to carry roan, two of which, however, are from doubtful sources, and the other two, of course, might easily be due to mistakes somewhere, I have found no case of roan carrying gray. Of the above four roans from gray by roan, two were reported as gray roans, one was a bay roan, and one either bay or chestnut roan. Of Wilson's, one is a blue roan, and the others are roans, the type not being made clear. The most probable view of the matter seems to me to be that the G R horses are blue or gray roans, which may perhaps be sometimes mistaken for grays and recorded as such. I did not treat dun in my paper, but Wilson has worked with it He assumes, though without saying so, that all duns are due to the same factor. This may be correct, but there are certainly at least two types of duns, differing from each other much more than do the bay and brown he is so careful to separate. One type has black " points," and Gray ? I Dan an(f black points (By Bay Trotter) 60 Colour Inheritance in Horses the other has very light, almost white, mane and tail, and no black on the feet. The light type is sometimes described as cream-coloured. That the two are related the pedigree given here indicates, though just what the relation is does not appear. Lumping all duns together and adding my figures to Wilson's give the following as the result of mating duns to various colours. Colour of other Parent Colour of Foals - — Chestnut Black Brown Bay Gray Roan Dun Chestnut ... 0 0 0 1 0 0 1 Black 1 1 0 1 0 0 2 Brown 0 0 0 1 0 0 1 Bay 1 1 0 2 0 1 2 Gray 0 0 2 0 7 0 5 Dun 1 0 0 1 0 0 2 Wilson gives a case of gray to gray producing dun, and the above pedigree shows three gray mares which carried a dun factor. Therefore, assuming that there is a single dun factor and making no attempt to explain the differences among the duns, we should agree with Wilson in placing the colour between bay and brown on the one hand and gray on the other. But such an assumption is rather hazardous as the case stands now, and I should not like to make any generalizations about dun until more evidence is at hand. Mr W. P. Newell has supplied me with information about an interesting family of white horses. The ordinary white horse is of course merely an old faded-out gray, but this is a family of real whites. Mr Newell gave Professor W. E. Castle some information about these horses, on the basis of which Professor Castle considered the colour to be an extreme spotted condition dominant to the ordinary colours^ I have now some further information, which makes the case an interesting one. These horses are said to be somewhat variable in colour. To use my informant's words : " The colour of skin is white or so-called pink, usually with a few small dark specks in skin. Some have a great many dark spots in skin. These latter usually have a few dark stripes in hoofs ; otherwise the hoofs are almost invariably white. Those that do not have dark specks in skin usually have glass or watch eyes, otherwise dark eyes.... I have one colt coming one year old that is pure white, not a coloured speck on him, not a coloured hair on him, and with glass eyes." The term "glass eye" means a white eye. Therefore the colt described above is almost an albino in appearance. 1 Breeder's Gazette, lix. 15, 1911, p. 948. A. H. Sturtkvant 61 However, his sire is one of the dark-eyed somewhat spotted whiten, bis dam being a brown Trotter. Since "glass" eyes occur not infrequently in pigmented horses it seems probable that this white-eyed albino (?) is really an extreme case of spotting, plus an entirely independent "glass" eye. Mr Newell writes that white mated to white gives about 50 7o white to 50 % pigmented. He reports only three matings of white to white. The results of these were, one white, one roan, and one gray. Apparently, then, the white factor stands at the very top of the series. However, I am not sure that this is the whole story, as it would be a peculiar coincidence if it were a mere accident that the only two non- whites produced from white by white are representatives of the only two other colours having white in the coat, and these both such un- common colours. On the other hand, it does not appear that white by pigmented gives a large percentage of grays and roans. In addition to those already mentioned I wish to extend my thanks to the following horsemen, who have supplied me with information used in this paper: Messrs S. J. Fleming, W. B. Wallace, D. W. Northrop, T. Stememan, and G. M. Garth. Summary. It seems probable that chestnut always breeds true. Therefore the placing of C (chestnut or yellow) at the bottom of the scale probably represents the condition of nearly all breeds of horses. Epistatic to it is H (black). Next comes, in the breeds studied, B (bay or brown), epistatic to both the preceding. G (gray) is next higher. Next is R (roan), which is probably always evident when present' and which probably merely causes a sprinkling of white hairs, without otherwise affecting the colour. Finally, we have W (white). 1 Unless sappressed by the next faetor, W. 4— S A FURTHER CONTRIBUTION TO THE STUDY OF RIGHT- AND LEFT-HANDEDNESS. By R. H. COMPTON. r In a paper presented to the Cambridge Philosophical Society in June 1910* I discussed the phenomena connected with the occurrence of two stereo-isomeric forms of seedling in two-rowed Barley (Hordeum distichum); the difference between them being shown in the mode of folding of the first leaf above the coleoptile (Diagram 1). The two L£/\f ri _.COL£OP- R.H »-•" Diagram 1. kinds of fold were called respectively right- and left-handed, the con- vention adopted being explained in the introduction to the paper. 1 B. H. Compton, " On Bight- and Left-Handedness in Barley," Proe. Cambridge PkiL Soc. XV. p. 495, 1910. 54 Study of Right- and Left- Handedness The difference between the two kinds of seedling may appear at first sight to be of an insignificant character, and not worthy of serious attention. That this is not so is amply evinced by the detailed account of the morphology of the plant given by Schoute^ in his fine work on the tillerage of cereals. It is there shown that the direction of folding of the first foliage leaf has a direct connection with the side towards which (in Barley, but not in other cereals) the axillary bud is displaced from the median line, and also with the side of the plant on which the first foliage leaf (after the prophyll) of every lateral axillary branch is produced. The successive leaves, arranged distich ously on the stem, normally alternate in their mode of folding^: thus all the leaf edges on one side of the plant lie beneath those on the other. Since the first foliage leaf of the axillary bud is always produced in the plane at right angles to that of its subtending leaf and on the side towards its under- lying edge, it follows that all the first foliage leaves of lateral axes of the first order are towards the same side of the main axis. In these and other ways it is demonstrated by Schoute that the mode of folding of the first leaf is closely bound up with the symmetry and morphology of the whole plant. Unfortunately reverse conventions are used by Schoute and myself in distinguishing right- and left-handed plants. In what follows, however, I propose to adhere to the convention adopted in my earlier paper. Two-Rowed Barley. In my previous paper it was concluded that there is no hereditary nexus between successive generations of two-rowed Barley in respect of right- and left-handedness. The statistics advanced were somewhat restricted, and did not attain the same degree of accuracy as those advanced in reply to the other questions proposed. Further results have since been acquired and are given below in order to place my previous conclusion on a firmer basis : for the conclusion itself is unaltered. ^ J. C. Schoute, "Die Bestockung des Getreides," Verh. d. K. Akad. v. Wetensch. t. Amsterdam (Tweede Sectie) ; Deal xv. No. 2, pp. 8—24, Feb. 1910. 2 Failure of this regular alternation of LH and RH leaves is mentioned by Schoute {loc. cit. p. 20), and also by Stratton and Compton, "On Accident in Heredity, with Special Eeference to Right- and Left-Handedness," Proc. Cambridge Phil. Soc. xv. p. 508, 1910. I made observations on twenty Maize plants, following them as far as 6 — 10 leaves, and noting the fold of each leaf as it appeared : of these thirteen showed the normal alternation of RH and LH throughout ; the other seven showed a disturbance of the regular sequence. R H. COMPTON 66 In 1910 a number of ears of " Kinver Chevalier" Barley were sown separately in the ground* and the plants produced were grown to maturity. Four of these families were selected for further study by reason of the eccentricity of the ratios of LH to RH which they exhibited': it being thought that such ears might give the clearest indications whether the characters or the ratios were hereditary. Ear I gave an unusual excess oi LH. seedlings; ears II, III and IV an excess oiRU. Table I summarises the results obtained. In the first column is given the reference numeral to the ear in question together with the fold of the first leaf of the plant which bore it (in 1909). Columns 2 and 3 show the numbers of LH and RH seedlings produced from the ear in 1910. The ears yielded by the 1910 plants were harvested separately and were sown on wet string-canvas. The numbers of LH and RH seedlings so produced were counted and the results are dis- played in the rest of Table I. The data are classified according as the TABLE I. Parent £ar, 1910 Offspring 1911 Na and LH RH From LH plants From RH plants Total twist 1909 LH RH LH/RH LH RH LH/RH LH RH LH/RH I R 15 3 845 622 1-36 123 69 1-78 968 691 1-401 II R 6 8 88 69 1-28 197 131 1-50 285 200 1-4-25 ni L 5 9 477 332 1-44 794 547 1-45 1271 879 1-446 IV L 7 12 550 378 1-46 495 409 1-21 1045 787 1-328 Totals 1960 1401 1-397 1609 1156 1-392 3569 2557 1-396 58-28 7 ,LH 58-19 °/o LH 58-26 %LH seedlings were the oflFspring of left-handed (cols. 4 and 5) or right- handed (cols. 7 and 8) parents in 1910. In columns 10 and 11 are given the totals from LH and RH parents added together. The ratio LH/RH is calculated in each case. In the lowest line of the table are given the sum of each column and the ratios for all four ears taken together. The bottom line of the table shows at once the extreme closeness of the ratios found among the offspring of LH parents on the one hand and RH parents on the other. This appears to be conclusive evidence 1 The results of these sowings were given in my earlier paper, p. 504. * The average ratio for this variety of Barley was found to be 1-390 LH : IRH (58-18 =/„Lfl). 56 Study of Right- and Left-Hmidedness that the direction of folding of the first leaf is not inherited. The same conclusion is to be reached from a consideration of the offspring of each individual ear I — IV. Ear I, which produced a great excess of LH plants in 1910, gave in 1911 a ratio almost identical with the average : ears II — IV, which gave an excess of RH plants in 1910, also approximated to the same ratio in 1911. Thus the conclusion of my earlier paper with respect to the non- inheritance of right- and left-handedness in the fold of the first leaves of two-rowed Barley is amply confirmed. This must be clearly distinguished from the fact that the ratio between rights and lefts is approximately the same in successive generations, as shown in Table II for Kinver Chevalier Barley. TABLE 11. Year LH RE LHIRH Per cent. LH 19091 730 546 1-337 57-21 19102 379 259 1-463 59-41 1911 3569 2557 1-396 58-26 Total 4678 3362 1-390 58-18 We may say that the ratio LHjRH is hereditary though right- and left-handedness themselves are not. Further confirmation was obtained at the same time of another conclusion previously stated : viz. that " The same ratio subsists among the seedlings whether produced from the odd or the even rows of seed on the parent ear " (p. 505). In the present experiments the offspring were as follows : — TABLE III. Bows LH RH LHIRH Per cent. LH Odd ... 867 615 1-410 58-43 Even ... 2702 1942 1-391 58-19 Total 3569 2557 1-396 58-26 The totals obtained for all the varieties of two-rowed Barley studied up to the present are as follows: — 11,185 LH, 7980 RH. Ratio LH/RH = 1-4016. Percentage LH = 58-362. 1 Compton, 1910, p. 497. 2 Ibid. p. 499. R H. COMPTON 57 Six-Rowed Barley. A few sowings were made with Hordeum heccastichum, var. pyrami- datum. In this species all three flowers produced at each notch of the spike are hermaphrodite and set seed. Thus there are three odd and three even rows on each ear. The three rows in each case are called "left," "middle," and "right": that row being called "left" which is towards the left hand when the ear is held upright with the middle row towards the observer. Diagram 2, which is partly ground-plan, partly elevation, will make this convention clear. O Q Diagram 2. The three odd rows in several ears gave in all 55 LH and 45 RH seedlings: the ratio LHjRH being 122 (% LH=5o). Exactly the same ratio was given by the offspring of the three even rows; the numbers being 60 LH, and 49 RH. In Table IV, therefore, the two TABLE IV. Bows LH RH LH/RH Per cent LH Left 60 46 1-304 56-6 Middle 87 76 1145 53-4 Bight 71 48 1-479 69-7 170 1-282 56-98 Total 218 " left " rows — odd and even — are added together, and the same is done for the " middle " and " right " rows respectively. Thus, as in the two-rowed varieties, there is an excess of left- handed offspring from all positions on the spike. The numbers studied 68 Study of Right- and Left- Handedness are rather restricted, so that the ratios cannot be considered exact ; and the discrepancies from the average given by individual rows are well within the limits of probable error. We may conclude with a certain degree of safety that the position of the seed on the ear has no effect upon the method of folding of the first leaf of the seedling. Oats. The mature plant of many species of the Gramineae exhibits a more or less pronounced tendency on the part of the leaves to twist loosely into the form of a screws The direction of the screw is constant, so far as known, in the case of the common cereals, and it is sometimes used as a diagnostic character for distinguishing different crops in their early stages I The leaf-blades of Barley, Wheat and Rye are rolled to the right (i.e. like an ordinary right-handed screw), while Oats show leaf-blades rolled to the left. This twist is quite independent of the mode of folding of the leaves in the young state (and of the leaf-bases when mature); all the leaves of the same plant showing the same twist, although the direction of folding is normally reversed at successive nodes. We have seen that in Barley, where the leaves exhibit a right- handed torsion, there is a definite and constant excess of seedlings whose first leaf shows the mode of folding which I have called left- handed. The question thus arises whether Oats, in which the leaf-blades show the reverse or left-handed torsion, will give a different ratio between seedlings showing the LH and RH modes of folding of the first leaf. A sowing was made of a random sample of "Thousand Dollar" Oats, and the seedlings were counted for rights and lefts. There were found 469 LH and 576 RH. That is to say, there exists in Oats a consider- able excess of right-handed seedlings. The ratio LHjRH is 0*814 (44*88 7o LH) ; whereas in Barley (adding together all the varieties studied) it is 1*4016 (58*362 7„ LH) (above, p. 56). We thus find that, in the cases of Barley and Oats, the reversed torsion of the mature leaves is accompanied by a reversal of the ratio between left- and right-handed seedlings in respect of the fold of the first leaf What is the nature of this association, or whether it is merely fortuitous, it is impossible at present to decide. 1 Hackel, " Gramineae," in Engler and Prantl, Nat. Pflfam. n. 2, p. 4. * Percival, Agricultural Botany, p. 511. R H. CoMPTON 69 Italian Millet. The seedling of Setaria italica exhibits a long internode between the scutellum and the base of the plumular sheath (cf. Maize). A number of seedlings were counted, and there were found 258 with the first leaf folded in the left-handed, 217 in the right-handed fashion. Ratio LH I RE = 119 (5^1 °I^LH). A marked excess of left-handed seedlings occui-s here also, though not so great as in Barley. Rye. Secale cereale is an unfavourable plant for this purpose, owing to the narrowness of the leaves and the frequency with which both margins are curled inwards in the upper portion^ Out of 30 seedlings, 16 were LH and 14 RH : showing that both conditions occur here also, though the numbers are insuflBcient to allow a ratio to be calculated. Maize. As is well known, the infructescence or cob of Zea Mais is com- parable with an annular fasciation : its construction is such as might be obtained by the fusion in a cylinder of a number of pairs of rows of fruits *. The number of such double rows varies from 2 to 1 1 according to the variety and the strength of the plant. In every case it is easy to distinguish the rows of a single pair from the adjacent rows of con- tiguous pairs. There is a shallow furrow on the cob between adjoining double-rows which becomes evident if the cob be cut or broken across, though it is usually impossible to see it in the ordinary condition owing to distortion of the rows at the ends of the cob. It is said that the Maize cob invariably possesses an even number of rows of grain : this being of course the result of its construction from pairs of orthostichies. In speaking of the Maize cob I propose to distinguish odd and even rows of seeds, in the same way as in Barley : for it may be considered as in a sense equivalent to a fused ring of ears of two-rowed Barley*. Holding a Maize cob upright, and considering a single pair of rows of grain towards the observer, that row which is towards his left hand will be called odd, and that row even which is towards his right hand. 1 Cf. Compton, 1910, p. 496, first foot-note. ' Another case of an hereditary ring-fasciation is described in my paper on "The Anatomy of the Mummy Pea," New Phytologist, x. p. 249, 1911. * Hackel (loc. cit. p. 20) remarks that "die einzelnen Doppelzeilen je Einer Aehre Ton Eachlaena entsprechen. " 60 Study of Right- and Left- Handedness The accompanying ground-plan (Diagram 3) will make this convention clear. The first leaf of a Maize seedling is folded in the same way as in the other species of Gramineae here considered : both right- and left- handed seedlings occurring. The question whether there is any connection between the direction of folding and the position of the seed on the cob was answered in the affirmative by Macloskie^ who stated that " The grains arising on adjoining rows in the ear of corn are of different castes, and produce antidromic plants^": and again, "In the particular ear examined the grains of the dextral row were all with dextral embryos, and those of the sinistral row had sinistral embryosV In 1910 I sowed the seeds of rather an old cob of Maize, keeping the Diagram 8. rows separate : comparatively few of the seeds germinated, but these were quite sufficient to show that both left- and right-handed seedlings were produced from the same orthostichy. I announced my failure to confirm Macloskie's statement in my earlier paper (p. 503). Later, Professor Macloskie sent me a copy of a further paper ^ of whose existence I was unaware, in which he corrects his previous statements as follows: — "I now find that two-thirds of the grains in the row opposite my right hand have the left margin of their leaves external, and the other third have their right margin external, these proportions being reversed for the row opposite my left hand." 1 "Dimorphism and Rhizomycete of Maize," Princeton Coll. Bulletin, v. p. 84, Nov. 1893. 2 "Antidromy in Plants," Amer. Naturalist, xxix. p. 973, Oct. 1895. 3 "Antidromy of Plants," Bull. Torrey Bot. Club, xxii. p. 379, Sept. 1895. * "Antidromy in Plants," Princeton Coll. Bulletin, vii. p. 107, Nov. 1895. R H. COMPTON 61 It was considered desirable to pursue the enquiry further, as statistics on the subject had not been published, and as both Macloskie's statements and my own were based (as now appears) on insufficient data. With this object a number of Maize cobs of different varieties' were procured, and sowings were made of the seeds from each row separately'. The seedlings were counted for rights and lefts as in previous experiments, and the results are summarised in Table V. If we consider the total number of seedlings of Maize frota all sources, given in the last line of the table (columns 12 — 15), we find that right- and left-handed plants occur in almost exactly equal numbers: the ratio LHjRH for 6189 seedlings being 1008*. This is a striking difference from what was found in Barley, Oats, etc., where there is always a considerable excess of one kind, LH in Barley, RH in Oats. Not only is the ratio for the total number of Maize seedlings very near unity, but the offspring of individual cobs, as recorded in columns 12 — 15 of Table V, also show a reasonably close approximation to the same ratio. In some cases it may appear that the ratio diverges from unity so far as to be outside the limits of fluctuating variability : this is the case in cobs I, II, V, VI, X, and XIV. In cobs I and VI this is clearly explicable as the result of the smallness of the numbers of seedlings ; in cob X it appears to be the result of an accidental excess of seedlings from odd rows — an occurrence to be explained below. But in the other three aberrant cobs the ratios may be significant. If we now consider the results given by odd and even rows of seeds taken separately (cols. 4 — 11 in Table V) a marked discrepancy from this ratio of equality becomes evident. On the whole, as shown in the last line of the table, the offspring of the odd roivs contain an excess of individrmls with the first leaf folded in the iright-handed direction : the even rows produce a corresponding excess of left-handed seedlings*. In the case of 2966 seedlings from the odd rows of 16 ears of Maize 1 I am mnch indebted to Mr A. O. D. Mogg, of Caias College, Cambridge, for kindly naming many of the varieties used. - The irregularly arranged seeds from the " butts and tips" of the cobs were discarded. ' It shonld be observed that there was an excess of 85 seedlings from odd over thuse from even rows ; and that, for reasons to be discussed later, this renders the ratio 1-008 slightly too small : after applying the necessary correction the ratio was foand to be lOlO. * A somewhat fanciful comparison may be made with the Cyprinodont fish AnabUps anableps, in which according to S. Garman {Amer. Naturalist, p. 1012, 1895), the males are 3/5 RH, 2/5 LH ; while the females are 2/5 RH, 3/5 LH : the dimorphism being shown by the external genital organs. 62 Study of Right- and Left- Handedness §ta ?'?'?' T* ■?* T' / po «5 lO O 0> »« « fe •* «S O -^ >* "* Oh c ©» •-? O tH fH O M5 ><5 00 t~ «5 -* OS o o »fll CO 05 00 05 05 05 :s OS o us s? ^ OS 00 00 § 00 § 85 CO OS o o O ^ i-H «S CO OS 50 00 CO OS >-l O IM OS iH 00 CO ?H OS 05 I— ) i-H CO rH i-H iH « Cfl »H C,-|-^»OeO0OOSOSOS»OCOO5(NiX> as, ^ ^ §g W CO a0 OS 00 t- C0 U5 US «5 »0 US CO 00 O O CD 00 -H CO IM O "-H 00 CO >— I OS us OS tH us OS 00 OS c- ^ lO CO (M OS f .-^ O H •^ .9 0° -2 ^ "^ ^ H E- 00 00 00 00 00 o >_, HH B > !> HH 1-4 I— I l|l - = 3 & > P g g a ^ a « g e g g g R H. CoMPTON 68 there were 228 more RH than LH individuals: and among 2881 seedlings from the even rows there were 243 more LH than RH individuals. There can be no doubt that so large a divergence from the mean ratio unity is significant : especially as in the majority of the Maize cobs considered separately a similar result was obt-ained. Further, individual rows of seed gave the same excess of LH or RH ofispring according as they w^ere even or odd. There is no need to give the data for all the cobs ; but one set of results will be quoted in full as an example of the phenomena encountered. The numbers shown in Table VI were yielded by cob XIII, and may be taken as typical of those given by the others ^ TABT.K VI. o< Number of Offspring Nnmber of Bow Id Rows Even BowB LH RR LU RB 1 14 19 — — 2 — — 14 16 3 7 25 — — 4 — — 18 14 5 16 16 — — 6 — — 20 11 7 9 22 — — 8 — — 20 12 9 13 14 — — 10 — — 19 9 11 15 13 — — 12 — — 15 11 13 13 9 — — 14 — — 15 13 15 8 21 — — 16 — — 16 13 Total 95 139 137 99 The reference numbers of the rows in the first column are arbitrary : No. 1 was a casually selected odd row, No. 2 the even row of the same pair, and so on all round the cob in the direction of the arrow in Diagram 3. The results for even and odd rows are displayed in different columns for the sake of clearness. It will be seen that five of the eight odd rows gave an excess of RH seedlings, and that seven of ^ In this particalar cob the seeds of each row were sown in order of position, with the object of determining whether there was any farther regnlarity in the distribation of rights and lefts over its surface : no such pattern, however, ooold be detected. 64 Study of Right- and Left- Handedness the eight even rows gave an excess of LH seedlings. There are alto- gether three exceptional rows in which the proportion of rights and lefts are reversed, and one in which LH/RH = 1. Such exceptions occur in all the cobs studied, both in even and in odd rows : they are probably to be attributed to fluctuating variability of the same kind as was found in the ratios for ears of two-rowed Barley, and represented by normal curves \ It must be noticed that exceptions to the general rule are not lacking in individual cobs, as will be seen on reference to Table V. Cob VII shows a ratio of very near equality for both odd and even rows. Cob XIV shows a ratio of about 1'2, not only for the even rows (where it is normal), but for the odd rows as well : a result which may be compared with that obtained in Barley. Cob XI is the most striking exception, for here the odd rows gave a definite excess of lefts and the even rows a similar excess of rights : this being the exact reverse of the usual results. Another cob, doubtfully of the same variety as cob XI — viz. cob XV — also gave somewhat abnormal results, so that it is possible that different varieties of Maize may behave in different ways^. No further cobs of the same varieties as VII and XIV could be procured. But despite exceptions, both in the offspring of individual rows and of single cobs, the general conclusion appears to be justified that odd rows on the Maize cob give an excess of right-handed, and even rows an excess of left-handed, offspring. It remains to find a reason for this behaviour. It was thought possible that the position of the ovule with respect to its neighbours, and the consequent differences in pressure which would be experienced by ovules according as they were produced in odd or even rows, might cause differences in the shape of the early environment of the embryo which would be to some extent reflected in the mode of folding of its first leaf. This hypothesis was tested by the following experiment. A cob (XVII) was chosen whose rows were considerably distorted, and whose seeds consequently showed much variety of shape. The seeds were divided into three lots according to the relative thickness of the two lateral edges : looking at the outer end of the seed with the embryo uppermost it was placed in class (a) if the LH edge was narrower than the RH, class (c) if the reverse was the case, or class (b) if it could not be definitely included among (a) 1 Compton, 1910, p. 501. 2 Further experiments are being made with this variety of Maize. R H. CoMPTON 65 or (c). (See Diagram 4 in which one is supposed to be looking at the distal end of the seed in situ, the shaded semilunar area repre- senting the position of the embryo.) The three classes were sown separately and the ofi&pring counted, the results being as follows: — TABLE vn. Offspring CUaB LH RH a 38 87 b 40 39 e 33 33 Total 111 109 Thus the three classes of seeds gave precisely similar results, the ratio being equality in each case — i.e. the same ratio as given by the whole number of Maize seeds studied. O c^ O ^ i c Diagram 4. Though this experiment failed to give a positive result, it cannot be considered as conclusive evidence against the theory it was designed to test. It seems probable a priori that a variation in symmetry of the seedling should be produced by asymmetry of the space within which the embryo develops, and that this asymmetry should be produced by the pressure of neighbouring seeds: in fact that asymmetry in the parent should directly produce asymmetry in the offspring-^ Several examples are indeed given by Macloskie^ showing that the seeds from opposite sides of a bilateral fruit (silicle or legume) are " antidromic " — a word which he uses very broadly to cover many diflferent kinds of asymmetry. ^ An essentially similar hypothesis has been suggested by van Biervliet to accoant for the apparent inheritance of homan right- and left-handedness (see footnote to p. 68). He remarks, " Nous savons que la femme droiti^re a la hanche droite plus d^velopp^, la gaach^re la hanche plus forte k gauche. La structure du bassin ne pourrait-elle influer sur la position de I'embryon ? ou da moins favoriser le d^veloppement preponderant du cdt^ gauche?" ("L'homme droite et I'homme gauche," JUvw philmophique de la France et de VEtranger, XLvn. p. 385, 1899). The existence of left-handedness among the children of a LH father and a RH mother, as recorded by Jordan, tends to cast doubt on this hypothesis but does not destroy it altogether. The collection of a greater number of pedigrees is the most obvious method of solving the problem. » Bull. Torrey Bot. Club, xra. p. 379, 1895. Jonm. of Oen. u 6 66 Study of Right- arid Left-Haiidedness An alternative explanation would be possible on the basis of a hypothesis of a somatic segregation of characters^ if it be assumed (i) that the characters of right- and left-handedness are represented in the gametes, and (ii) that odd rows tend to produce an excess of female gametes bearing right-handedness, even rows an excess of those bearing the character for left-handedness. (The male gametes may be left out of account when considering large numbers, for pollination is anemo- philous and promiscuous.) The alternative explanations are (i) that the influence determining the fold of the first leaf is of a feeble nature and acts upon the gamete or embryo, which itself has no special inclination to become right- or left-handed, but must choose one or the other : or (ii) that the female gametes are definitely right- or left-handed, and that a certain amount of somatic segregation occurs. In order to maintain this latter alternative explanation it would be necessary to show that right- and left-handedness in Maize are in some way hereditary characters. The statistics give strong evidence that this is not the case. Almost all the 17 Maize cobs studied gave ratios between lefts and rights very nearly alike (p. 62) : but it is almost inconceivable that in every case the parent plant should have had the same direction of fold in its first leaf. It is evident that in different cereals there are tendencies to produce different ratios ; Barley for instance gave a very different result from Oats : and it is possible that the inverse ratios given by Maize cob XI point to some similar divergencies between different varieties of Maize. But the fact that different plants give different ratios and that (in the case of Barley at least) the ratio is constant for successive generations does not imply that the characters of right- and left-handedness them- selves are hereditary. We may therefore conclude that there is no evidence for the inheritance of the direction of fold in the first leaf of Maize : and this conclusion harmonises with the similar one previously reached in the case of Barley. A further argument against the second alternative explanation is the fact that, though much sought, up to the present no direct proof of somatic segregation affecting gametes of the same sex has been obtained, however probable the existence of such a phenomenon may appear. 1 See W. Bateson and E. C. Punnett, "On the Inter-relations of Genetic Factors." Proc. Roy. Soc. B, Vol. lxxxiv. p. 6, 1911. " On Gametic Series involving Reduplication of certain Terms," Verh. d. naturf. Ver. inBrilnn, Bd. xlix. 1911 and Journ. Genet. Vol. 1. 1911. R H. CoMPTON 67 It seems therefore almost impossible to maintain the hypothesis of somatic segregation of right- and left-handedness in Maize : and despite the negative result of the single experiment on the subject, I incline to embrace the first explanation proposed, viz., that the difference in offspring between odd and even rows is due to the direct influence of spatial relationships on the developing embryo. In two-rowed Barley no such dependence upon position was found : it may be suggested that the difference in the case of Maize is due to the close packing together of adjacent rows of seeds ; for in two-rowed Barley the individual seeds develop without lateral pressure from neighbouring rows such as they experience in Maize. Genetic Spirals. It is not proposed at present to enter fully into the question of genetic spirals from the point of view of ratios and heredity : experi- ments are in progress, and it is hoped to publish a paper on this subject at some future date. Meanwhile it is interesting to remark that the ratio LHjRH, for genetic spirals also, diverges from equality in a more or less marked degree in the cases investigated. Bonnet* found 43 RH and 30 LH stems among a collection of 73 plants of Chicory. Out of 458 plants oi Lepidium sativum I found 241 LH and 217 RH. Not only is there a divergence from equality in the case of different plants, but on one and the same plant there may be a considerable excess of branches showing one spiral over those showing the reverse one. Valuable data were obtained- by Dr A. H. Church and Mr K G. Broome with respect to the genetic spiral of certain individuals of difierent species of Pines : the following numbers were recorded : — Tree axutriaca Average P. fumilio P. larieio TABLE VIII. Year Number of cones Percent LH Percent RE 1900 100 41 59 1901 1000 46-4 53-6 1902 500 46-0 54-0 1900—02 1600 45-94 54-06 1900 100 47 53 1900 100 68 38 1901 400 71 S9 1902 600 69 16 30-83 1900—02 1100 69-82 30-18 ,, Average ^ Reeherches sur I'tuage des feuillet, p. 179, 1754. 2 A. H. Church, The Relation of PhyUotaxU to Mechanical Lawi, pp. 92, 351, London, 1904. 6—2 68 Study of Right- and Left-Handedness These results are very striking, and as Church remarks, " The element of chance appears quite out of the question" in the case of P. laricio. It may be suggested that perhaps differences in conditions (sunlight, prevailing winds, etc.) between opposite sides of the tree may have caused an excess of cones to arise towards one side : and that, since the genetic spiral of a lateral branch depends on its mode of insertion, this might be the cause of the excess of one kind of spiral among the cones. At present, however, it is impossible to decide whether this accounts for the phenomena, or whether there is a definite tendency in the apical meristem to produce a genetic spiral of one direction rather than the other. Conclusions. The present paper deals with the dimorphism found in certain Gramineae in respect of the mode of folding of the first leaf of the young plant. It is in part a continuation of a previous paper (cited above) to which reference should be made for an explanation of the conventions used in describing the phenomena, and for a summary of the literature of the genetics of right- and left-handedness in general/, ^ Further references to works on the genetics of right- and left-handedness may be inserted here. H. E. Jordan ("The Inheritance of Left-Handedness," American Breeders^ Magazine, Vol. II. pp. 19, 113 : 1911) gives a number of human pedigrees which show that functional left-handedness is hereditary, in certain cases apparently in conformity with a simple Mendelian scheme. The most remarkable human pedigree on record is perhaps that given by Aim6 P6r6 (Les courbures laterales normales du rachis humain, Toulouse, 1900, p. 71 : quoted by D. J. Cunningham, "Eight-Handedness and Left-Brainedness," Journ. Anthropol. Inst, xxxii. p. 273, 1902). In this family no fewer than twenty-six left-handed individuals are recorded : the marriage of a LH ? x RH i gave eight sons and six daughters, all left-handed, — a fact in strong contrast to Jordan's hypothesis of the dominance of right-handedness, and suggesting the reverse assumption. A number of other instances of inheritance of left-handedness in man are given by F. Lueddeckens (Rechts- und Links- hdndigkeit, Leipzic, 1900). A general summary, with a full bibliography, of our present knowledge of human asymmetry is given by K. von Bardeleben ("Ueber bilaterale Asym- metrie beim Menschen und bei hoheren Tieren, " Anat. Anz. Ergdnzungsh. z. Bd. xxxiv. p. 2, 1909). H. de Vries, The Mutation Theory (Engl, trans.), ii. p. 561, London, 1911, finds that the peculiar torsion of the stem in certain races of Dipsacus sylvestris is partly hereditary, but that the direction of the twist is not transmitted. About equal numbers of RH&nd. LH individuals are produced normally : and in an experiment in which only RH plants were allowed to flower in two successive years the offspring comprised 245 RH and 239 LH plants. R H. COMPTON 09 In this paper it was shown that in two- rowed Barley there is no inheritance of the right- and left-handed characters, nor is there any regularity in the distribution of right- and left-handed seedlings on the ear. (1) The previous conclusions in the case of two-rowed Barley are confirmed by new experimental results : in particular it is amply shown that, while the ratio of lefts to rights is maintained through three successive generations, the kind of asymmetry itself is not inherited. (2) In Maize it also seems clear that there is no inheritance of right- and left-handedness as such. (3) In six-rowed Barley an excess of LH seedlings was also found : and there was no conspicuous variation in the ratio LHjRH as between different rows of grain on the ear : but the numbers examined were too small to be decisive. (4) In Maize the ratio LH/RH is very near unity (I'OIO). Setaria italica shows, like Barley, an excess of left-handed seedlings, LH/RH being 1*19 (54*1 °/^ LH). Both stereo-isomeric forms are also present in Rye. (5) Oats show a considerable excess of right-handed seedlings, the ratio LH/RH being 0814 (44-88% LH): it thus gives a result the inverse of that found for Barley. This may be in some way connected with the fact that in Barley the mature leaf-blades are generally slightly twisted into a right-handed screw, while in Oats the torsion is in the reverse direction. (6) The seeds on a Maize-cob give different ratios of left-handed seedlings according to their position : the seeds on odd orthostichies giving an excess of right-handed, those on even rows an excess of left- handed, offspring. This result is obtained in the majority of cobs, but there are exceptions. On the average, deduced from 5847 seedlings from 16 cobs of several varieties and sizes, the ratio LH/RH for even rows is 1"184, for odd rows 0*857 (54"22 and 4616 % ^^ r®* spectively). (7) An attempt to discover whether there was any connection between the shape of the seed of Maize and the direction of fold of the first leaf gave a negative result. 70 SUidy of Right- and Left- Handedness (8) Nevertheless, it appears probable that the diverse ratios given by odd and even rows in Maize result from a difference in shape of the environment of the developing embryo, this being connected with the different spatial relationship of the seeds on the cob. An alternative hypothesis involving somatic segi-egation of symmetry characters repre- sented in the gametes is dismissed as highly improbable. Botany School, CAMBRroGB, November 1911. Volume II JUNE, 1912 Na 2 SPECIES HYBRIDS OF DIGITALIS. By W. NEILSON JONES, M.A., University Collegey Beading. Historical. Hybrids between the various species of Digitalis have been recorded from time to time and in many cases can be produced without difficulty or even occur in nature. A general summary of the literature up to 1881 is given by Focke in Die Pjiamen-Mischlinge. The hybrids most frequently recorded are those between D. purpurea and D. luiea, and between D. purpurea and D. grandijiora^, the accounts being somewhat contradictory. The cross between D. purpurea and D. lutea has been re-investigated by J. H. Wilson and described in the Report of the 3ixl International Conference on Genetics (1906). Briefly, the fects concerning the hybridising of these two species are as follows: It was found much easier to eflEect the cross when D. purpurea was used as pollen parent. The reciprocal crosses diflFered from one another as to their flowers, in each case more closely resembling the seed-parent. The hybrid having D. purpurea as seed-parent had larger and wider flowers of a rose colour although the D. purpurea used was a white flowered variety without coloured spots (from which it was concluded, incidentally, that colour may be latent in a white foxglove). In the cross in which D. lutea was used as seed-parent the flowers were narrower and in colour creamy-yellow or almost white with a pale rose flush, even when a purple D. purpurea was used. The flowers of both hybrids had purple spots inside the corolla tube. In a subsequent series of experiments Wilson found that the Fi plants of both hybrids varied considerably among themselves as to flower colour. This was possibly due to the use of an impure strain. The reciprocals were indistinguishable until they flowered. No fertile seed was obtained from either hybrid. * D. grandiftora L. is synonymous with D. -^ o8 H -»3 ^ £3 c8 S r^^ ■1 X -H p nn n n D D nn n nn n a a — a a a an a a a .. s [HI 1 ■ 1 ■ s e8 1^ 3 ^ ^ ^ Hs i-» -.ri^f^uC,-;>ej o I— I C5 tH t- a be S O a PQ II n t3 c 03 P5 Cm O >. -a 3 c8 I? nn BD Bnnnn nn D C4 M ^ □ a 9SS H 1^ O Q © 00 o Ci © 3 3 © 02 © O) o " © » © be "S, ^ J2 '^ 3 © <^ S i-s © O) o ^ .9 ? 3 a, ^ « a- a Joom. of Gen. n 120 On Heterochromia Iridis in Man and Animals The parti-coloured iris occurs during inter-breeding fairly frequently, but the parti-coloured skin must be extremely rare in the human species, whereas in domesticated animals the parti-coloured or " wall " eye and the pied coat pattern both occur when self colour is mated with white. This fact strongly suggests that the factors for self colour in the negro or the factor for self white in the European, or both of them, offer more resistance to disintegration than the factors which control the self-duplex and the self-simplex patterns in the human eye. Some association no doubt does exist between skin-colour and eye- colour even in the human subject; the dark brown or black iris and the black skin of the negro, the blue eye, reddish hair and white skin of the northern races illustrate this point. The same is certainly true in many wild species and domestic varieties of animals, and even in extracted hybrid varieties we have the association of the agouti and blue coat colour with black eyes in mice, and chocolate eye pigment with pink eyes in the Himalayan rabbit, the Siamese cat, and the Cinnamon canary (Bateson, p. 114) (9). But owing possibly to the difficulty of breaking up the self-skin colour factor in man and the greater susceptibility of the self-colour iris factor to disintegration, the association which originally existed between these factors is partially dissolved when they are called upon to behave independently under the altered conditions of inter-breeding. Thus in the native Irish race, dark complexion and dark hair have become associated with the violet or simplex eye. It is a suggestive fact that the factor for self skin colour in man which refuses to dis- integrate when mated with absence of colour or white skin should apparently refuse also to segregate in gametogenesis. Some anomalous facts about albinism in the human subject are also of interest in this connection. A woman M. T. aet. 35, the daughter of a brown-haired, blue-eyed father, and a red-haired, blue-eyed mother, and one of four children (two daughters with red hair and blue eyes, and one son with dark brown hair and blue eyes) was herself born with white pigmentless hair, eye- brows, and eyelashes and pink pupils and pigmentless irides, in fact with all the characters of the true albino. At the age of puberty the hair of the head of this woman gradually assumed a red tinge and is now the ordinary fiery red colour, while the eyebrows and eye structures have remained free from pigment. In this case the association which existed in childhood between the recessive characters, absence of skin and hair pigment, and absence of C. J. Bond 121 iris and choroid pigment, has been broken up by the appearance in later years of a dominant character, e.g. hair pigment in the hair of the scalp though not in the hair of the face. Bateson also refers to the existence of similar abnormal cases of albinism in man, pp. 226-7. See also Pearson, Nettleship and Usher, On Albinism in Man. In one sense this unusual development of a later epistatic character in an albino is only what happens normally in the case of other characters in the young of animals and of human beings. All children, with few exceptions, are bom with simplex or blue eyes, the epistatic duplex character only begins to appear from 4 to 8 or 10 weeks after birth. The same is true of kittens and the young of other animals, while in young pigeons hatched with " bull " eyes the red or yellow anterior pigment begins to be deposited 6 or 8 weeks after hatching. In fact following the law of recapitulative ontological development with the addition of characters, epistatic characters seem usually to follow and not precede recessive characters in the process of the unfolding of unit characters in the zygote. Hair Pattern in Man. But although as we have seen the self-skin colour seems in the case of man at present to resist influences of a genetic kind calculated to break it up into subordinate factors (a state of things tvhich may account for the fact that, when in the human species irregular or parti- coloured iris pigmentation occurs a like particulate distribution of skin colour pattern does not usually occur along with it, at any rate in European races), yet there are some facts which suggest that the factors which control hair structure and its distribution in the human subject are subject to similar disturbances to those which affect the factor for iris pigmentation. As a general rule the corkscrew (ellipsoidal sectioned) black hair of the negro is dominant over the wavy (oval sectioned) brown hair of the European. And in the majority of the Fi hybrids of the negro and white cross this dominance extends over the whole of the scalp area to all the hair of the head. (Plate IX, fig. 2.) In two children L. M. and B. M. out of a family of nine the Fi offspring of an English woman with straight brown hair and a West African negro (Plate VIII) some segregation of the corkscrew and 9—3 122 On Heterochromia Iridis in Man and Animals straight hair pattern occurred. The curly tufted negro hair covered the sides of the scalp while the straighter hair formed a patch on the vertex in each case. (Plate IX, fig. 1.) In these two children genetic factors which normally control the epistatic negroid type of hair structure over the whole scalp behaved differently in different scalp areas, and the true explanation of irregular or ray iris pattern and of harlequin coat colour pattern will probably apply to this abnormality in hair structure pattern also. PART II. Irregular Eye Colour Pattern and the Constitution of Gametic Factors. Irregular iris pigmentation of the ray type and indivisibility of gametic factors in heredity. It has been customary to regard gametic factors as indivisible units like the chemist's atom. Thus Punnett(lO) (Mendelism, p. 39) says, " Unit characters (in the zygote) are represented by factors in the gametes which behave in the process of heredity as indivisible entities." Now if in the " B " human family we regard the male parent G. B. as a heterozygous duplex of irregular or ray pattern, then, according to expectation, the mating of this individual with a simplex recessive should result in equal number of simplex and duplex offspring in the* Fi generation, whereas the actual numbers are seven simplex and three duplex of which two are of the irregular ray pattern like the father in a family of ten. But a further difficulty arises with the manner of transmission of the irregular or ray pattern itself. For although two of the three duplex children are partial or ray duplex like the father, the distribution of the irregular pattern is not the same in the parent and the children. The pigmented ray appears in the right eye instead of the left (as in the father) in one child, and in the opposite half of the same iris in the other child. How can we explain this different incidence of pattern in the passage from father to children ? Either the complication arises in some alteration in the composition or arrangement of the duplex factors in the gametes of the male parent, C. J. Bond 123 or it is brought about by some modifying factor introduced by the gametes of the recessive female parent. It will be best to leave the final conclusion till all the evidence has been considered. There are certain facts about these irregular forms of eye colour in man and pied coat colour in animals which suggest that presence and ahsence of gametic factors though essential are not the only conditions which determine the behaviour of unit characters. These facts suggest that much depends on the manner in which any factor is present in the gamete which carries it, and on the way in which it incorporates itself with, or is incorporated by, the gamete which bears the alternative fector during the process of gametic union. We already know that much depends on the volume of the factor present. The difference in appearance between the homo- and heterozygous dominant depends on doubleness or singleness of dose of the dominant character. In this connection I should also mention Davenport's (11) remarks on the imperfection of dominance in which he concludes that alongside of dominance we must place an important modifying factor, the factor of the strength or potency of the representative of the given character in the germ plasm. The influence of reciprocal gametic contribution is well known in the case of sex controlled characters. In other cases the appearance and behaviour of unit characters have been shown to depend on the intermediation of a second and in some cases of a third ftictor, introduced by the gamete which carries the recessive character. Thus black eye colour in certain varieties of mice is attributed to the interaction of two factors (Bateson, p. 112)(9). The limitation of colour to certain skin areas in pied individuals, in which pied pattern is dominant, has been explained by the restraining influence of the factor for pied pattern on the &ctor for self colour (Bateson, p. 84)(9). The investigation of irregular types of eye colour pattern throws some light on the relationship between these different factors for eye colour especially in the human subject, because in man the close association which exists in the black and white races between eye colour and skin colour has been partly dissolved in the European or mixed races, and one complicating element has been thereby removed. 124 On Heterochromia Iridis in Man and Animals It seems clear, from the familial distribution of these cases of irregular eye colouration, that the ray eye pattern in man, like the Dutch coat pattern in the rabbit, is controlled by a definite gametic factor or factors. The duplex ray pattern like the duplex self-colour and ringed patterns seems to be dominant over the simplex character (see Baines, 2nd family). It would also appear to be dominant over self-colour duplex when the self-colour duplex is of lighter shade than the ray duplex (see Frake), but this dominance only occurred in two out of nine children and then only in one eye, the other being self-colour duplex. In two others the self pattern was associated with irregular patches of darker pigment as we shall see later. These facts suggest that ray pattern is only dominant over self-colour pattern when the original factor for self-colour has undergone some dilution or when disintegration and dilution are operating together. Concerning this difference in genetic composition between self-colour and Dutch pattern, Bateson says (p. 142)(9): " Physiologically we should I suppose refer it to differences in the distribution of one of the chromogenic factors, rather than to the presence or absence of an additional factor." It is these differences in the distribution of one or both of the chromogenic factors that I wish to consider in the light of the evidence derived from cases of irregular iris pigmentation in man and animals. It is true that there are albinos of different factorial composition in every species, just as there are dominant and recessive " whites " among fowls. There are some extracted albinos which carry the factor for self- colour, and others which carry the factor for pattern, though both lack the colour developer factor itself. The different results obtained in mating the black fantail cock with the two white fantail hens of different gametic composition afford another example of the influence of the gamete which bears the recessive character on the determination of colour pattern. But while by the assumption of a colour factor awaiting development in the albino and a developer factor in the self-coloured animal we can by the mutual interaction of these factors explain the restriction of the colour unit character to certain skin areas, and the dilution of colour in zygotes of different gametic composition, we have not thereby explained the original distribution of these different factors in either the albino or the pied animal. Thus it appears that besides presence and absence, and besides C. J. Bond 125 singleness and doubleness of dose, and besides reciprocal gametic introduction, and besides intermediary action of restraining factors, and besides colour basis and colour developer factors, there still remains some crametic cause for the abnormal behaviour of the unit characters in the individuals with irregular eye colour pattern. Now it seems possible that this unexplained irregularity of pattern may depend on some deficiency in integration of the dominant duplex factor, some ab- normality in the way in which it is present in the gamete which carries it, or in the way it is incorporated by the gamete which carries the recessive or the alternative character. Of the characters possessed by gametic factors two of the most important are total VOLUME, that is, size or quantitative presence, and DEGREE OF INTEGRATION, that is, qualitative presence. The reduction of the total volume of gametic factors occurs probably during gameto-genesis and the important point is that this reduction of volume may apparently take place with or without a concomitant process of partial disintegration of the factor concerned into its subordinate units. That is to say reduction of volume may be quanti- tative or qualitative. Thus the factor for self-eye colour has, in ordinary individuals, a definite total volume ; it possesses a definite capacity to spread epistatically over the whole anterior surface of the iris and so to produce the self-coloured pattern in the duplex eye. In certain less numerous cases (14 out of 88) the volume of this factor is reduced and its epistatic influence is restricted to a circular zone of iris round the pupil, and the ring pattern appears ; in other individuals it is limited to certain patches on the iris and produces the SPOTTED PATTERN ; in Others again (3 cases in 200) it only operates over certain sectors of the iris and then forms the RAY pattern. In all these cases the reduction in volume occurs in association with a qualitative and disintegrative change, a change in which the original factor for self-colour undergoes subdivision into subordinate factors independently controlling different areas (either rings, or patches, or rays) of the same iris. In other cases again there is no subdivision or disintegration and the reduction in volume (or capacity of influence) of the factor for self- colour takes the form of quantitative dilution, of deficient pigment saturation over the whole of the anterior surface of the iris, and in this way the various tints of the duplex pattern, from yellow to dark brown, are brought about. 126 On Heterochromia Iridis in Man and Animals If we imagine the factor for self-eye colour to have been built up during animal evolution out of two primary factors, one for each eye, and each of these again out of a number of secondary factors controlling the deposit of pigment in different sectors of each iris, and that all these factors have become welded together into one whole, and now act as one integrated factor, then we shall associate the appearance of the irregular or ray pattern with some preceding disintegrative change in the factor for self-eye colour in the human species. Some disintegration, partial or complete, is the first step in the origin of a new colour pattern. ' It provides the opportunity for the re-arrangement of subordinate factors which takes place during gameto- genesis and gametic union. From this standpoint we should regard the appearance of pied coat colour and pied eye colour, or " wall " eye, in the Orkney rabbits, and the appearance of the heterochromic eyes in the blue and white tumbler fantail hybrid of the third generation, as the outcome of a disintegration of the factors for self-coat colour or self-eye colour and this disintegration appears during the mating of individuals of different genetic com- position. The establishment of these new types on a basis of genetic stability (such as we see in the ring and ray iris pattern in man, and the Dutch coat pattern in rabbits) becomes then a matter of the re-welding of these secondary factors into one newly integrated factor, of a fresh type, and with a different arrangement of its component parts. Thus regarded the only way to obtain the piebald variety in the human species is to disintegrate the factor for self-colour which at present controls, in different degrees of dilution, the whole skin area, and which at present ordinarily refuses to segregate in gameto-genesis. The way to bring about irregular iris colour pattern is to break down (as we have done in the heterochromic pigeon) the self-colour factor for both eyes into two factors controlling its different eyes, and these again into subordinate factors controlling different areas in the same iris. The inter-mating of dark duplex with light duplex and duplex with simplex types results in the appearance of partial or complete hetero- chromia in some cases in the human species, and the familial distribution of these cases of irregular iris pigmentation suggests the possibility of the establishment under certain conditions of a heterochromic variety of the human race. Looked at from this point of view, the difference between individual C. J. Bond 127 and racial, and between varietal and specific characters would seem to be a matter of integration of the gametic factors concerned in each case, dominance and integration being closely associated. Thus the difference between the human mulatto hybrid and the spotted negro hybrid and the pied animal hybrid is this. In the mulatto the reduction of the volume of the colour factor is brought about by a quantitative process of equal dilution over the whole skin area, and not by a qualitative process of disintegration and segregation of component factors for different skin areas as in the spotted negro or the pied animal hybrid. By assuming the occurrence of a process of disintegration we extend the principle of segregation into the constitution of gametic factors, we assume intra as well as inter factorial segregation. If inter factorial segregation can explain the behaviour of unit characters on Mendelian lines in the normal heterozygote, then intra factorial segregation can explain the irregular behaviour of unit characters in the abnormal heterozygote. If the foregoing conception of gametic architecture be at all true, then it must stand the test of experience. The establishment of the Dutch pattern of coat colour in the rabbit should show some evidence of the welding process by which two or more less integrated unit characters (or rather the subordinate factors which control them) have been integrated into one factor for Dutch pattern. In the same way the resolution of the Dutch pattern into its component unit characters should show, as it does in the case of the Orkney rabbits (p. 112, Part I), the various steps in the disintegrative process. The same process can be seen at work in the genesis of the hetero- chromic pigeon. If true, this disintegrative theory should be applicable also to the problem of the comparative sterility of inter-special and the com- parative fertility of inter-varietal hybrids. Moreover it is quite independent of any pre-conceived notion as to the ultimate nature of genetic factors. It is equally applicable to the theory of gametic constitution which rests on an architectural or mechanical as to one which rests on a chemical basis in the organisation of the germ plasm. Since writing this paper my attention has been called to H. H. Laughlin's(12) observations on the inheritance of colour in 128 On Heterochromia Iridis in Man and Animals shorthorn cattle, in which he puts forward a chemical theory of " Ihtra- zygotic inhibition and reaction in response to specific set (? genetic) conditions," p. 27, and alludes to the possibility that " unit characters may arise from a partial destruction of larger units, and that a determiner for a unit character behaving precisely in unit fashion may be a 'complex' capable of being shattered into a large number of inde- pendently behaving characters," p. 26. BIBLIOGRAPHY. (1) J. Hutchinson. Ophtkxdmic Hospital Reports. London, 1869. (2) Marcos Gunn. The Ophthulmic B^view. London, 1889. (2a) Sym. The Ophthalmic Review. London, 1889. (26) Thorpe. Brit. Med. Journal, Vol. ii. 1906. (2c) Allen. Brit. Med. Journal, "Vol. i. 1906. (2cO Ross. Brit. Med. Journal, Vol. i, 1906. (3) Malgat, Recueil d^Ophthalmologie. 1895. (4) FucHS. Zeitschrift fur Augenheilkunde. 1906. iii. (5) Anton Lutz. Inaugural Dissertation. Berlin, 1908. (6) Hurst, C. Proceedings Royal Society, B. Vol. viii. 1908. On the Inheritance of Eye Colour in Man. (7) Przibram, Hans. Archiv fur Entuncklungs Mechanik der Organiamen. 1907. (8) Pearson, Nettleship, and Usher. A Monograph on Albinism in Man. 1911. (9) Bateson. MendeVs Principles of Heredity. 1909. (10) Punnett. Mendelism. 1911. (11) Davenport. "The Imperfection of Dominance." American Breeder^ Maga- zine, Vol. L No. 1, p. 42. (12) H. H. Laughlin. The Inheritance of Colour in Shorthorn Cattle. American Naturalist, December, 1911. American Naturalist, January 1912. (13) A. R. Gunn. Article on " Albino," Ency. Brit. 11th Ed. (Mudge.) JOURNAL OF GENETICS, VOL II. NO. 2 PUTE VI Fig. I. Fig. 2. JOURNAL OF GENETICS, VOL II. NO. 2 PLATE VII Fig. I. Fig. 2. Fig. 3- JOURNAL OF GENETICS, VOL II. NO. 2 PLATE VIII Fig. I. JOURNAL OF GENETICS, VOL II. NO. 2 PLATE IX Fig. I. Fig. 2 C. J. Bond 129 EXPLANATION OF PLATES. PLATE VI. "Wall eye" in an Orkney rabbit of Dutch colour pattern. Note absence of anterior pigment in the upper half of the left iris and the apper and anterior portions of the right iris. Fig. 1. Left eye. Fig. 2. Right eye. PLATE VM. Heterochromic Pigeon. Fig. 1, Left bull eye. Fig. 2. Bight orange eye with a gap in the orange pigment (colour coloboma) in the lower half of the iris. Fig. 3. Adult frog shows gap in anterior pigment of iris (colour coloboma) in downward vertical direction in the lower half of the iris. PLATE Vm. • Fig. 1. Nine children, the ofifspring of an English woman (wavy hair) and a West African Negro (corkscrew hair). The two youngest (marked x ) show differentiation of hair pattern with wavy hair on the vertex. PLATE IX. Fig. 1. L. M. Boy aet. 4, shows wavy European hair pattern on vertex, corkscrew pattern on sides. Fig. 2. Girl aet. 6, shows corkscrew negro hair pattern all over scalp. SECOND REPORT ON THE INHERITANCE OF COLOUR IN PIGEONS. TOGETHER WITH AN ACCOUNT OF SOME EXPERIMENTS ON THE CROSSING OF CERTAIN RACES OF DOVES. WITH SPECIAL REFERENCE TO SEX-LIMITED INHERITANCE. By RICHARD STAPLES-BROWNE, M.A. CONTENTS. PAGE Introduction .......... 131 Brief statement of resalts 132 Account of the experiments ....... 136 Bock Doves used in experiments 136 Bock Doves x White Domestic Pigeons 139 White Pigeons used in experiments 141 Types of birds produced 141 Details of the matings 144 Carriers, Dragoons, Owls and Fantails used in the experiments . 155 Silver Owls x Black Fantails 156 Dun Carrier ? x Blue Dragoon s 167 Crosses between dark and white races of doves . . . 158 Introduction. In the Proceedings of the Zoological Society for 1908, p. 67, I published a report on the inheritance of colour in Domestic Pigeons with special reference to reversion. The colours chiefly dealt with there were black and the "reversionary blue" of Darwin, which may be considered as the type classed as " black -chequer " by the fanciers (Plate V of the report), and also the relations of these two forms to white. 132 On the Inheritance of Colour in Pigeons In the present paper the colours considered are black, dun, blue and silver, their relations to each other, and the behaviour of the two latter when mated to white. The experiments on these colours are still in full progress, so that this report must be considered merely as a preliminary account of them. Enough has, however, been done to demonstrate a sex-limited inheritance in silver. The same may possibly be shown in dun also on further experiments It is found that, whereas a black f^ mated to a silver $ gives all the offspring black, yet a silver ^^ mated to a black $ gives black (/s and dun $ s. This result is similar to that obtained from the reciprocal matings of Cinnamon and Green Canaries. Further evidence on the sex-limited inheritance of silver in pigeons is obtained from the experiment, alluded to below, on the matings of Silvers and Reds. The possibility of a sex-limited inheritance of white in pigeons was suggested in my previous paper (p, 85) to account for the great excess of whites produced from the matings of heterozygous reversionary blues with whites in Exps. 16 — 23. This was particularly marked when the white was (/•, and Prof. Whitman's experiments with Doves were instanced in support of this proposition. The present series of crosses gives very little further evidence as regards white, but reference to experiment 58 of the present report would suggest that it is not sex-limited in these pigeons. I have recently made a series of experiments on the crossing of coloured and white doves. These matings confirm Prof Whitman's results, and an account of them is included in the present report. Specimens of the Turtle Dove {T. turtur) and the Barbary Cage Dove {T. risorius var. domesticus) have been mated respectively with the Java Dove, a white variety of the latter. These crosses show con- clusively that the phenomenon of sex-limited inheritance of white does occur in these species. Brief Statement of Results. In the present report three sets of experiments with pigeons are included, 1. Matings of blue and silver Rock Doves to each other and to whites. The whites used being Fantails and whites extracted from the Barb- Fantail crosses already described. * The evidence from this year's matings, as far as it goes, supports the suggestion that the inheritance of dun is also sex-limited. (See p. 156.) R Staples-Browne 133 2. Matings of Silver Owl and Black Fantail pigeons. 3. Matings of Blue Dragoon and Dun Carrier pigeons. In the case of the Rock Dove crosses the sex-limited inheritance of silver was not discovered, as in all cases the matings made were those of silver $ x blue (/. In this case the oflFspring were all blue according to expectation. The reciprocal mating of blue ? x silver ^ is being tried this year. To test fully this question the following matings are necessary : 1. dun $ X black <^. 2. black ? X dun «/•. 3. silver ? x blue ^. * 4. blue $ x silver » >» 51, 54, 55, 65, 66, 74 27 blue, 9 white 28 B., 8 W. »» »» 60 6-6 B., 1-9 S., 2-5 W. 6 B., 2 S., 2 W, Silver X Silver 61, 68, 69 All silver 31 silver Blue X Silver 48, 71 AU blue 12 blue „ ,, 69, 67 11 B., lis. 11 B., 11 8. Blae X White 60, 52 AU blue 10 blue M »» 64 18-75 B., 6-25 S. 20 B., 5 S. Silver X White 56, 57, 58, 62, 73 All blue 15 blue >> i» 63 9-5 B., 9-5 S. 13 B., 6 S. Experiments on Chequering. Non-chequered x Non-chequered 47, 48, 49, 69, 70, 71, 72 All non-chequered 50 non-chequered f» >t 55, 74 10-5 non-C, 3-5 W. 9 non-C, 5 W. Chequered x Chequered 67 All chequered 10 chequered »» »« 65, 66 6 C, 2 W. 4C, 4W. »» »i 61 12 C, 4 non-C. 11 C, 5 non-C fi »» 54, 60 11-81 C, 3-94 non-C, 5-25 W. IOC, 7 non-C, 4\^ Chequered x Non-chequered ... 68 All chequered 8 chequered i> j« ••• 53, 59 9-5 C, 9-5 non-C. 9 C, 10 non-C, i White rump x White rump Blue rump x Blue rump White rump x Blue rump Experiments on Colour of Rump \ 47 5-25 W, rump, 175 B. rump 5 W. rump, 2 B. rump 49, 70 72 All blue rump All white rump 12 blue rump 2 white nmip is to say, that blue is a dilute form of black, and silver a dilute form of dun. Messrs Bonhote and Smalley, however, take the contrary view, regarding silver as dilute blue. As there is no other colour on which the two middle terms of the series can be tested (as they can for example in mice), it is not possible absolutely to distinguish between these two alternatives. In appearance some silvers may confidently be said to contain no black (e.g. Silver Rock), but in others (e.g. Silver Owl and Dragoon) the colour of the wing bars hardly differs at all from that of blues, and probably contains true black pigment. Unfortunately microscopical tests, though extensively tried, have not hitherto provided any satisfactory criterion between the various pigments, and do not add much to what can be seen on ordinary inspection. ^ In these experiments silvers are not included, as it is impossible to be certain whether the rump is coloured or not. Joom. of Gen. u 10 136 On the Inheritance of Colour in Pigeons Account of the Experiments. The numbering of the experiments herein described is continued from that of my previous experiments on colour contained in my first report. Thus the present series of crosses begins at Exp. 47 and ends at Exp. 91. The matings are divided as follows : Exps. 47 — 49 (see Table I) consist of matings of Rock Doves. Exps. 50 — 74 (see Tables II — V) consist of the matings of Rock Doves to Whites. Exps. 75 — 78 (see Table VI) consist of the matings of Owls, Fantails, Carriers and Dragoons. Exps. 79 — 91 (see Table VII) consist of experiments with Doves. Rock Doves used, in Experiments. The specimens of Columba livia used in these crosses came from two sources : (1) three (fs from a pair taken at the Isle of Achill, on the west coast of Ireland, and (2) one ^ and one $ from Lincolnshire. Irish birds were used in Exps. 50, 71 and 72 of the matings with white, whilst Lincolnshire birds were used in all the other crosses with whites. Irish Rock Doves. These birds were kindly sent to me in 1906 by Mr J. L. Bonhote, who had bred them in his aviaries from a pair obtained by him from Achill. He informed me that the birds there were quite wild, and he saw no varieties in the flocks there. There were no tame pigeons within a radius of about thirty miles. Mr Bonhote has bred 19 birds in all from his Irish Rock Doves and their descendants, all true to type, with one exception. This bird was chequered, but since it was produced from a pair of birds that had their liberty, the possibility of a cross was not excluded^ From the descriptions of the crosses made with these birds it will be seen that no question arises as to their purity, and it was found that in F^ from the mating of an Irish Rock Dove with a white the pattern factor segregated out quite cleanly (v. Exp. 51). Lincolnshire Rock Doves. These birds were obtained in 1903 from Lincolnshire through a dealer's advertisement. They were stated to be wild caught, but I could obtain no reliable information concerning them. Their appearance was identical in every respect with that of the Irish birds and other pure 1 V. Bonhote and Smalley, P. Z. S. 1911, p. 605. R. Staples-Browne 137 specimens. On breeding them together, however, certain varieties were obtained, and, as will be seen, they were obviously heterozygous in at least two characters. It frequently happens that semi-domesticated birds join the Rock Doves in their breeding haunts, and no doubt often cross with them. It is stated in Yarrell's British Birds, 4th edition, Vol. Ill, p. 14, that " even in Yorkshire and Northumberland the birds found are open to the suspicion of not being pure wild birds." In 1906 I visited several caves near Flamborough Head which were frequented by these birds in large numbers, and, although no varieties were seen on the day of my visit, I was informed by the boatmen that lighter and darker birds and whites were frequently seen and shot there. Test earperiments with Lincolnshire Rock Doves. When the Lincolnshire Rock Doves were bred together a mixed generation, consisting of three distinct types, was produced. To determine the relationship of these types to one another matings in the direct line were continued and two more generations were raised. The details of the experiments are given in Series A (Exps. 47 — 49), and the results tabulated in Table I. Types of birds produced from the Lincolnshire Rock Doves. (1) Typical C. livia. These birds were identical in every respect with their parents. This type behaves as a dominant to the other two types produced. One of these birds, mated to white, forms a starting point of some of the experiments described in Series B (v. Exp. 52). (2) Blite rumped Rock Dove. The plumage of this type resembled that of the sub-species of Rock Dove known as G. intermedia. The colour of the rump was slightly lighter than that of the back. On examining the series of skins of C. intermedia in the British Museum I noticed that specimens varied slightly in the colour of the rump, some being darker than others. Mr Blyth informed Darwin that the rumps of C. intermedia were sometimes albescent {Animals and Plants under domestication, 2nd edition, Vol. I, p. 193). In the whole of these experiments thirty birds of this type have been produced, and although the colour of the rump varied slightly, the character was always quite distinct, and could be seen at a very early age. In the following descriptions this type is alluded to as Blue with no white feathers (Bl. no wh.). 10—2 138 On the Inheritance of Colour in Pigeons (3) Silver. In this type the blue of the Rock Dove is replaced by a light silver, or, more strictly speaking, cream colour, whilst the wing and tail bars are dun. The head, neck, flight feathers, and tail are much lighter and browner than those of the blue, but are considerably darker than the plumage of the wing coverts, back, breast, and under parts. It is extremely difficult to determine with certainty whether the rump is a very light shade of silver or white, but after examining a large number I did not feel convinced that I had found a really white-rumped bird. In the figures given for the experiments on rump character, therefore, these birds are omitted. Silvers are recessive to both types of blues described above, and breed true when mated together. The silver $ produced by breeding together the two Lincolnshire Rock Doves formed the starting point of the experiments described in Series C. TABLE I. Experiments with Lincolnshire Rock Doves. Offspring Bjcp. No. Female Origin Also from used in Exp. Exp. Origin from Male Exp. Also Blue used in no Exp. white Typical Coiumba livia Silver 47 Typical C. livia (no number) — — Typical C. livia — (no number) — 2 5 1 48 Silver 43 47 55,56,57, 58, 68 Typical C. livia 44 47 50, 52 4 6 — 49 Blue, no wh. 27 48 — Blue, no wh. 39 48 — 8 — 2 Details of the mating s of Lincolnshire Rock Doves. Series A. (v. Table I.) Exp. 47. Typical C. livia $ x Typical G. livia ^. These two birds, when mated together, produced eight offspring, of which five were typical C livia, two were blue with no white feathers, the rump being a lighter shade of blue, and one was silver with dun wing and tail bars, the rump being apparently a very light silver. Now since subsequent experiments show that blue is dominant to silver and white rump dominant to blue rump, we may conclude that had this experiment been prolonged we should have had White-rumped blues, Blue-rumped blues and Silvers in the proportion of 9:3:4, The observed figures were 5:2:1. R. Staples-Browne 139 Exp. 48. Silver ? 43 x Typical G. livia 05 05 00 W lO >o »o U5 tr- ti ^ ^ ^ 1 co •" 00 00 t~ t- 1 "S 'O lO «5 U5 (n' S - «« ^ ^ to a t- to to to o »o U5 US >o 3 CO 00 00 00 -* -* -* •a t- t- t- t- CO 05 o Tjt ^^ ^ Ui »o lO « CO «5 CO i-H TJI ■* -* ■* iH 1-< CO .i i i i CD f 1 § > o a o a > o > i d o d jn m w. m pq OQ 4: s ^ § l_^ Cm 5. „ „ no white feathers, 1 J Silver, non-chequered, much white, 1\ Silver, chequered, very few w.f. on rump and thighs, 1. White 2. The numerical results of this F^ mating follow fairly closely the Mendelian ratios, the observed figures being : Coloured : White : : 8 : 2. Of the coloured birds : Blue : Silver : : 6 : 2. Chequered : non-chequered : : 6 : 2. White feathers in plumage. No w.f. : very few w.f. : several w.f : : 1 : 2 : 5. Eocp. 61. Silver chequer, no wh. $ 31 x Silver chequer, no wh. j/* 2. Both these birds were raised in Exp. 59. The mating was continued for two years, and 16 young were produced, all silver, of which 11 were 1 This bird was an intermediate and is counted as non-chequered. R Staples-Browne 151 chequered and 5 nun-chequered'. The Meudelian expectAtion on this mating was chequered : non-chequered :: 12 : 4, with which ratio the figures obtained comply. No white feathers were produced; we may therefore presume that white was not being carried by either parent. In the nest a few of the birds were thought to have some white feathers, but these were seen to be light silver when the birds grew up. One bird also which had some feather deformity was counted as non- chequered in the nest ; on maturing, however, when the feathers attained their normal development, it was seen to be a light chequer. Series D. {v. Table IV.) Exp. 62. Silver chequer, no white ? 31 x White • CO GIGANTISM IN PRIMULA SINENSIS. By FREDERICK KEEBLE, Sc.D., Professor of Botany in University College, Reading. CONTENTS. SECTION '*<*■ I. Introdaction 163 n. Gigantism in Primula tinensis 164 TTT. The Mode of Origin of Giant White Queen Star 165 rV. A Histological Comparison of Giant with Normal White Qneen Star. 168 V. The Genetical Behaviour of Giant White Queen Star: with Notes on the Genetics of Gigantism in P. sinensis 172 VI. Theoretical Considerations 185 Vn. Summary 186 L Introduction. Many of our cultivated plants are to be met with in giant and dwarf, as well as in normal forms, and anyone who will consult the pages of nurserymen's catalogues may learn that the plant breeder has exercised his genius no less successfully in moulding the form of plants than in modifying the colours of their flowers. The universality of the phenomenon of gigantism justifies the belief that a comparative study of the physiology of giant and normal plants will lead to results of interest and value. But such a comparison must be accompanied by a study of the genetical relations which exist between giant and normal forms ; for although, so far, only one type of giant has been observed among plants, analogy with animals would suggest the possi- bility of the occurrence of more than one type of gigantism in the vegetable kingdom. {Cf. Gilford, 1911.) The plants in which the phenomenon of gigantism has been investi- gated include Oenothera gigas (de Vries 1901, Gates 1909), Lathyrus odoratus (Bateson 1909 a) and Pisum sativum (Mendel, see Bateson 1909 B and Keeble and Pellew 1910). In all these cases the giant is 164 Gigantism in Primula sinensis dominant to the normal form. Thus, as de Vries {op. dt.) has shown, Oenothera gigas, a giant mutant from 0. Lamarckiana, when crossed with the latter form (0. Lamarckiana x 0. gigas) yields an F^ consisting of plants all possessing giant characters. In both Lathyrus odoratus and Pisum sativum the character of gigantism is known to be deter- mined by two factors, the presence of both of which is necessary for the manifestation of the character. The purpose of the present communication is to record the origination of a giant form of P. sinensis from a normal form the pedigree of which is known, to describe the histological characteristics which distinguish this giant, to present the results of the experiments which have been made on the genetics of gigantism and to discuss a few of the problems which are suggested by the results of these experiments. II. Gigantism in Primula sinensis. x^mong the many varieties of Primula sinensis in cultivation at the present day there are not a few which are characterised by an imposing massiveness of both floral and vegetative organs. The races thus distinguished are known to florists as giants and are prized by them on account of the remarkable size and substantiality of their flowers. Associated with the gigantic habit are a somewhat slow rate of growth, a certain leisureliness of flowering, and, not infrequently a considerable measure of infertility. It is probable that the last mentioned characteristic accounts for the opinion, not uncommon among gardeners, that giants are more difficult of cultivation than are the normal races. This opinion, however, is not well founded, nor is there any ground for the assertion, which is often made, that the gigantism of these strains depends on methods of cultivation. The fact that seedsmen are able to offer giant races year after year i^ well nigh sufficient evidence that gigantism is a " fixed " character, and that giants, when self-fertilized, breed true to form. Beyond these facts little is known either of the origin of the giant races of P. sinensis {cf. Gregory 1909 and 1911) or of their genetical behaviour. The physiological peculiarities of giants are likewise unknown. The giant races of P. sinensis which are offered by seedsmen are almost invariably of the sinensis type both with respect to habit and flower: that is to say the flowers are produced in a compact and F. Kbkble 165 massive head and bear large petals the outer margins of which are not notched {stellata type) but fimbriated. There is however no necessary connection between gigantism and the sinensis type of petals: indeed Messrs Sutton's remarkable collection of Primulas contains at the present time numerous giants which bear flowers of the stellata type. Further, although the giants in general cultivation have the sinensis style of inflorescence, there is no essential connection between gigantism and that type of inflorescence. For example, the giant (Giant White Queen Star), the origin of which is about to be described, is charac- terised by the possession of a typically stellata habit of inflorescence (c/ Plate XI, fig. 2). III. The Mode of Origin of Giant White Queen Star. Among the Primulas which have been grown at University College, Reading, during the last ten years is a well-known variety, White Queen Star (Plate XI, fig. 1). This variety — a reddish stemmed, palm- leaved strain with white flowers of stellata form and habit — was raised from seed presented by Messrs Sutton in 1903. Plants of the first generation of White Queen Star raised from these seeds were self-fertilized in 1904 and yielded an Fi generation which, so far as the records show, presented no departure from the normal form. Of the plants of the F, generation raised from seed obtained by self-fertilizing F^ plants, some bore an occasional 6-petalled flower among the other normal 5-petalled flowers. Fluctuations of this kind are of course common in P. sinensis, and also in many other cultivated plants. They appear suddenly in one generation, may be lost sight of in the next; and their advent and disappearance are generally ignored. Notwithstanding the strength of the evidence that such fluctuating variations arise and fade away without leaving recognizable after- effects in the descendants, experiments were undertaken with the object of testing the degree of permanence of the variation which results in the production of flowers with supernumerary petals. As may be seen from the results of the experiment summarised in Table I the attempt to produce a race characterised by the p)ssession of flowers with more than the normal number of petals was not successful. Inasmuch however as the experiment of breeding from 166 Gigantism in Primula sinensis abnormally petalled plants was continued only for a few generations the negative results cannot be claimed as decisive. TABLE I. Selection Experiments tvith White Queen Star {W. Q. S.). W. Q. S. from seed no abnormality observed. Selfed W. Q. S. some plants with occasional 6-petalled flowers selfed a 6-petalled flower plants with most flowers normal, a few with 6 or 7 petals selfed a 7-petalled flower ditto ditto 5A bB oD 1903 1904 Fi 1905 Fi 1906 F, 1907 ^4 1908 Ffi 1909 Fe F7 F» 1910 1911 1912 10 plants petals slightly fimbriated 8 with 1 or more 6-petalled flowers flowers large selfed 11 plants all giant (see Table 11) selfed I = 10 plants all giant selfed = giant — giant = giant 12 plants 11 normal 1 with 1 flower with 6 petals selfed 6 plants all normal 4 plants some flowers with supernumerary petals selfed 2 plants selfed all normal 13 plants aU normal selfed all normal selfed = normal = normal = normal 12 plants 5 normal 7 with occasional 6-petalled flowers It may be added that, as shown in Table I, the habit of forming supernumerary petals has not manifested itself in later generations to so marked a degree as it did during the time of the selection- experiment, and that the race of White Queen Star now growing at Reading, which consists of descendants of the 6- or 7-petalled plants, is not distinguished to any marked degree by this abnormality. The interest attaching to the series of experiments which has just been described lies, however, in another direction. For it was in the course of these experiments that a Giant variety of White Queen Star was obtained. The mode of origin of the Giant race is set forth in detail in Table II. As indicated in Table II, gigantism manifested itself in the Ft generation (1908). That generation, which was composed of 11 plants, consisted of giant forms only. The striking appearance presented by F. Kkeblk 167 TABLE II. The Origin of Giant White Queen Star. 1903 White Qaeen Star (W. Q. S.) raised from seed 1904 = normal W. Q. S. selfed 1905 F] plants with occasional 6-petalled flowers : most flowers normal I selfed 1906 Fi plants with occasional 6- and 7-petalled flowers : most flowers normal | selfed 10 plants, all with slight fimbriation of edges of petals, 8 with 1 or more 6-petalled flowers : petals large, selfed 1 plant 11 plants : flowers large, of good substance ; petals meeting or over- lapping. Flower slow in opening ; slight fimbriation of edges of many petals = (Hant White Queen Star, see Plate XI, fig. 2. selfed 3 plants all true to Giant form and habit 7 ,, ,, ,, selfed 8 plants all trne to Giant form and habit 1 plant 2 plants „ „ „ selfed 6 plants all true to Giant form and habit selfed 1 plant „ „ „ 30 plants all true to Giant form and habit these "mutants" is illustrated in Plate XI, figs. 1, 2 — and a com- parison of the giant and normal forms shows how markedly they dififer the one from the other. The differences between the two forms are not confined to the flowers, though they are most patent in the petals (cf. Fig. 1). Giant White Queen Star, the origin of which is now 1907 ^3 {oA) 1908 Fi 5A1 1909 Fs 5A1I1 5A1I2 1910 Fa 5A 1/2/7 5A 1/2/8 5A 1/2/9 1911 F7 5^9 5^10 1912 Fs G. W. Q. S. Fig. 1. Corolla of Normal and of Giant White Queen Star. Joam. of Gen. n 12 168 Gigantimn in Primula sinensis under consideration, is a more massive plant, with slower growth than that of the ordinary form of White Queen Star from which it arose. The corolla of the Giant is about half as large again as the corolla of the normal form and adjacent petals of the former either meet or overlap instead of leaving a narrow space between them as is the case with the normal plant. The overlapping of the contiguous petals is a characteristic of giants both of the sinensis and stellata type of flower, and is due to the fact that the oblate petals are much broader basally than are the more ovate petals of varieties of normal stature. In one respect only does Giant White Queen Star differ from most giants. It has retained the tiering habit of the typical stellata form of inflorescence. This example of a giant form characterised by a typically stellata habit of inflorescence is not unique. Thus the Giant White Star which has been investigated by Gregory (1909 and 1911) possesses also the stellata type of inflorescence. Giant White Queen Star conforms in all respects with the known giant races of P. sinensis. Its flowers are large, of good substance, and slow to open. Its notched petals — the edges of which are apt to be slightly fimbriated — are rounded in the bud stage, grow slowly, and as they grow broaden basally till those adjacent to one another meet or overlap. The flowers are much more lasting than are those of the normal parent, and the massive stigma persists for weeks if pollen be denied access to it. IV. A Histological Comparison of Giant with Normal White Queen Star. The most comprehensive examination of the histological character- istics of a giant form of plant is that made by Gates (ioc. cit.) in the case of Oenothera gigas. Comparison of 0. gigas, a giant mutant of 0. Lamarckiana, with the normal form from which it arose has led Gates to the conclusions that the cells in the giant are conspicuously larger than in 0. Lamarckiana; that the cells of certain tissues of the giant are almost exactly twice the size of those in the corresponding tissues of 0. Lamarckiana ; and that the giant, 0. gigas, has double the number of chromosomes present in 0. Lamarckiana. Histological comparison of Giant and Normal White Queen Star, shows at once that the former is a giant because its constituent cells are gigantic. The marked difference between Giant and Normal White F. Kkkblk 169 Giant 5/a/lO. Normal 5/d/4. Fig. 2. Primula sinensis. Cross sections of the flower peduncles of Giant White Qaeen Star and of Normal. Magnification the same in both. 12-2 170 Gigantism in Primula sinensis Queen Star with respect to size of cells is demonstrated by the Camera Lucida drawings reproduced in Figs. 2 — 5, and appears to be in every way similar to that obtained by Gregory (1909) in the case of the Normal and Giant Star Primulas which he compared with one another. The difference in size between the nuclei of the giant and normal form which Gregory demonstrated, obtains also in the example now under ctx, X, Fig. 3. Primula sinensis. Longitudinal sections of the flower peduncles of Giant White Queen Star and Normal. Magnification the same in both. e = epidermis, cte= cortex, a; = pericycle. consideration. The larger size of the nuclei of the giants is well shown in Fig. 5, which represents the pollen grains of the pin-eyed normal and giant varieties. The histological aspect of the phenomenon of gigantism in P. sinensis deserves more detailed examination than has been devoted to it, and in particular an enquiry into the relation between size of nucleus and size of cell should prove of interest ; but beyond confirming Gregory's con- clusion that the numbers of chromosomes (12 and 24) are the same p. Keeblk 171 Fig. 4. Primula sinensis. Stomata and Epidermal cells of Giant and Normal White Queen Star. Magnification the same in both. Fig, 5. Primula tinejuis. Pollen grains of Giant and Normal White Queen Star. Both Pin-eyed. Magnification the same in both. Grains swollen and stained by Methyl green Acetic. 172 Gigantism in Primula sinensis in the giant and in the normal varieties, no detailed histological study of the two forms has, as yet, been undertaken. The rough histological analysis, which has been made, suffices, however, to prove that in the variety Giant White Queen Star, the gigantism of the individual is the expression of that of its con- stituent cells. The mutant is a giant because its cells are gigantic. Further, the facts that, in the cortex, the number of cell-layers is fewer in the giant than in the normal plant, and that the rate of growth of the plant as a whole is slower in the former than in the latter, suggest that cell gigantism may be due to a reduction in the normal rate of cell-division. Examination of Figs. 2 — 5, shows that the cells of the giant are larger than those of the normal form in all three dimensions — radial, tangential, and longitudinal, and that gigantism is not peculiar to the elements of any one tissue-system, but is common to the cells of all — epidermal, cortical, and stellar. The extent of the differences in size between the cells of the giant and ordinary White Queen Star is illustrated by the following figures, in which the size of a giant's cell is represented by 100. Flower peduncles (see Figs. 2 and 8, and Plate XI). Cortical cells of the layer immediately external to the endodermis : — Giant : Normal Radial measurement 100 : 48 Tangential „ 100 : 81 Longitudinal „ 100 : 57 V. The Genetical Behaviour of Giant White Queen Star: with Notes on the Genetics of Gigantism in P. sinensis. The race, Giant White Queen Star, the origin of which is described in Section III, differs markedly from the parent race with respect to fertility. Whereas plants of normal White Queen Star produce seed freely both when self-fertilized and when crossed with other varieties of P. sinensis, the giant race is relatively sterile. Nevertheless self- fertilization of the original giants (1908, Table II) resulted in the production of enough seed to continue the strain. The partial self- sterility which characterised the first generation of giants has not been maintained, and the descendants of the original Giant White Queen Star now yield when self-fertiUzed a fair amount of seed. F. Keeble 173 As indicated in Table II, the giant race has been continued by self- fertilization for four generations, and several hundreds of descendants have been cultivated. All of these have been observed with care, and in the case of 58 plants (Table II) written records of their characters have been kept. All have without exception been true to the giant habit. On the other hand all attempts (see Table III) to cross Giant White Queen Star with other varieties have — with one doubtful ex- ception to be mentioned immediately — proved abortive. Whether the giant race be used as the seed parent or as the pollen parent ; whether it be crossed with normal or giant forms, with stellata or sinensis forms, the result is the same. Even when it is crossed with the parental race of White Queen Star, and this cross has been repeated many times, no seed has been obtained except on one occasion, when two fertile seeds were produced (see Table III). The last-named cross has been repeated often and advisedly, because it was recognized that in the success thereof lay by far the best chance of obtaining material suitable in all respects for an experimental study of the genetics of gigantism. Despite the negative results obtained by crossing Giant White Queen Star with other varieties, and inasmuch as it would appear probable, from analoofous cases, that this refractoriness to cross-fertilization will disappear, it seems worth while to record the varieties (see Table III) with which the attempts to cross fertilize Giant White Queen Star have been made. The Mendelian phenomena presented by the offspring produced by crossing giant and normal plants are not altogether easy of interpre- tation. This, however, is due rather to the difficulties of experimentation and particularly of observation, than to the complexity of the problem. Since it is evident from the facts recounted already that gigantism is a cell phenomenon, it follows that the genetics of gigantism should be investigated with reference to the cell, and that, if Mendelian results of certain value are to be obtained, the attention of the observer must be directed not only to the dimensions of the individual plants or flowers, but also to those of the cells of the plants with which breeding experiments are made. The experiments, the results of which are now to be discussed, were undertaken before this need was realised, and the experiments which were suggested by the advent and cell characteiistics of the mutant Giant White Queen Star have never been made, owing to the obstinate sterility of that form. Hence it must be admitted at 174 Gigantism in Primula sinensis TABLE III. Record of Crosses between Giant White Queen Star and other varieties of P. sinensis. Giant White Queen Star = ? Year 1908 1909 Catalogue Number of the Variety bB 15 Name of Variety Normal White Queen Star Euby Star 552 Normal White Queen Star 5 Z)3/2 Normal 5D3/2 1908 1909 1911 16/1 (SBpl 20/3/1 130 B/2 25 190 £ 180 62/6/1 White Star Ivy Leaf Cambridge Blue Snow King I\, Reading Pink x Crimson King Snow Drift ... Orange (Coral) Pink Carnation Mauve Flake Crimson King Duchess Queen Alexandra Giant Primula sinensis A Blue, Semi-giant ... Character of the Variety Besult Normal stellata no seed " " 2 seeds 1 which germinated ,, ,, no seed ,, ,, a few seeds, none germi- nated .. no seed Normal sinensis Giant form Semi-giant form Giant White Queen Star = Giant Pink Giant Primula sinensis Giant White (Sutton) Giant F2 , Royal White x Snow King F3, Reading Pink x Crimson King A White Sinensis ... Snow Drift Duchess Normal White Queen Star Cambridge Blue Snow King ... no seed Giant form Normal sinensis Normal stellata ^ The plants thus raised were typical Giant White Queen Star. Though possibly they were due to accidental selling of the female parent, they yielded no seed neither when self-fertilized nor when crossed with normal White Queen Star. F. Keeble 175 the outset that the method by which the following records were made — that of appraising gigantism by reference to the gross ap- pearances presented by individuals — is open to grave objections. For example, a large and massive flower may, if seen among a family of small-flowered plants, be classed unhesitatingly as a g^ant, although if it were judged side by side with indubitable giants, it might be relegated to the class of doubtful or semi-giants. On the other hand an observer who spends a considerable amount of time among plants of a certain kind gains in some measure a sureness of judgment with respect to the " points " of those plants which is not likely to lead him into very serious errors of observation. In spite of the drawbacks of the method of classification which has perforce been adopted in these experiments, the results which have been obtained appear to be worth recording both for their inherent interest and for the purpose of demonstrating that the phenomena of gigantism presented by P. sinensis are less simple than those exhibited by other plants which have been the subject of like investigation. For example, the genetics of the gigantism of sweet peas (Bateson, 1909 a) and of culinary peas (Keeble and Pellew, 1910) is of a fairly simple kind. Thus in Pisum sativum gigantism depends for its ex- pression on the presence of two factors.* Of these growth-factors, the one induces excess of growth in length (factor for long intemode), the other induces excess of growth in thickness (factor for thick intemode). In the presence of both these factors, either in a pure or hetero- zygous state, the plant is a giant (6 ft.), in the absence of either it is of mid-stature (3 — 4 ft.), and in the absence of both it is a dwarf (1-H ft.). The gigantism of P. sinensis is built on less simple Mendelian lines. The experiments which lead to this conclusion, recorded in Tables IV, V, and VI, were made mainly at University College, Reading; though some records which are included in the tables are those of breeding-experiments carried out by Messrs Sutton and Sons at their trial grounds. The records derived from the latter source are indicated in the tables; those in which the source is either not mentioned or indicated by the letters U. C. R. were carried out at University College. For access to the records of Messrs Sutton and Sons' experiments 176 Gigantism in Primula sinensis the very cordial thanks of the writer are due to Mr Leonard Sutton and to the firm's Primula experts, Messrs Macdonald and Tufnail. From what has been said already with respect to the mode of inheritance of gigantism in Pisum sativum it follows that the Fi generation resulting from a cross between a pure giant and any other form consists of giants. A like result is not exhibited by Primula sinensis in the ^i generation derived from a similar cross. In some instances (see Table IV) the Fi progeny of a cross between a giant and a normal form are of giant type : this is the case, for example, in the Fi of the crosses Giant Pink x Reading Pink and Duchess X Giant Lavender. In other instances, for example, in the TABLE IV. Crosses between Giant and Normal P. sinensis. Results in F^. I. Cases in which Gigantism appears to be more or less completely dominant. Source of Experi- Record ment No. Nature of the Parents F, Results U. C. R.i 06—7 31 Giant Pink x Reading Pink (Normal) 12 plants of giant type 32 A Royal White (Giant) x Pink Stellata 14 plants, flowers like (Normal) those of Royal White ( Giant) but freer in habit . and on longer pedicels „ ,, 32 F Pink Stellata (Normal) x RoyalWhite 11 plants, flowers like (Giant) those of Royal White (Giant) but freer in habit and on longer pedicels Sutton 2 II, p. 71 Duchess (Normal) x Giant Lavender Giant II. Cases in which the F^ is intermediate with respect to Gigantism of flowers. Sutton II, p. 20 Giant Royal White x Crimson King Semi-giant (Normal) „ II, p. 32 Crimson King x Giant White ... 8 plants, semi-giant „ I, p. 22 Royal White x Crimson King ... "Habit of flowers spoiled " III. Cases in which Gigantism of flower appears to be I'ecessive. U. C. R. 80/07 Giant Pink x Lord Roberts Star Petals slightly fimbri- (Normal) ated ; flowers not re- corded as showing giant habit „ 39/08 Lord Roberts Star x Giant Pink ditto ditto ^ U. C. R. = record of experiment carried out at University College, Reading. ^ Sutton = record of experiment from the record books of Messrs Sutton and Sons, Reading. F. Keeblb 177 cross Giant Royal White x Crimson King the F^ family consists of semi-giants, and in yet other cases the ^i plants were not recorded as showing giant habit of flower: whence it is to be inferred that if they possessed any symptoms of gigantism those symptoms were too slight to attract notice. The one case in which seed was obtained as a result of crossing Giant White Queen Star, though it must be recorded, is open to suspicion. The cross in question (Table III) was one between Giant White Queen Star and its parent, the normal White Queen Star. Two seeds only were obtained, and yielded plants of giant habit. It is highly probable that they were the result of chance self-fertilization of the giant, although it is to be noted that these ^i plants were sterile both with their own pollen and that of normal White Queen Star. A superficial consideration of the differences which subsist between the ^1 generations might lead to the conclusion that two types of gigantism occur in P. sinensis : one type in which the gigantism behaves as a dominant character; and the other in which it behaves as a recessive. Such a conclusion is not open to objection on general, theoretical grounds, and meets with support from the known facts of gigantism in human beings. For as Gilford (op. cit) shows, overgrowth in man may be of a normal type or of a pathological nature. In the former it is due to an exaggerated but normal development : in the latter it would appear to be the consequence of the lack of a growth-controlling factor. So in plants, excessive growth may be the outcome of the presence of a factor for growth-acceleration (or for inhibition of cell-division), or may be due to the absence of a factor which in the normal plant controls and limits the amount of cell- growth in the interests of the several organs or of the organism as a whole. More careful examination of the results obtained with P. sinensis serves to show, however, that to apply such an hypothesis to their interpretation is certainly premature and probably unnecessary. Hence in the argument now to be developed it will be assumed that there is but one type of gigantism in P. sinensis and that gigantism is dominant to normality. The variable extent to which the giant character manifests itself in Fi generations must therefore be susceptible of explanation in terms of this hypothesis. An inspection of Table V shows that, as with the P, generations, so with Pj generations, there is a remarkable lack of uniformity with respect to the manifestation of gigantism. Thus the P, and subsequent 178 Gigantism in Primula sinensis generations produced from self-fertilized F^ plants may show either a considerable preponderance of giants or an even more marked excess of normal (non-giant) forms. TABLE V. Crosses between Giant and Normal P. sinensis. Results in F^ and subsequent Generations. I. Cases in which Giants appear in large numbers. Source of Record Experi- ment No. Sutton II, p. 32 Sutton I, p. 71 Nature of the Cross Results in F^ and f , Crimson King x Giant White F2 = 18 giant : 5 non-giant. Fi = Semi-giant Several F^ = plants true to giant Duchess X Giant Lavender i^2> giant, F3, 18 plants all giant Fi = giant U. C. R. U. C. R. U. C. R. II. Cases in which Giants appear in small numbers, ... 120/07 ... 80/07 Royal White x Pink Stellata Fi = flowers in massive heads Giant Pink x Lord Roberts Star Fi — flowers slight sinensis none recorded as giant Royal White x Snow King Fi = no flowers recorded as giant F2, 67 plants of which 1 = giant, another F2, 54 plants no giants F2, 211 plants of which 9 = giant 14/2/1 Royal White x Snow King F^, 13 plants, 11 non-giant, ~ " " " 2 giant ^3 > of giant -F2 = 7 plants all giant, petals overlapping with very fimbriated edges ■^3) of giant ^3 = 14 plants all giant, petals overlapping with very fimbriated edges F3, of non-giant F2=& plants, 5 non -giant, 1 giant F3, of non-giant ^2=11 plants, none giants F3, of non-giant F2=l plants, none giants A clue to the significance of this diversity of behaviour is provided by Exp. 120/07 (Table V). In this experiment a true-breeding giant. Royal White crossed with Normal Pink Stellata yielded an F^ in which the flowers though non-giant were borne in massive heads — owing, as shown in Table VI, to dominance of the stellata habit of free-tiering. TABLE VL Dominance of Stellata over Sinensis Habit of Inflorescence. Royal White x Pink Stellata Fi= habit freer than Royal White, pedicels longer 2^2=22 stellata habit: 6 sinensis habit in 3 : 1 ratio = 21 stellata habit : 7 sinensis habit F. Keeble 179 The Fi plauts yielded an F, generation which in one case consisted of 54 non-giants and no giants and in another case was composed of 67 plants of which one was a giant. The ratio 66 non-giant : 1 giant suggests the further hypothesis that gigantism depends for its expression on three factors. Let it be assumed that a giant differs from a Normal Pink Stellata in the possession of at least three factors, which factors are either different in nature or — and more probably — similar in nature and of different distribution in the germ-plasm. Then when the giant of constitution AABBCC is crossed with a form which by virtue of its size and habit may be assumed to lack all three factors the result is an Fj, the members of which are heterozygous for all three factors. Thus : AABBCC X aabbcc, Fi = AaBbCc non-giant, and F2 = 1 AABBCC : 63 plants of other constitutions, = 1 indubitable giant : 63 other plants, none of which appears as a giant when viewed beside the pure dominant giant form. Results which point to ratios of this order are exhibited in Table V. Thus in No. 120/07 the F. of the cross Royal White x Pink Stellata consists of 67 plants of which 66 are non-giant and 1 is giant. In another F^ family of the same origin no giant appeared among 54 plants. The case of Snow Drift x Snow King (Table VII, I) is similar and of interest in another direction also, inasmuch as it affords an example of the origin of a giant race as the result of crossing two non-giant varieties. The F2 generation consists of 33 plants of which one was recorded as a "doubtful giant" (51/2/1). This plant yielded an F^ consisting of seven plants of which five are non-giant and two are giants. Four of the non-giant ^2 plants yielded F3 families which together contained 59 plants none of which are giants. The results of this experiment and those of the experiments described previously conform with the requirements of the hypothesis that floral gigantism is determined by three factors, all of which must be present in the homozygous condition for the phenomenon to be exhibited. The results now to be described require a slight modification, or rather extension, of this hypothesis. As mentioned already the "doubtful giant" (51/2/1) obtained in the F.i of the cross Snow Drift x Snow King (Table VII, I) yields an Ft consisting of 5 non-giants and two giants. Whence it follows that 180 Gigantism in Primula sinensis the floral habit of a plant may approach so nearly that of a giant as to be recorded as a "doubtful giant" although, as shown by its progeny, it is not pure for the giant character. TABLE VII. The Production of Giants hy the Crossing of Non^Giant Forms. 1909 1910 F,= 1911 JF,= 1912 F3 = Snow Drift x Snow King^ 22 plants : no giants recorded selfed 2 plants 51/1 = 17 plants : no giants 51/2 = 16 plants of which 1 = (? giant) (51/2/1) selfed (? giant) and 4 non-giants ' 5\j2j\ (from (2 giant) ^2)= 7 plants, viz. 5 non-giant, 2 giant 51/1 (from non-giant F2) = 13 51/H „ „ = 5 511 M „ „ =14 51IW „ „ =27 - 59 plants : none giant II 1909 1910 1911 1912 Fi = F2= F.= Snow Drift x White Queen Star (Normal) 9 plants, no giant 8/2/2 r= 15 plants, no giant {8/2/11 =1 plant non-giant 8/2/2/11 = 9 plants, no giant 8/11 = 18 plants J 1 with giant flowers t 1 ? giant = 17 (? 16) non-giant : 1 (?2) giant It must therefore be assumed that combinations of the factors A, B, C other than the combination AABBCC may give rise to giant-like forms. The assumption which appears to fit the facts most nearly is as follows : Of the three factors A, B, C two, but not any two, must be present in the homozygous state for the definite manifestation of gigantism. Provided that the plant have the constitution AABB, it may exhibit well-marked gigantism even though the third factor C be present in the heterozygous condition. Thus the three factorial combinations AABBCC, AABBCc, AABBcC may all produce giant-like plants, albeit the homozygous giant is recognizably more gigantic than the hetero- zygous giants. The genetical behaviour of the "doubtful" giant 51/2/1 (Table VII, I) demonstrates that it is not pure to gigantism. If there be ascribed to it a constitution AABBCc, it should yield an F3 of lAABBCC : 2 { ^^^^^^ : lAABBcc. (AABBCc 1 Snow King has flowers of fair size (stellata type) and should perhaps be classed as a semi-giant. F Keeble 181 The first is an undoubted giant, the last is an undoubted non-giant, the two others are of the constitution of the F, parent, namely hetero- zygous for C, and although the parent arising in a family of smaller forms was classified as a doubtful giant, these plants now that they are seen side by side with the more massive (AABBCC) plant are thrown unceremoniously into the category of minor forms. Classified thus the Fi consists of : 1 giant : 3 non-giant in seven plants 1'75 „ : 525 „ expected 2 „ : o „ found. If the ultimate object of Mendelian analysis were merely to fashion constitutions on slender experimental bases, an hypothesis such as that now in course of formulation would be scarcely worth the making ; but it must be remembered that one of the prime objects of Mendelian analysis is to provide classification with more subtle methods than those on which it relies at present ; and, as is shown immediately, the hypothesis now in course of adumbration does lead to a better system of classification of the cultivated forms of P. sinensis than could be obtained by any other method whatsoever. The validity of the assumption that well-marked gigantism may only be manifested by plants which, whilst possessing a certain factor C in the homozygous or heterozygous condition, are pure with respect to the presence of the other two factors (AA) and (BB) is borne out by the results of experiment 80/07 (Table V). In this experiment a true breeding giant (AABBCC) Giant Pink was crossed with Lord Roberts Star which by reason of its delicate habit of flower may be regarded as of the constitution aabbcc. Giant Pink x Lord Roberts Star, AABBCC X aabbcc = F, : AaBbCc, and such an F^ on self-fertilization yields an F2 composed of : giant and r 1 AABBCC approximately J lAABBcC : 61 non -giants giant forms I lAABBCc = in64 plants: 3 : 61 = „ 211 „ 9-9: 2011 actual F, = 9 : 202. 182 Gigantism in Primula sinensis Although not directly germane to the subject of gigantism it may be recorded here that the form of inflorescence of the cultivated varieties of P. sinensis appears also to be determined by the mode of distribution of three factors in the zygote. Thus in the cross just described (Giant Pink x Lord Roberts Star) the somewhat delicate and few-flowered type of inflorescence character- istic of Lord Roberts Star disappears in the F^ generation and in the Fa generation there are produced : 212 non-" Roberts" inflorescence : 8 "Roberts": which ratio points to the oonclusion that general type of inflorescence is determined by three factors (XYZ) ; that the Lord Roberts Star type which is patently more feeble than either the sinensis or stellata types is produced in pure form only when the zygote has the constitution xxyyzz ; and that Roberts-like inflorescences are also produced in plants of the constitutions Xxyyzz and xXyyzz. If this be so then in 64 F2 plants there are to be expected : 3 Roberts : 61 non-Roberts and in 211 plants 99 „ : 2011 whereas 8 „ : 203 „ were found. The finer details of form of inflorescence, length of peduncle, length of pedicel etc., appear also to be determined by Mendelian factors. The evidence in support of this statement must be reserved for a further communication, although the fact, that the Mendelian method may aid the plant-breeder to trim up a plant to almost any desired form — to straighten the leaves, to elongate the pedicels, to lower or heighten the inflorescence — deserves to be brought to the attention of professional plant-breeders. To return to the subject of gigantism : the hypothesis that this phenomenon depends on not less than three factors drives us directly to the conclusion that the classification of varieties of P. sinensis into giant and non-giant forms is illusory. For it is a necessary corollary to that hypothesis that the mode of distribution and combination of these factors must be very different in the different races. In other words the conclusion is inevitable that, as is notoriously the case in other cultivated plants, P. sinensis must contain not only giant and dwarf strains but also semi-giant races. Further the hypothesis enables us to understand our failure to recognize, previous to Men- delian experiment, the existence of such diverse races. For with three factors concerned in the determination of stature the number of F. Kej^blk 183 intermediate forms must oecessarily be large, and these forms must al»o produce the illusion of a continuous series rather than of a series made up of a large but definite number of forms, each of definite constitution and each therefore distinct from the others in genetical behaviour (c/. Nilsson-Ehle (1909) and Baur (1911)). Mendelian analysis thus leads to the recognition that just as with antirrhinums, peas, sweet-peas and hosts of other cultivated plants, so with Primula sinensis we have to deal with giants, dwarfs and middle races. The case already described in which giants arose as the outcome of the mating of non-giants is at once intelligible when this fact is grasped. Two instances of the appearance of giants in this manner are recorded in Table VII. In one, Snow Drift x Snow King a doubtful {F2) giant yielded an F3 of 5 non-giant : 2 giant ; in another, an ^s from Snow Drift x White Queen Star, the ratio is 16 (? 17) non- giant : 1 (? 2) giant. Now although both Snow King and White Queen Star are alike in their stellata flowers and stellata habit of inflorescence, the flower of Snow King is distinctly larger and the petals more massive than is the case with White Queen Star. In other words Snow King is a semi-giant. It must therefore contain more of the factors for gigantism than are borne by White Queen Star. By ascribing factorial formulae consistent with their apparent con- stitutions to the varieties Snow Drift, Snow King and W^bite Queen Stai- it is possible to account for the several results obtained by crossing each of the two latter varieties with the former variety. Thus and by way of illustration only, if the constitution of Snow Drift be aabbCC and that of Snow King be AABBcc then Snow Drift x Snow King = aabbCC x AABBcc and ^1 = AaBbCc Fn = S giant and giant-like = 3 as compared with 1 = 4 61 non-giant 61 15 „ found 60 A further point worth bearing in mind is that the conception of three factors admits of the explanation of minor but constantly recurring variations in shape and size of flower. For in a family which lacks the Journ. of Gen. n 13 184 Gigantism in Primula sinensis C factor, one of the other factors may be present in homozygous condition in some members and in heterozygous condition in others. In such a family, which can never produce a giant form, the constituent individuals may be characterised according to their respective factorial constitutions by different modes of growth of the corolla and other parts. In the variety Mont Blanc Star for example there are to be met with constantly plants which bear smaller flowers than the type. These peculiar flowers are characterised not only by their smaller size but also by the fact that the basal parts of the petal-lobes are more fused with one another — approaching slightly to gamopetaly — than are the corresponding parts of the flowers typical of the variety. The assumption of the existence of three factors for size of corolla throws light on this phenomenon. Lacking altogether one of the three factors for gigantism the variety Mont Blanc Star cannot throw giants but if its constitution be AaBBcc it may throw both AaBBcc and AABBcc forms, the former in larger numbers than the latter. If the AABBcc form differs, as differ it must, from the AaBBcc form it is described as a fluctuation. In other words fluctuations or minor variations may owe their origin to the hetero- zygousness (for one or more factors) of a factorial complex which is completely lacking in one factor essential for the production of a given Mendelian character. On the assumption that growth factors may condition cell chemistry this hypothesis of the origin of fluctuations may be found to supply the key to an explanation of the facts discovered by H. E. and E. Frankland Armstrong (1912) with respect to the sporadic dis- tribution of cyanophoric glucoside in herbage plants, such as Lotus corniculatus. Their studies have brought to light the interesting fact that the glucoside may be present in one plant or group of plants and absent from another, and although it may be that climatic conditions may play a part in the phenomenon it seems also probable that this fluctuation is dependent on the genetic constitutions of the individual plants. Again it will be at once evident that the cases enumerated in Table IV, in which giants appear in Fi, are susceptible of explanation on the hypothesis which has been put forward. Thus Crimson King (see Table IV) is itself a fairly massive plant and may be supposed to contain two of the three growth factors. Hence when crossed with a giant it gives an jPj which the expert describes as semi-giant and an F2 (Table IV) composed of 18 giant : 5 non-giant. Thus : F. Keeblk 186 AABBcc X AABBCC yields i\ = AABBCc F, of 3 giant and giant-like forms : 1 non-giant = in 18 plants : 135 giant : 4'5 non-giant as compared with 13 „ : 5 „ found (Table V). In supposing that plants which, in F^, were ranked as semi-giants are liable to be classed with giants in F2 no violence is done to probability ; for in the first place the judgment is a rough judgment and in the second place the habit of inflorescence in Fi is apt to be free, and the flowers borne in such an inflorescence are less likely either to appear or to be gigantic than those which are borne on a more massive flower-stalk. Indeed the records of the behaviour in sub- sequent generations of plants recorded as giants show that not all plants to which gigantism is ascribed with confidence prove to be pure to that character. Finally with respect to the Mendelian phenomena of gigantism it appears reasonable to conclude that gigantism in P. sinensis depends for its full expression on the simultaneous presence of three factors : that pure giants are homozygous for these factors ; that giant-like forms occur when the plant is heterozygous for the C factor ; and that an intergrading series of semi-giant races occui-s in which the grades are represented by appropriate combinations of factors and their "absences." Theoretical Considerations. Numerous considerations, some of no small interest and importance, arise out of the results which have been described in the previous sections ; but of these considerations only few can be discussed in the present paper. First among them is the question concerning the origination of Giant White Queen Star. Is the fact that it arose in course of "selection" of flowers with supernumerary petals a mere coincidence or did the selection process play any part in the liberation of the giant ? If the hypothesis on the nature of fluctuation (see p. 184) be accepted it is evidently susceptible of application in the present case. For a form of P. sinensis of the type of constitution AaBbcc though, for lack of the C factor, it may not produce giants, may produce gametes of various constitutions and these in turn combining in the various ways open to 186 Gigantism in Primula sinensis them may give rise to zygotes characterised by minor peculiarities which are the outcome of the several constitutions. However this may be, the origin of Giant White Queen Star appears to provide an example of the appearance of a " new," dominant character and is noteworthy because of the small number of cases in which this form of evolution has been observed. For, as is well known, the " dropping out " of a factor is common enough in the descent by reduction which cultivated and wild plants are undergoing; whereas the number of known examples of the appearance of new dominant characters (Punnett, 1911) is much fewer and none is known in which the phenomenon has, as it were, been witnessed in a pure strain. Second, the complete infertility of the giant when crossed either with its parent or with other strains and its original relative infertility on self-fertilization are remarkable and suggestive facts. Third and last : the phenomena of gigantism appear to have a bearing on those which concern the origin and nature of certain of our cultivated plants such as fruit trees and shrubs. Thus a culti- vated variety of apple, pear or plum is evidently a giant with respect to its fruit. It may well prove that this cell-gigantism is the origin of all the differences between the large and luscious fruit of the cultivated apple and the astringent puny fruit of the crab. Alter the size of the cell-laboratory and the operations of that laboratory are altered. Events which mark the waning of the life of the small and rapidly maturing cell of the crab may never — for reasons of time or space — occur in the large and slow growing cell of the apple. The character of astringency would seem to have been lost by the dropping out of a factor for that character ; whereas on the view now presented it is only lost because under the new conditions of growth the character cannot appear. How far all or most Mendelian characters depend directly or indirectly on such growth acceleratory and growth inhibitory factors must be left for subsequent consideration. Summary. 1. A giant form of White Queen Star originated from a normal strain of known pedigree. 2. The giant arose in the course of selection-experiments made with plants possessing flowers with supernumerary petals. 3. Histological comparison indicates that the gigantism of the mutant is due to that of its cells. F. Kkeble 187 4. The giant arose suddenly and breeds true. 5. It is now moderately fertile with its own pollen but proves absolutely sterile when crossed with all other varieties (including the parent form) of P. sinensis. 6. Gigantism in P. sinensis is due to three factors and the character is dominant to normal-character. 7. Owing to the number of factors involved in the production of the character of gigantism numerous semi-giant races exist. These races intergmde one with another and hence their existence is not generally recognized. 8. Giants which breed true may be produced by crossing non- giant races of P. sinensis. 9. Fluctuating variations may owe their origin to the heterozygous state of one or more factors in a form from the genetic constitution of which is lacking entirely one of the factors for the production of a Mendelian character. In conclusion, the author has pleasure in expressing his thanks to Miss C. Pellew for her assi.stance during the early part of the experiment, and to Mr G. Coombs both for his kindness in drawing the figures in the text and for much other help. The author has also to add that a portion of the expense incurred in the plant-breeding experiments has been met by a grant from the Royal Society. REFERENCES TO LITERATURE. 1901. De Vries, Hugo. Die Mutationstheorie. Leipzig, Bd. i. 1909 a. Bateson, W. MendePs Principles of Heredity, p. 19. 1909. Cambridge University Press. 1909 a Bateson, W. Op. cit. "Translation of experiments in Plant Hybridiza- tion." By Gregor Mendel, p. 337. 1909. 1909. Gates, R. R. "The Stature and Chromosomes of Oenothera gigas, de Vries." Archiv fur ZeUfonchung, Bd. in. Heft. 4. 1909. Gregort, R. p. "Notes on the Histology of the Giant and Ordinary Form of Primvla ginengi*." Proc. Camb. Phil. Soc. xv. Pt iii. 1909. 1909. Nilsson-Ehle, H. Krevaungsunterguchungen an Hafer und Weizen. Lund. 4. S. 122. 1909. 1910. Keeble, F. and Pellew, C. "The Mode of Inheritance of Stature and Time of Flowering in Peas {Pisum mtivumy Joum. of Genetics, i. 1910. 188 Gigantism in Primula sinensis 1911. Baur, E. Einfuhrung in die experimentelle Vererbungslehre. Berlin (Born- traeger). 1911. 1911. Gilford, Hastings. The Disorders of Post-natal Growth and Development. London. Adlard and Son. 1911. 1911. Gregory, R. P. ^^Ex'periments with PrimtUa sinensis." Joum. of Genetics, I. 1911. 1911. PuNNETT, R. C. Mendelism. (Chap, vii.) London. Macmillan and Co, 1911. 1912. H. E. and E. Frakkland Armstrong. Herbage Studies I. Lotus comiculatus a Cyanophoric Plant. Proc. Roy. Soc. B. 84. 1912. EXPLANATION OF PLATE XL Fig. 1. White Queen Star normal variety. Fig. 2. Giant White Queen Star — a mutant from the normal variety. JOURNAL OF GENETICS, VOL II. NO. 2 PLATE XI be to Volume II NOVEMBER, 1912 No. 3 THE CHROMOSOMES IN THE OOGENESIS AND SPERMATOGENESIS OF PIERIS BRASSICAE, AND IN THE OOGENESIS OF ABRAXAS GROSSULARIATA. By L. DONCASTER, M.A.. Fellow of King's College, Cambridge. In a recent note^ on some stages in the spermatogenesis of Abraxas grossulariata I showed that there is apparently no inequality in the chromosomes of the spermatocyte divisions, corresponding with the hetei'ochromosomes which have been found in other orders of insects. The same result has been found by others in Lepidoptera. I know, however, of no careful study of the oogenesis in this order, and since in Abraxas the sex-limited character is transmitted by the female only to her male offspring, it seemed possible that there might be inequality in the chromosomes of the eggs, corresponding with the male-deter- mining and female-determining eggs, which are shown to exist by the facts of sex-limited inheritance. In Abraxas the chromosome number is large (the reduced number being 28), and in my first attempts to investigate the subject I failed to find oogonial mitotic figures which were suitable for accurate observation. On searching for other Lepi- doptera which might provide more suitable material, I found that the Large White Butterfly {Pieris brassicae) has a much smaller number of chromosomes, with clearer mitotic figures. After following out the available stages in Pieris, the experience so gained, and especially the fact, unknown to me in my earlier attempts with Abraxas, that the oogonial divisions take place chiefly in the larva, enabled me to return to the study of Abraxas with more success, and to work out the earlier stages of oogenesis with some completeness. In the following ' Journal of Genetics, Vol. i. p. 179. Joam. of Gen. ii 14 190 Chromosomes in Pieris and Abraxas account I shall describe first the behaviour of the chromosomes in the oogenesis and spermatogenesis of Pieris brassicae in which the phenomena are more easily followed, and then return to the oogenesis of Abraxas. Pieris brassicae. Ovaries and testes were dissected out of fresh adult larvae and young pupae (1 — 5 days) fixed immediately in Flemming's fluid, cut into sections of about 6/i- in thickness, and stained with Heidenhain's iron haeraatoxylin. Some also were stained with Breinl's process^ for comparison, especially for the study of nucleoli. A point of some interest is the fact that in adult larvae in August and September, from which the imago will not emerge until the following spring, the testes have attained their full size, and contain every stage of spermato- genesis from spermatogonia grouped around a conspicuous Verson's cell in each compartment to practically mature spermatozoa. The ovaries in adult larvae, on the other hand, are extremely small, consisting of four parallel tubes or rather columns of cells showing as yet no division into egg-chambers, and with no deposition of yolk in the eggs. That almost fully formed spermatozoa should exist in the larval testis before the six months' hibernation of the pupa is a somewhat surprising fact. Oogenesis. In the tubes of the ovary — if they may be called tubes when they are in no sense hollow — when the tube is cut longitudinally, a continuous series may be seen from early oogonia through multipli- cation stages to the earlier growth-phases of the oocytes before the deposition of yolk. Since all the stages occur in fairly regular order from the apex of the tube downwards, they are easier to follow than are the corresponding stages of spermatogenesis, for in the testis the cells are grouped in follicles which are very irregularly arranged. An examination of the stages of spermatogenesis, however, shows that they are closely similar to the series found in the egg-tubes. The apex of the tube is packed with small oogonia, of which the nucleus in the resting stage shows a faint reticulum and a conspicuous nucleolus. The latter in favourably-stained cells appears to consist of a double or bilobed chromatin mass associated with a plasmosome, or enclosed in a mass of achromatic material ; not infrequently the two halves of the chromatin mass may lie apart, and sometimes appear compound. Division figures commonly occur in groups; when equatorial 1 Ann. Trap. Med. and Parasitology, Vol. i. 1907, p. 470. L. DONCASTER 191 plates are seen in face, 30 chromosomes are easily counted (Fig. 1). They include eight which are smaller than the rest, and 22 larger chromosomes. The larger ones, however, are not all of exactly equal size, so that it is not easy to separate the smallest of them from the larger members of the eight small ones. The eight small ones usually appear to be graded into two smallest, four rather larger, and two somewhat larger still, but these paii"S cannot always be recognised with certainty. Pairs similar in size often but not always appear near to- gether in the equatorial plate ; this is most conspicuous in respect of the two smallest. Some figures give the appearance of an odd number of small chromosomes, but more careful study suggests that this is due to one member of a pair being seen end-wise, and the other more side- ways. I conclude therefore that there is no evidence for the existence of an unequal pair, and still less for an unpaired chromosome in the female of Pieris hrassicae. Outside the circle of chromosomes in the equatorial plate, two or more small chromosome-like bodies are often visible ; their position and size usually distinguish them without difficulty from true chromosomes. Miss Cook^ has described similar " chromatin granules " in the spermatocytes of Lepidoptera. After the last oogonial division, the nucleus begins to enlarge, the reticulum becomes more clearly visible, and then, apparently suddenly, the "synizesis" stage supervenes. The nucleus now contains a very fine thread or group of threads, tightly coiled on one side of the cavity ; the chromatin nucleolus at this stage is single, small, with sharp outline (Fig. 2). The synizesis condition ceases as suddenly as it began, giving place to nuclei with the chromatin thread in separate segments, much inter- twined, and noticeably thicker than in the preceding stage (Fig. 3)» The transition from this to the following stages is gradual ; the nucleus enlarges, the threads become less interwoven, and finally become ar- ranged not quite regularly but approximately in a meridional manner round the nuclear membrane (Fig. 4). In the earlier stages it is im- possible to count them, but in favourable cells at this stage it is sometimes not difficult to see that there are 14 separate threads, and as the unravelling of the segments is a quite gradual process there are probably 14 at the close of synizesis, although some of the earlier cells give the impression of a larger number (possibly twice as many). The nucleolus during this process has enlarged and finally again become conspicuously double ; occasionally the two parts are separated ; it is 1 Proc Acad. Sei. Philadelphia, 1910, p. 294. 14—2 192 Chromosomes in Pieris and Abraxas now more obviously composed of an achromatic sheath enclosing a double mass of chromatin. There are thus at this stage fourteen chromatin threads and a double nucleolus. The last stage represented in my series is the .contraction of the threads into short thick chro- mosomes, which at the close of the process are conspicuously split (Fig. 5), so that the ordinary chromosomes are no longer clearly dis- tinguishable from the " chromatin nucleolus " which must be regarded as a double, equallj' paired, heterochromosome such as is described by Miss Cook {loc. cit.) in the spermatogenesis of several species. I have not been able to determine with certainty how the doubleness of the thick chromosomes arises ; in Abraxas, and also in Pygaera bucephala, the ovary of which I have examined for comparison, and in which the double chromosomes are most beautifully shown, the appearance strongly suggests that the doubleness is produced by the contraction of looped threads, which break apart at the bend of the loop and give rise to a double rod. Certain exceptional conditions should be mentioned. At various stages from the oogonia onwards degenerating cells, singly or in groups, are not infrequent, in which the nucleus develops into a more or less compact staining mass. The same thing is not infrequent in spermato- genesis of various insects. Among cells shortly after the synizesis stage in one ovary a single nucleus is present in which, in addition to chromatin-nucleolus and plasmosome, there are about 25 short thick chromatic threads like those of the last stage in the production of the 14 double chromosomes de- scribed above. As will be seen below in the description of Abraxas, such cells with the diploid number of shortened chromosomes are abundant in the latter insect, and appear to arise by the more or less complete separation of the chromosomes which have paired in synapsis. In Abraxas these cells probably do not give rise to eggs, but become nutritive or follicle- cells. In Pieris the nuclei of all cells at this stage, whether they will ultimately become eggs or follicle-cells, normally show the reduced chromosome-number; the single cell found with the double number is quite exceptional. Finally it should be mentioned that in cells of the ovarian epi- thelium enclosing the developing oocytes, mitotic figures occur with many more than thirty chromosomes ; a similar reduplication of the chromosomes in the ovarian sheath appears also in Abraxas, and has been seen in other orders of insects. L. DONCASTBR 193 Spermatogenesis. The testis is divided into about three compart- ments, and at one side of each of these is a large Verson cell round which the spermatogonia are packed, not as yet visibly arranged in follicles. Here and there groups of spermatogonia may be found in division, indicating that the grouping which later shows itself by the arrangement in follicles already exists. The equatorial plates of the spermatogonia are so crowded that I have found no case in which accurate observation of the chromosomes is possible ; the best figures merely indicate that the number is about 30. A little distance away from the Verson cell, beyond the zone in which mitoses are found, the early spermatocytes, now clearly grouped in follicles, are seen to pass through phases comparable with those described in the young oocytes. Since the arrangement of the follicles is irregular, and all the cells in any follicle are nearly at the same stage, it is less easy to place the stages in their right order, but by comparison with the oocyte stages no doubt remains that the cells pass through a similar synizesis followed by a stage with long intertwined chromosomes, and then appear to contract into chromatic bodies round the nuclear mem- brane, connected by a fine reticulum. I have not found anything exactly comparable with the short double chromosomes found in the oocytes, but the last stage mentioned, which persists until the sper- matocyte is ready for division, doubtless represents it The chromatin nucleolus is also similar to that of the oocyte at the corresponding stage. During these processes the cells enlarge considerably, and the follicles grow still more rapidly, so that each comes to contain a considerable cavity. In some testes, perhaps occasionally in all, certain follicles and their contained cells fail to grow, and when these cells come to divide in the spermatocyte divisions, the mitotic figures are irregular and very similar to the abnormal divisions which I have described in Abraxas. They are much less frequent in PieHs, and I have not followed the subsequent fate of the spermatids in this form. The normal spermatocyte divisions are remarkably regular and clear, and show with great regularity 15 chromosomes when the equatorial plate is seen in face (Figs. 6, 7). Three of these are always smaller than the rest, and usually one somewhat less small is distinguishable ; these four correspond with the eight smaller chro- mosomes observed in the oogonial equatorial plates. Side views of the metaphase and early anaphase stages of the first division show that the chromosomes divide in the heterotype manner, the separating halves being connected for a time by two strands. 194 Chromosomes in Pier is and Abraxas The second spermatocyte division is easily recognisable by the smaller size of the cells and of the chromosomes (Figs. 8, 9). Fifteen chromosomes can always be seen in well-placed equatorial plates, of which usually four, sometimes only three, and occasionally as many as five, are noticeably smaller than the rest. The two smallest are often, but not always, lying side by side. In some figures, these two small ones are very conspicuous ; in others, often in the same follicle, only one very small one is found. This led me at first to suppose that in the first spermatocyte division there must be an unequal pair of heterochromosomes such as Wilson has described in Lygaeus and Euschistiis, but if it exists, the difference in size between the two members of the pair is not sufficient to reveal itself in a side view of the metaphase and early anaphase. I have examined many such figures with great care, and have never found a dividing chromosome in which the two halves were certainly unequal. In some figures a very slight inequality is suggested occasionally, but in others, where every chromosome is visible, no inequality can be seen. Further, the fact that there may be three or four all equally small, and in other equatorial plates in the same follicle only one (or rarely none at all) which is conspicuous by its small size, makes the interpretation of these difierences as being due to unequal heterochromosomes very doubtful. It may be concluded therefore that in both sexes of Pieris jbrassicae the somatic number of chromosomes is 80 ; these vary in size and include one pair which differ from the rest in the growth-phases in constituting a "chromatin-nucleolus," so resembling heterochromo- somes. The reduced number in both sexes is 15, and although there is a suggestion that the heterochromosomes form an unequal pair in the male, the evidence for this is quite inconclusive, and the appearance is very probably deceptive. Abraxas grossulariata. The material consisted of ovaries removed from larvae varying from somewhat more than half-grown up to nearly full-grown, and treated similarly to those of Pieris. The ovarian tubes are shorter, but the stages follow each other nearly as regularly as in Pieris. The oogonia contain a nucleolus more or less conspicuously double, which consists chiefly if not entirely of chromatin. Among the oogonia, especially near the apex of the tube, groups of cells are constantly found under- going degeneration ; their number is often considerable. Oogonial mitoses, even in rather young larvae, are less numerous L. DONC ASTER 195 than in Pieris, and it was not easy to find examples cut so that the chromosomes could be counted with perfect accuracy. Counts were always made by drawing the chromosome group, not by eye. In the most satisfactory figures the number appears to be 56, i.e. twice the number found in the spermatocyte divisions. Two of these equatorial plates are figured in Figs. 10 and 1 1 ; the only doubt about the number 56 in these cases consists in the facts that in Fig. 10 a pair of chromo- somes (at the left upper edge, apparently consisting of a larger and smaller member) might possibly be two halves of a dividing chromo- some, but the only reason for this suggestion is that they are at slightly different levels; and in Fig. 11 the double body outside the circle at the bottom might conceivably not be a chromosome, for other stained bodies occur in the cytoplasm outside the spindle. Careful examination has convinced me, however, that in fact there are 56 in each figure. The same number has been found in three other figures in which the number 56 is quite clearly seen, and in a further three in which part of the equatorial plate was seen in the next section to the main group, so that a chromosome might possibly be cut so as to appear in two sections, although this is very improbable when they are so small. In a number of other figures it was impossible to decide with certainty between 56 and 55, and in some only 54 were clearly visible. Since in the best figures obtained there is no reasonable doubt that 56 is the true number, and since an even number is certainly present in Pieris brassicae, it may be concluded with some confidence that iu the female Abraxas grossulariata the unreduced number of chromo- somes is 56. With regard to the variety lacticolor I am less certain ; in all my figures of this form counts may be interpreted as 55 or 56, according to whether a double chromosome is regarded as one or two. As de- scribed in my paper on the spermatogenesis, the lacticolor male does not dififer recognisably from the grossulariata ^ in its chromosome group, the spermatocyte number being 28 in each. It is to be ex- pected therefore that the number in the female should not differ from that of grossulariata. That the number in lacticolor is either 55 or 56 is certain ; a final decision on the matter can only be arrived at when more material is available. It should be said that the chromosomes of the oogonial divisions in both forms are not all equal in size ; in some cases two are noticeably larger than the others, but these two are not always recognisably different from the next in size, so that accurate identification of chromosomes is hardly possible. 196 Chi'omosomes in Fieris and Abraxas After the oogonial stage, the nuclei begin to enlarge and pass through a synizesis stage closely similar to that of Pieris, except that the nucleolus is less conspicuous and is sometimes difficult to find. On emerging from this condition the nuclei, now considerably enlarged, contain interlaced chromatic threads, the number of which cannot be accurately determined. The nucleolus at this stage varies somewhat in appearance ; it usually consists of an achromatic mass enclosing a double mass of chromatin, but may be almost or completely divided into two parts, and each half is sometimes seen to contain a compound mass of chromatin. In the rather later stages this com- pound nucleolus, consisting of a number of globules or irregular lumps of chromatin, and usually clearly divided into two more or less separate parts, is very commonly seen in sections from which the stain is rather thoroughly washed out or which are stained with Breinl's stain (Figs. 13 a, 15 a). The individual chromatin globules sometimes appear double, but I doubt whether this is more than accidental. The chromatin threads contract and thicken, but do not in general assume the meridional arrangement often seen in Pieris. They appear to become bent into loops (Fig. 12), and these shorten to form horse- shoe or even ring-like figures (Fig, 13). The nuclei are now so large as to extend through two sections, but careful counts show that the number of the contracted loops is not far from 27. From this stage onwards, or possibly from one somewhat earlier, a differentiation appears to set in among the cells. In the majority, the chromosomes contract still further and appear to break apart at the bend of the loop, giving rise to a double body, the halves of which are commonly completely separated, so that the nucleus contains about 27 pairs of short, thick chromosomes, and an additional larger pair apparently derived from the chromatin-nucleolus. Each member of a pair may itself show signs of doubleness, possibly owing to the precocious appearance of the longi- tudinal split of the next division, traces of which may sometimes be seen in an earlier stage (Fig. 13). In many nuclei the two members of each pair are completely separated, giving the diploid number of small irregular chromosomes scattered in the nucleus (Fig. 14). I was at first inclined to believe that these single chromosomes came together in pairs, giving rise to the stage in which the reduced number of doubles is seen, but a careful study of the various stages has convinced me that in the nuclei here described the paired condition precedes that with the full number of singles. The nuclei just described are the most numerous class at the lower L. DONCASTER 197 end of the egg-tube, but are not, I believe, true oocytes. Among them, fewer in number, are somewhat larger nuclei, which always lie in the middle of the egg-tube, either surrounded by those of the type just described, or extending further down the egg-tube than the latter. I believe these larger nuclei are those which will give rise to the egg-nuclei; the kind described just previously probably belong to the cells of the egg-follicles or the nutritive cells. The large nuclei, which appear to be those of the true oocytes, contain a very large nucleolus and interlaced chromatin threads, which have not contracted into short loops. Their number is difficult to count, but appears to be the haploid number (27) rather than the diploid (54). The nucleolus appears to consist chiefly of chromatin ; it is either double or divided into two separate parts, and each part consists of a number of masses of chromatin apparently embedded in achromatic material, which, however, is not very easy to demonstrate. The irregularity of the chromatin masses composing the two parts of the nucleolus makes it difficult to determine whether the two parts are equal or unequal ; in some nuclei no difference in size is visible, in others one mass is certainly larger than the other. The nucleolus appears rather as a store of chromatin than as a definite chromosome. In addition to the two classes of cells described in the lower region of the ovary-tube, there are cells of varying size with many scattered chromatin granules in the nucleus. These are occasionally found in division, and show about 56 chromosomes ; if the cells described above give rise to the nutritive cells and true oocytes, these last are probably the follicle-cells. In conclusion, then, it appears that Abraxas, like Pieris, has an even number of oogonial chromosomes, with no evidence of an un- equal pair. In the early stages of the meiotic phase two of these give rise to a double chromatin-nucleolus, and the remainder undergo synizesis, from which they emerge in the haploid (reduced) number of chromatin threads. In cells which probably do not become tnie oocytes, these threads then become contracted into short loops, which break at the bend and give rise to the reduced number of pairs of chromosomea The members of the pairs may then separate. In the true oocytes the bivalent threads persist to the latest stage observed — possibly till the prophase of the polar divisions. The halves of the chromatin-nucleolus, though not always identical in size, do not show any constant differences which would justify the assumption that it may be regarded as an unequally paired heterochromosome. The 198 Chromosomes in Pieris and Abraxas chromosomes in the earlier stages of oogenesis therefore do not pro- vide any visible basis for the sex-limited transmission of characters. If Spillman's suggestion be correct, that in the normal grossulariata male there are two G-bearing chromosomes, while in the female one of these is replaced by a sex-chromosome {"X ") which does not bear G (the factor for grossulariata), this is what would be expected ; but since, in the male at least, the variety lacticolor has the same number of chromosomes as grossulariata, the G'-bearing chromosomes would have to be supposed capable of losing the factor G without becoming visibly different. Note. — Since the above was written, I find that Payne (Journ. Morphol. Vol. XXIII, 1912, p. 331) has come to conclusions closely similar to mine about both the constitution of the nucleolus and the origin of the oocytes in Reduviidae ; and Miss P. H, Dederer has published a preliminary note on the maturation of the eggs of Philo- samia cynthia {Biol. Bull, xxill. p. 40, June 1912) in which she finds no dimorphism among the egg- chromosomes. Postscript, Sept. 26, 1912. In ovaries of larvae derived from the 1912 pairings, although I have no perfectly trustworthy figures in individuals derived from the cross lact. X lact, I have several which show 56 chromosomes quite clearly in larvae from the cross gross. $ x lact. cf. Since this cross always gives only lacticolor females, it may be concluded with confidence that the chromosome number in the lacticolor female is not dififereut from that in the grossulariata female. EXPLANATION OF FIGURES. All the figures except 13a were drawn from sections stained with iron-haematoxylin ; for all Zeiss apochromat n. a. 1.40, 3 mm. and oc. 12 were used. In figs., 3, 4, 5, 13a, 15a the shaded area round the nucleolus represents the non- chromatic portion from which chromatin stains are easily washed out. Figs. 1 — 9. Pieris brassicae. 1. Oogonial division, equatorial plate. 2. Synizesis stage. 3. Shortly after synizesis. The upper half of the nucleus is not included in the section. 4. Rather later stage ; complete nucleus with 14 threads and chromatin-nucleolus. 5. Contraction of threads to form double chromosomes ; only 13 of these are visible in this nucleus, the fourteenth is probably hidden by the nucleolus. L. DoNCASrER 199 200 Chromosomes in Pieris and Abraxas 6, 7. Ist spermatocyte equatorial plates; four small chromosomes in fig. 6, three in fig. 7. 8, 9. 2nd spermatocyte equatorial plates : three small chromosomes in fig. 8, only one conspicuously small in fig. 9. Figs. 10 — 15. Abraxas grossulariata. 10, 11. Two oogonial equatorial plates, each with 56 chromosomes. 12. Oocyte after synizesis. Reduced number of chromatin loops. The nucleolus and a number of loops are not included in the section. 13. Slightly later stage ; not all the chromosomes are shown. (26 double chromo- somes and two nucleoli were visible in this nucleus.) 13a. Nucleolus of similar nucleus stained with Breinl. 14. Later stage, probably a nutritive cell. Somatic number of short chromosomes, most of them split, some lying in pairs. (54 chromosomes and two nucleoli were visible in this nucleus.) 15. Oocyte, at about the stage of fig. 14, with reduced number (approximately, not all shown) of chromatin threads. 15a. The two halves of the nucleolus of a nucleus in about the same stage, from which the stain is so far washed out that the chromatin threads were nearly colourless. SOME RECENT WORK ON MUTATION IN MICRO-ORGANISMS. By CLIFFORD DOBELL. Much work has been done in the last few years upon mutation' in several different groups of micro-organisms. This work has been published in many different places, and has been largely carried out in connexion with investigations of a medical nature. The records are therefore somewhat diffuse, and not always easily accessible to the biologist who applies himself mainly to the study of genetics. In the compass of the following few pages I shall endeavour to chronicle — in a somewhat critical spirit — some of the more important observations which have been recently recorded in this branch of biology. I. Mutations in Trypanosomes. In this first section, I shall describe some recent work upon mutation phenomena observed in several species of flagellate Protozoa belonging to the genus Trypanosoma. The mutations may be grouped in two different classes — morphological and physiological. A. Morphological Mutations. In several cases, structural modifications have been induced in Trypanosomes, and found to be permanent and transmissible for a variable number of succeeding divisions. These cases will now be described. ' I Qse this term — as others have ah-eady done — to denote those heritable modifications which have been indaced in various ways in various micro-organisms. I believe that a " matation " in a Trypanosome is essentially the same sort of thing as a " mutation " in a molticellolar organism. Bat I must also point out that I use the words "inheritance," "heritable," and similar terms in the customary manner — applying them to the trans- missible characters of such organisms as Trypanosomes, Bacteria, etc. I do not wish to assert, however, that "inheritance" in Trypanosomes means exactly the same thing as " inheritance " between parent and offspring in sexual multicellular organisms. 202 Mutati07i in Micro- Organisms The Trypanosomes are Protozoa possessing a very remarkable structure. It is necessary, therefore, to recall at the outset the structures present in a typical animal of this sort. This can be most clearly done with the aid of a diagram (Fig. A), an inspection Fig. A. Structure of a typical Trypanosome. (1) Trophonucleus, or chief nucleus. (2) Kinetonucleus, or smaller nucleus, in connexiou with locomotory apparatus. (3) Blepharoplast — a minute basal granule at the root of the flagellum. (4) Undulating membrane, used in locomotion. (5) and (6) Flagel- lum — marginal (to undulating membrane) and free parts. Ant., anterior or flagellar end. Post., posterior or aflagellar end of organism. of which will, I think, make further explanation unnecessary. The terminology of the parts is that of Minchin. To avoid any confusion I should mention that the organ which is called kinetonucleus through- out this paper is called centrosome by Laveran, and blepharoplast^ by the majority of German workers. It should also be added that Try- panosomes reproduce by longitudinal division — both nuclei dividing into two. I will now describe the curious structural changes which have been brought about in certain Trypanosomes, and will begin with the work of Wendelstadt and Fellmer (1910). It has been found by these workers that Trypanosomes — they used two species, Trypanosoma brucei and T. lewisi — which normally live in the blood of certain mammals {e.g. rats) may be inoculated into cold- blooded vertebrates and invertebrates, in which they can live for a certain time. The Trypanosomes used were from well-known strains which had been cultivated in rats and under observation in the laboratory for several years. Trypanosoma brucei was, after some difficulty, passed from the blood of rats into the blood of grass snakes {Tropidonotus natriwy. In the snake's blood, the Trypanosomes become smaller (Fig. B, 2), as compared with the initial forms in the blood of the rat (Fig. B, 1). 1 Frequently — and incorrectly — also written " blepharoblast." ' The authors refer to the snake merely as the " Eingelnatter," but presumably this animal is meant. C. DOBELL 203 Frequently no parasites could be demonstrated microscopically in the snake's blood, although subsequent inoculation experiments proved them to be present. The small forms were apparently formed by the divisions of the original larger forms, and themselves underwent 1. 3. Fig. B. (1) Normal T. brucei in blood of rat. (2) T. hrucei in blood of grass snake- eight days after inocalation. (3) T. brucei, giant form produced by inoculation from grass snake back into rat. (4) T. lewiti, normal form during chronic infection in blood of rat. (5) T. leicisi, form produced by passing the strain from rat through grass snake, then frog, and then back into rat. Fourth rat passage, five days after inoculation. [From Wendelstadt and Fellmer (1910), slightly diagrammatized.] The organisms are all drawn to the same scale, so that the differences in size are correctly shown. division. They showed a slight change in their staining capacity. When these small forms in the snake were inoculated back into rats, they became very large, thus giving rise to a race of giant Trypano- somes (Fig. B, 3). The increased size persisted for many divisions, during passage through several rats\ In later passages, however, the Trypanosomes diminished in size, and returned to their normal dimensions. Closely similar results were obtained by passing the Trypanosomes through tortoises ("europaische Sumpfschildkrote") and lizards ("graue und Smaragdeidechsen ") : but no difference in size was observed after passage through the salamander ("EIrdmolche"). A temporary increase ^ When Trypanosomes (or other micro-organisms) are passed into a fresh host, or culture medium, the new race which thus arises is frequently termed a new "generation " — a vicious usage of the word borrowed from bacteriology. 204 Mutation in Micro- Organisms in the size of the Trypanosomes was also brought about by passing them through beetles {Gychrus rostratus and Aphodius sp.), or through a slug (Aiion impiricorum), and then back again into rats^ Several other passages (axolotl, caterpillars, etc.) were unsuccessfully attempted. Wendelstadt and Fellmer made similar experiments with T. lev/isi. They succeeded in passing this species from the rat through lizards, frogs, and grass snakes. In the cold-blooded host, no Trypanosomes could be found microscopically after inoculation : but inoculation of the blood back into uninfected rats gave rise in them to an infection with parasites of increased size^ When the normal Trypanosomes (Fig. B, 4) from the rat were passed through a snake, or through a snake and then a frog, and then back into a rat, a remarkable modifi- cation was finally produced. The Trypanosomes were not only much larger, but they were also greatly elongated at the aflagellar end^ (Fig. B, 5). No divisions were observed in these forms. Moreover, they showed certain differences in their staining properties as compared with the original forms. It may be added that no results similar to these of Wendelstadt and Fellmer have been recorded by other workers. A much more interesting — because more thoroughly investigated — morphological mutation in Trypanosomes has been discovered by Wer- bitzki (1910). In the course of some researches on the effects of certain organic dyes upon living Trypanosomes, this worker made the following observations. (The researches were carried out in Ehrlich's laboratory, on a strain of T. brucei known as " Nagana ferox," and cul- tivated in mice.) When certain dyes were injected into infected mice, the Trypanosomes (Fig. C, 2) lost their kinetonuclei (Fig. C, 1). The modified Trypanosomes were found to remain permanently devoid of this organ during subsequent divisions. They divided normally and actively, and could be passed in the usual way through other mice by subinoculations. A race of Trypanosomes with a permanent morpho- logical modification has been thus produced. The dyes used successfully 1 The blood containing the Trypanosomes was injected into the body of the invertebrate, which was subsequently ground up in salt solution and the liquid so obtained injected into a rat. It is somewhat surprising that any positive results were obtained by such crude methods. Besides the Trypanosomes, very many other things must have been injected into the rats. 2 The incubation period in the rat was also found to be shortened. 3 Forms similar to these are of constant occurrence during the multiplication period of normal T. lewisi in the rat (Minchin). They have been described as a distinct species (" T. longocaudense ") by Lingard. C. DOBELL 205 by Werbitzki were chiefly substances belonging to the pyronin, acridin, and oxazin groups (mde infra) — the best results having been obtained Fig. C. T. brucei, strain "nagana ferox." (1) Form without kinetonncleos, after treatment with pyronin. (2) Normal form. [From Werbitzki (1910), slightly diagrammatic] with oxazin. The action of the dye upon the Trypanosomes is rapid. In one experiment in which oxazin was injected into a mouse on the second day after infection with the Trypanosomes, the following ob- servations were made : — Hours after injection of dye 1—2 4 6 8 10—12 24 Number of TrypanoBomes without kinetonucleus Isolated specimens 10-12 7„ 25-30 % 40-50% 70—80 % 80— 90°/o The strain containing about 80 % of individuals devoid of a kineto- nucleus, when inoculated into other mice, shows a smaller — but still large — percentage of the modified organisms. By passing this strain through mice 6 — 10 times, however, and treating with oxazin each time, a strain of Trypanosomes in which every individual is devoid of a kinetonucleus has been obtained. This strain remains constant after numerous sub- sequent passages through untreated mice^ The Trypanosomes devoid of kinetonuclei are — as regards motility, general appearance and behaviour, etc, — indistinguishable from normal organisms save in this one feature. Their rate of multiplication is, moreover, unchanged. They show, however, a slight difference in resistance, ^ Kndicke (1911 a) reports that one snch strain has been passed throagh 115 mice, without any treatment, and still retains its morphological peculiarity unaltered, Joam. of Gen, ii 15 206 Mutation in Micro- Or ganisTUS It should be noted that it is the kinetonucleus only which has been removed from these organisms. The blepharoplast (end-knob, or basal granule) and the rest of the locomotory apparatus remain intact^ Werbitzki endeavoured to obtain a race of Trypanosomes with kinetonuclei by further treatment of the race from which this organ had been removed. The parasites were treated with various dyes, and passed through various animals (rats, guinea-pigs, rabbits) ; but the results were not always the same. In one case, passage through 50 animals, and treatment with dyes, left the strain quite unaltered. In another case, however, it was found that 7 % of the Trypanosomes had acquired kinetonuclei at the 16th passage: and this percentage increased during siibsequent passages, until at the 27th practically every individual possessed a kinetonucleus. How the kinetonuclei were " regenerated " was not determined. Microscopically, the individuals of the new race did not differ in any way from normal Trypanosomes. But it was found that their new kinetonuclei were susceptible to the action of drugs which were without effect upon ordinary organisms. For example, the kinetonuclei in the new race were removed by the action of arsacetin — a drug which has no action in this respect on normal Trypanosomes. An important question now arises as to the exact way in which the kinetonucleus is removed from the strain of Trypanosomes which has been subjected to the action of dyes of a certain sort. Werbitzki suggested that its disappearance might be accounted for in three different ways. First, the kinetonucleus might have been destroyed by the dye, or eliminated from the organism ; secondly, it might have fused with the trophonucleus ; thirdly, it is possible that the kineto- nucleus has really not been removed, but its apparent absence is due to the fact that it no longer takes up chromatin stains in the usual way — owing to the action of the drug — and therefore is invisible in microscopic preparations. The second and third suppositions were shown by Werbitzki to be unsupported by any direct evidence. He inclined to the supposition that the kinetonucleus had been destroyed in some way. He figured, moreover, dividing forms of the Trypano- somes in which one daughter individual contained a kinetonucleus, whilst the other contained none. The suggestion therefore seemed justified that the new race arose in this manner — by an irregular distribution of the organ during division. No definite conclusions in this respect were arrived at, however, by Werbitzki. 1 This is stated on the authority of Dr v. Prowazek, to whom the strain was submitted for a careful cytological examination. C. DOBELL 207 Kudicke (1911) made a further attempt to discover how Werbitzki's straia had lost its kiuetonuclei. He says that even in normal races of Trypanosomes — that is, in organisms untreated with dyes — as many as 5 7o of the individuals may lack kinetonuclei. It is therefore possible that the drug selects these organisms : they may be more resistant to the drug, and therefore survive after treatment and so give fise to the new race. Kudicke was unable, however, to discover exactly how the kinetonucleus disappeared. He found that acridin would remove the kinetonucleus from T. levrisi — in a certain percentage of cases — but here again he was unable to decide with certainty how the removal was brought about. Kudicke's work, on the whole, did not show whether the races of Trypanosomes without kinetonuclei were pro- duced by selection, by an irregular division, or in some other way. An important sequel to Werbitzki's work has been furnished by Laveran and Roudsky. Laveran (1911) obtained the"nagana ferox" and " Werbitzki " strains of T. brucei from Ehrlich. He was able to confirm Werbitzki's observations on the structure of the individuals composing these strains. In collaboration with Roudsky (1911), he reinvestigated the action of oxazin upon T. brucei. These workers found that the dye removed the kinetonucleus — as Werbitzki had stated. They were able to extend the investigations, moreover, to seven other species of Trypanosomes (evansi, soudanense, gambiense, diniorphon, pecorum, congolense, lewisi). In all of these, oxazin caused a disappearance of the kinetonucleus : and the peculiarity was trans- mitted hereditarily in subsequent divisions, so that strains were produced in which the individuals were — to a greater or less extent — deprived of kinetonuclei. Laveran and Roudsky (1911, 1911 a) appear to have decided how the kinetonucleus is removed. They have found that when oxazin' is injected into a mouse infected with Trypanosomes, the kinetonuclei of the latter are stained pink or violet with the dye. The rest of the Trypanosome is uncoloured, and it remains actively motile — provided that the dye is not present in such concentration as to kill. The action of the dye can be observed in a drop of infected mouse's blood under the microscope. It seems certain, therefore, that the dye has a special affinity for the kinetonucleus. It can be seen further that the kineto- nuclei which have been stained by the dye — in the living Trypanosomes — dwindle in size, and finally disappear. Laveran and Roudsky ac- cordingly believe that the dyes used have a direct and specific action * Similar results were obtained with acridin. 15— a 208 Mutation in Micro- Organisms upon the kinetonucleus, which they attack and finally remove. They suggest further that the actual destruction of the kinetonucleus is brought about by autoxidation in situ. Certain experiments appear to support this view. It is known that potassium cyanide and alka- loids— when present in very small quantities — retard autoxidation processes m the tissues. Laveran and Roudsky made a number of different preparations of heavily infected mouse blood. To some they added oxazin alone : to others oxazin with minute quantities of KCN or certain alkaloids ^ The results were very striking. Oxazin alone coloured and removed the kinetonuclei of the Trypanosomes (as usual): oxazin + KCN, or oxazin + alkaloid, did not affect the kinetonuclei, which remained quite colourless. It appears certain, from the above observations of Laveran and Roudsky, that the production of a race of Trypanosomes devoid of kinetonuclei by the action of dye-stuffs, is due to the specific action of the dye upon the kinetonucleus. The latter is attacked by the dye — as is shown by its becoming coloured — and then removed, probably by autoxidation. Laveran and Roudsky find no evidence to show that the kinetonucleus is ever removed by an irregular division — as sug- gested by Werbitzki and Kudicke. They suggest that " if in certain cases the kinetonucleus does not divide at the moment of bipartition, it is probably because it is already dead or altered " — through the action of the dye. Werbitzki (1910) found that the only difference — save as regards the nuclei — between his strain and normal 1\ bruc&i was that the former was less resistant to pyronin. Laveran (1911, 1911a) and Laveran and Roudsky (1911) found that the two strains differed in that the Werbitzki strain had an attenuated virulence for laboratory animals. They also found that injections of oxazin caused the appear- ance of giant forms of the Trypanosomes in infected mice {T. brucei, T. evansi, T. soudanense). All the species of Trypanosomes without kinetonuclei appear to possess a diminished virulence. Laveran and Roudsky (1911 a), by imitating Werbitzki's procedure, have now obtained by oxazin injections a strain of T. evansi which has no kinetonuclei and is apparently fixed in this respect. It breeds true in untreated mice. From T. soudanense, however, they have only suc- ceeded in obtaining a race which — at the 50th passage — contains 66 % of individuals without kinetonuclei. ^ The authors do not state which alkaloids they used. C. DOBELL 209 There is a point of considerable interest in connexion with the dyes which bring about the disappearance of the kiuetonucleus in Trypanosomes : for there appears to be a definite relation between the chemical structure of the dye and its action upon the Trypanosome. Werbitzki found that those dyes which destroy the kiuetonucleus are substances belonging to the pyronin, oxazin, and acridin groups (" ani- line " dyes with acid chromophores). These dyes all possess a structure which Ehrlich (1909) calls an orthoquinoid structure. Orthoquinoid substances possess a structure which is essentially thus: That is to say, they consist of two benzene rings united together as shown. Dyes of the pyronin series have the general structure : Those of the acridin series have the structure : Both the pyronins and the acridins are derivatives of diphenylmethane. This is not the case with the dyes of the oxazin group, however, which have the general structure : and are diphenylamine derivatives. 210 Mutation in Micro- Organisms The orthoquinoid linkage appears to be the important thing in these substances. Those drugs with it act upon the kinetonucleus : those without it have no action. (Many dyes and other substances were tried in this respect, e.g. atoxyl, arsenophenylglycin, trypan-red, etc.) It must be noted, however, that dyes with a structure which Ehrlich (1909) terras paraquinoid (e.g. parafuchsin) also have a slight action upon the kinetonucleus. These dyes have the general structure : parafuchsin being a derivative of triphenylmethane with the structure NH2CI When parafuchsin is employed in doses large enough to affect the kinetonucleus, it also injures the rest of the Trypanosome, and finally kills it\ ' The parafuchsin-resistant strain of Ehrlieh, kept at the Speyer Haus, has its kineto- nuclei intact. The strain was produced by acting upon normal T. brucei with increasingly large doses of the dye. C. DOBELL 211 It seems legitimate to conclude, therefore, that dyes with an ortho- quinoid structure have a specific action upon the Trypanosome kineto- nucleus*. They fix themselves to it in some way, and bring about its disappearance. Concerning the Trypanosomes without kinetonuclei there are but a few additional facts of importance to record. These are results of the work of Kudicke (1911 a). He has not found it possible to obtain from the Werbitzki strain — either by drug treatment or transplantation into other animals — a strain in which the kinetonucleus is present once more. (But compare Werbitzki, p. 206, supra.) From certain immunity experiments, he has concluded that the original strain of "nagana ferox" and the Werbitzki strain derived from it, are — as regards immunity reactions — alike. Kudicke has also made some interesting observations on relapse strains of " nagana ferox." He inoculated a mouse with the strain together with trypan-blue^ Four days later, all the Trypanosomes had disappeared from the blood of the mouse. But four days after this, many Trypanosomes were found in the blood — that is, a relapse race arose. Nearly all the individuals of this race were devoid of a kinetonucleus. After passage through a second, and then a third mouse, all the Trypanosomes were without kinetonuclei. They persisted in this condition during 79 subsequent passages. An explanation of this phenomenon was not arrived at, but it seems that something dififerent from what occurs in the case of ortho- quinoid drugs must have happened^ It is perhaps of some interest to recall here — in connexion with the experiments just recorded — certain observations which have been made upon some flagellate Protozoa closely related to the Trypanosomes. Several observers have found occasional individuals which have lost their trophonuclei. Hartmann and Prowazek (Arch. Protistenk. X Bd. 1907) noted, for example, that 5-day cultures of the kala-azar parasite contain individuals which have lost this organ. Similarly, Flu (ibid. XII Bd. 1908) and Berliner (ibid. XV Bd. 1909) describe the occurrence of individuals with a similar defect in Crithidia melophagia and 1 Ehrlich has foand that Trypanosomes which have become resistant to orthoqainoid substances are also resistant to arsenic compounds — a very curious phenomenon. He also found, conversely, that races resistant to arsenic are resistant to the orthoquinoid diphenyl- methane derivatives. ^ Not an orthoquinoid substance, and usually without action upon the kinetonucleus. ' The relapse race may have been formed by selection — a chance individual with no kinetonucleus having been more resistant to the drug than the normal forms, and having survived the injection and given rise to the new race. 212 Mutation in Micro- Organisms Herpetomonas jaculum respectively. It is unfortunate that the origin of these forms is as yet quite unknown ^ Probably they were merely degenerate individuals. B. Physiological Mutations. It has now been known for some years — largely through the work of Ehrlich and his collaborators — that drugs and antibodies may modify profoundly the physiological properties of Trypanosomes. As early as 1907, Ehrlich showed that treatment of Trypanosome-infected animals with atoxyP might cause — under certain conditions — the Trypanosomes to acquire an immunity to the drug. By subjecting the Trypanosomes to the action of minute but increasing quantities of atoxyl or other drugs, Ehrlich has succeeded in obtaining strains of the parasites which are highly resistant^ to these poisons. In 1909 he stated that he had pro- duced an arsenic-resistant strain of T. brucei, which had, in the course of three years, undergone passages through some four hundred untreated animals — without any loss of resistance to arsenic. Many similar ob- servations have since been recorded, so that it may now be stated as a fact that physiologically modified races of Trypanosomes can be made by artificial means from the races which occur normally in nature. Mesnil and Brimont (1908) also succeeded in obtaining a race of Trypanosomes resistant to atoxyl. But they pointed out that the re- sistance was only manifested " in a given organism " {i.e. host). More definite in this respect, however, were the statements of Breinl and Nierenstein (1908)^ From inoculation experiments, they concluded that a Trypanosome's resistance to the drug is manifested in that ^ It should be remembered that Trypanosomes which have grown in artificial culture media frequently display morphological peculiarities — as regards size, shape, relative position of nuclei, etc. These modifications are, however, transitory : they do not persist after the organisms have been reinoculated into the animals which are their normal hosts. 2 Atoxyl is sodium arsanilate — the Na salt of jj-aminophenyl-arsenic acid (Ehrlich and Bertheim). * A Trypanosome which tolerates the action of a drug is generally said to be " fast " to the drug in question, e.g. a Trypanosome which has been rendered tolerant to atoxyl or other organic arsenicals is spoken of as "arsenic- fast." The word "fast" has, however, an older and very different usage in bacteriology. For instance, tubercle Bacilli — and certain others — are called "acid-fast." This does not mean that the living organisms tolerate, or are resistant to, acids : it means that dead organisms when stained with carbol-fuchsin are stained "fast" (in the dyer's sense) against mineral acids. I therefore prefer to use "resistant" rather than "fast" when discussing the phenomena of living Trypanosomes. * These workers, it may be noted, give an incorrect account of the results of Mesnil and Brimont. C. DOBELL 213 species of host animal alone in which it was acquired. Fur example, they found that T. brucei in donkeys became resistant to atoxyl after injections of the drug. Transplanted into rats, however, it rapidly lost its resistance and became susceptible again. Races of Trypanosomes with a changed virulence, produced by passages through a different host, have several times been recorded. Fellmer (1907), for example, stated that the virulence of T. brucei was diminished by passage through the hedgehog. The structure of the parasites was also stated to be modified by a sojourn of the race in this animal. Fellmer's experiments were repeated by Gonder and Sieber (1909), who used both T. brucei and T. equiperdum. They com- pletely failed, however, to produce any change in either the virulence or the structure of these Trypanosomes in this way. Wendelstadt (1909) and Wendelstadt and Fellmer (1909, 1910) also announced that the passage of T. brucei and T. leivisi through cold- blooded vertebrates — to which reference has already been made — greatly modified their virulence. They found, for example, that T. ieivisi when passed through the grass snake becomes modified into a race which is pathogenic for rats — in which the infection is normally harmless. Inoculation of Trypanosomes from the snake back into the rat kills the latter. Twenty-four passages with a similar result were thus made. Laveran and Pettit (1909) repeated these experiments. They injected both T. lewisi and T. evansi from rats into snakes, and then back into clean rats. But they failed entirely to produce any change in the virulence of the Trypanosomes. They state, moreover, that the blood of the snake is very toxic to rat« — which may account for the results of Wendelstadt and Fellmer. It seems, therefore, that these experiments in which changes of virulence are said to have been produced in Trypanosomes should be regarded with considerable scepticism for the presents Levaditi and Mutermilch (1909) found that they could produce races of T. brucei which were resistant to certain antibodies. Later, Levaditi, in collaboration with Twort (1911), has shown that a race of T. br^ucei can be made which is resistant to the toxin produced by Bacillus subtilis — a substance which is usually very toxic to T. brucei. If a normal race of this Trypanosome (from the blood of the mouse) is subjected in vitro — even for only a few minutes — to the action of 1 It should be recalled, however, that the Werbitzki races of Trypanosomes have undergone a diminution in their virulence — a fact which appears to be established (Laveran and Roudsky). 214 Mutation in Micro- Organisms the toxin, it is found, after reinoculation into a mouse, to have acquired a marked resistance to it. A sM6^t7iS-toxin-resistant strain of T. hrucei has been thus produced which remains as such during subsequent animal passages — that is, the acquired resistance of the Trypanosomes is trans- mitted hereditarily. Most important and extensive work in this direction has been recently published by Gonder (1911), whose results may now be con- sidered in some detail. The work was carried out in Ehrlich's laboratory, where it was begun by Werbitzki. In his experiments, Gonder used two strains of T. lewisi. These were, first, a strain from wild rats, grown in tame laboratory animals and susceptible to arsenic. An injection of 0"1 gm. of arsenophenyl- glycin per kilogram body-weight of rat sufficed to kill all the Trypano- somes in its blood. This strain, after numerous passages through normal rats, always remained susceptible to arsenic. The second strain was one which had been made arsenic-resistant \ It was made by accustoming the Trypanosomes to minute but gradually increasing doses of arsenophenylglycin. Finally — after two years — a strain of T. lewisi was obtained which was resistant to the drug to such an extent that it was unaffected by injections of 0*2 gm. per kilo. Passage of this strain through untreated rats showed that the arsenic -resistance had become fixed, and was transmitted hereditarily. At the twentieth passage, the resistance was unchanged I As regards structure, and behaviour in other ways, the Trypanosomes were found to be indis- tinguishable from the normal race. Gonder found that both the non-resistant and the resistant race could be transmitted from rat to rat by the rat louse, Haematopinus spinulosus — which is supposed by some workers to be the usual inter- mediate host of T. lewisi in nature. The incubation period in the louse was found to be 15 — 17 days in the case of the normal race (3 rats); 25 — 30 days (6 rats) in the case of the arsenic-resistant race. The two races of Trypanosomes were then tested as regards their resistance to arsenic. And the results showed that all the Trypanosomes — whether they had previously been resistant or not — were non-resistant to arsenic after development in the body of the louse. 1 Arsenic-resistant strains of Trypanosomes can be rapidly produced by treatment with a number of different organic arsenic compounds, and also by the action of dyes of an ortboquinoid type (Ehrlich, 1911). 2 Similar arsenic-resistant races of other Trypanosomes had, of coarse, been previously produced by Ehrlich and others. C. DOBELL 215 Id two cases, where mechaDical transmission by the louse was believed to have occurred — that is, where no development of the Trypanosomes in the louse intervened — it was found that the transmitted Trypanosomes had retained their power of resisting arsenic. The incubation period in the louse in these cases was only 5 days. By daily injecting emulsions made from the bodies of lice — in which Trypanosomes were developing — into uninfected rats, Gonder was able to determine how long the arsenic-resistance of the Trypanosomes persists. He found that it persists for 12 days in the body of the louse. After this period, the Trypanosomes lose their resistance to arsenic, and become normal. Cultures of normal and arsenic-resistant races of T. leivisi were made in artificial medial Both races behaved exactly alike. The non-resistant races, when reinoculated into rats, were still non-re- sistant: the arsenic-resistant races remained arsenic-resistant. Both races underwent similar structural changes in the cultures — being gradually converted into Grithidia-\\ke forms in the course of some 3 months. These forms, when injected back into rats, assumed the normal Trypanosome form once more — the incubation period being 3 — 11 days. Multiplication occurred in the artificial cultures. Ehrlich (1911) and Gonder (1911) have interpreted the foregoing facts in the following way. They suppose that the development which T. levnsi undergoes in the louse constitutes a sexual cycle in the life- history of this species. They suppose that the resistance to arsenic, which the Trypanosomes have been made to acquire, persists only so long as the asexual cycle endures — that is, during the period when the Trypanosomes are in the blood of the rat, or in artificial culture media. When the sexual cycle takes place in the body of the louse, the acquired resistance of the race is lost, and the individuals revert to their original non-resistant condition. The "acquired character" is thus "inherited" in asexual reproduction only*. This extremely interesting and suggestive idea cannot be regarded at present as anything more than a hypothesis. For in the first place, • T. lewisi was first successfully cultivated in an artificial medium by Novy and MacNeal, in 1903. Since then, many other workers have succeeded in cultivating a number of other species. ' Far-reaching conclusions regarding " the inheritance of acquired characters " can be drawn from these experiments only by those who are content with words and unable or anwilling to analyse the facts. 216 Mutation in Micro- Organisms it has not been proved that the louse is the normal intermediate host of T. lewisi. There is much evidence to show that it is the flea, and not the louse, which is the normal carrier of the Trypanosome from rat to rat : though the louse may occasionally be the means of infection. Secondly, it has never yet been proved that the development which the Trypanosomes undergo in the gut of the louse is a sexual development. A sexual cycle in the louse was first described by Prowazek, and has since been alleged to occur by Baldrey, Rodenwaldt, and others, whose results Gonder says he can confirm " almost entirely." To prove the existence of a sexual cycle in the louse, however, something more than the arbitrary seriation of certain stained specimens is requisite. Until the publication of more convincing evidence — derived from a study of the living organisms, and from careful cytological research — it is not justifiable to conclude that conjugation of the Trypanosomes occurs in the body of the louse. It is, moreover, obvious that Gonder's own results cannot be held to prove that conjugation occurs in the louse : his interpretation is based on the supposition that conjugation does occur. And there is really no reason why the development — which the Trypanosomes appear undoubtedly to undergo — in the louse, should be regarded as necessarily of a sexual nature. Since it has been found (Mesnil and Brimont [1908], Breinl and Nierenstein [1908]) that the resistance of a race of Trypanosomes to arsenic is manifested only so long as the race remains in a given host, it is not impossible that Gonder's results are explicable on the same principle. T. lewisi may remain arsenic-resistant so long as it con- tinues in the blood of the rat, or in an artificial medium : but a change of host {i.e. from rat to louse) may abolish the resistance — just as T. brucei, arsenic-resistant in donkeys, becomes non-resistant when trans- planted into rats. (Of. p. 213.) If one substance can bring about arsenic-resistance, it is at least conceivable that another substance can remove it. And it is possible that the body of the louse may furnish such a substance. At all events, there is no need to assume the existence of sexual phenomena to account for the results of the experiments, Ehrlich and his followers regard resistance to drugs or sera as a direct consequence of the action of the substances in question upon the living protoplasm. That is to say, they suppose that when a Trypanosome is treated with a minute quantity of arsenic, its proto- plasm becomes changed in such a way as to make it resist the drug when applied subsequently. New races of Trypanosomes are thus C. DOBKLL 217 supposed to be directly produced by a modification of the individuals of the old race'. Erhlich's views in this respect are not shared by some other workers. Levaditi, with Mutermilch (1909) and Twort (1911), interpreted his own results as due to selection by the poison employed. The toxin of B. sttbtilis was found to kill or affect many Trypanosomes, when observed in vitro. And it was concluded that " certain races of Trypanosomes, considered as homogeneous, are only, in reality, a mixture of a large number of individuals endowed with unequal susceptibility towards a given trypanocidal poison." (Levaditi and Twort [1911].) In other words, resistant pure lines may be formed from a mixed population by the selective action of a poison — onl}' those naturally most tolerant Trypanosomes being able to survive, and to perpetuate the race. It therefore seems uncertain how resistant races of Trypanosomes arise. It is possible, however, that both a direct action of the drug and an indirect selection by it play a part in their formation. In conclusiou the more important results noticed in the foregoing pages may be very briefly summarized. I will limit myself to only those conclusions which appear to me to be justified at the present moment. (A) It has been stated that the passage of certain Trypanosomes, which normally occur in mammals, through cold-blooded vertebrates and certain invertebrates, causes them to undergo certain structural changes which persist during subsequent divisions (VVendelstadt and Fellmer). This work has not yet been confirmed. It has further been stated (Werbitzki) and confirmed (Laveran and Roudsky, Kudicke) that certain dyes can destroy a definite organ (kinetonucleus) in a Trypanosome, without killing or injuring it or impairing its power of propagation. Thus new races of Trypanosomes may be produced which completely lack this organ. It has, moreover, been rendered highly probable that the dyes which have this power possess a certain chemical structure (orthoquinoid substances of Ehrlich): * According to Ehrlich (1911), resistant races of Trypanosomes are of two quite different sorts: (1) Serom-resistant, i.e. resistant to specific antibodies; (2) Chemo-resistant, i.e. resistant to various chemicals. Such races are supposed to arise in different ways. In the terms of Ehrlich's theory, a serum-resistant race is formed by the serum causing a certain receptor (nutriceptor) to disappear, when it is replaced by an altogether new kind of receptor. A chemo-resistant race, on the other hand, is produced, not by the replace- ment of one receptor by another, but by the diminution (" Uerabminderung '') of a certain chemical function. 218 Mutation in Micro- Organisms and that the dyes have a specific action upon the kinetonucleus — but upon no other organ in the Trypanosome — and bring about its destruction by autoxidation (Laveran and Roudsky). New races of Trypanosomes are thus produced by modifying the individuals of the old — not by selection. (B) Races of Trypanosomes without kinetonuclei possess a lowered virulence (Werbitzki, Laveran and Roudsky). By the action of various drugs and antibodies, races of Trypanosomes may be obtained which are resistant to these substances (Ehrlicb, Mesnil and Brimont, Breinl and Nierenstein, Levaditi and Twort, etc.). These races subsequently breed true — though it may be a necessary condition of this that they be kept in the same sort of host as that in which they originally acquired their resistance. Races of Trypanosomes with a changed virulence are said to be pro- duced by passage through certain animals (Wendelstadt and Fellmer) : but this has been denied (Gonder and Sieber, Laveran and Pettit). By treating T. lewisi with arsenophenylglycin, a race may be obtained which is resistant to this drug. This race breeds true — retaining its resistance during numerous passages through untreated rats. Resistant and non-resistant races remain unchanged, as regards this character, when grown in artificial cultures. When the resistant race undergoes a development in the louse — the exact nature of which is not determined, though it is possibly sexual — resistance is gradually lost, and the race returns to the original non-resistant condition (Gonder). It has not been definitely determined whether resistance is brought about by the direct action of the poison on the living Trypanosome (Ehrlich, etc.), or whether it is the result of selection (Levaditi, etc.). That some of the observations noticed in the course of this review are of great interest, I think nobody would deny. And that they may lead to a better comprehension of the phenomenon of mutation in general is at least possible. In his Dresden address in 1911 Ehrlich said: "...Aber, meine Herren, in der Natur ist nichts spontan, alles hat seine Ursache, und wenn es sich urn biologische Fragen handelt, meistens eine chemische Ursache.... So glaube ich, dass gerade diese Studien an Parasiten, an klinstlich herbeigefiihrten Mutationen durch bestimmte biologische Eingriffe, deren Mechanismus genau erklarbar ist, uns auch ein belles Licht iiber die so dunklen Fragen der Mutation iiberhaupt bringen werden." (Ehrlich [1911], p. 95.) Though all may C. DOBBLL 219 not take so confident and hopeful a view, this expression of opinion is noteworthy, and indicates the vast possibilities which the future still holds for one bmnch of biological research. LITERATURE. 1908. Bbeinl, a. and Nierenstein, M. "Weitere Beobachtungen iiber Atoxyl- festigkeit der Trypanosomen." Deutscli. med. Wochengchr. No. 27. 1907. Ehrlich, P. "ChemotherapeutischeTrypanosomenstudien." Berliner klin. Wochenschr. XLIV. p. 233, etc. 1909. . " Ueber die neuesten Ergebnisse auf dem Gtebiete der Trypanosomen- forschung." Arch. Schiffs- u. Trapenhyg. Beih. vi. p. 91. 1911. . "Ueber Chemotherapie." Ber. iib. 5 Tagung d. Freien Vereinig. f. Mikrobiol. i. d. Internat. Hygiene- Austell ung Dresden, in : C. B. Bakt. Beil. z. Abt. i. Bd. L. (Ref.), p. 94. 1907. Fellmer, T. " Veranderungen an Nagacatrypanosomen durch Igelpassaga" C. B. Bakt. I. Abt. (Orig.), Bd. xlv. p. 512. 1911. Goxder, R. " Untersuchungen iiber arzneifeste Mikroorganismen. I. Try- panosoma leicisi." C. B. Bakt. i. Abt. (Orig.), Bd. lxi. p. 102. 1909. and Sieber, H. " Experimentelle Untersuchungen iiber Trypano- somen." C. B. Bakt. I. Abt. (Orig.), Bd. XLix. p. 321. 1911. KuDiCKE, R " Die Wirkung orthochinoider Substanzen auf Rattentrypano- somen." C. B. Bakt. i. Abt. (Orig.), Bd. Lix. p. 182. 1911a. . "Beitrage zur Biologie der Trypanosomen." C. B. Bakt. i. Abt. (Orig.), Bd. LXI. p. 113. 1911. Laveran, A- " Contribution k I'etude du Trypanosoma hrucei sans bleph- aroplaste de Werbitzki." Bull. Sac. Path. exot. No. 4, p. 233. 1911 A. . " Au sujet de Trypanosoma brucei sans blepharoplaste de Werbitzki." Bull. Soc. Path. exot. No. 5, p. 273. 1909. and Pettit, A. " La virulence des trypanosomes des Mammiferes peut-elle etre modifiee aprfes passage par des Vertebres h, sang froid ? " C. R. Acad. Sci. Paris, Tom. CXLIX. p. 329. 1911. and Roddsky, D. "Au sujet de Taction de I'oxazine (chlorure de triaminophenazoxonium) sur les Trypanosomes." C. R. Acad. Sci. Paris, Tom. cuiL p. 226. 1911 A. . " Au sujet de Taction de Toxazine (chlorure de triaminophen- azoxonium) et de Takridine (diphdnylmethane) sur les Trypanosomes." C. R. Acad. Sci. Paris, Tom. CLiii. p. 916. 1909. Levaditi, C. and Mdtermilch, S. " Le m^canisme de la creation des varidtes de Trypanosomes resistants aux anticorps." C. R. Soc. Biol. Tom. Lxvii. p. 49. 220 Mutation in Micro- Organisms 1911. Levaditi, C. and Twort, C. "Sur la trypanotoxin du Bacillus suhtilis, etc." A series of seven papers, in C. R. Soc. Biol. Vol. lxx. pp. 645, 753, 799, 927, 962, 1024, and Vol. Lxxi. p. 127. 1908. Mesnil, F. and Brimont, E. " Sur les propridtds des races de trypancsomes r^sistantes aux medicaments." Ann. Inst. Pasteur, Tom. xxii. p. 856. 1909. Wendelstadt, H. " Ueber Form- und Virulenzanderungen von Trypano- somen durch Kaltbluterpassage." Med. Klin. Nr. 16, p. 608. 1909. and Fellmer, T. " Einwirkung von Kaltbliiterpassagen anf Nagana- und Lewisi-Trypanosomen." Zeitschr. f. Immunitdtsforsch. Bd. ill. p. 422. 1910. . " Einwirkung von Kaltbliiterpassagen auf Nagana- und Lewisi- Trypanosomen. II. Mitteilung." Zeitschr. f. Immunitiitsforsch. Bd. v. p. 337. 1910. Werbitzki, F. W. " Ueber blepharoblastlose Trypanosomen." C. B. Bakt. I. Abt. (Orig.), Bd. Liii. p. 303. INHERITANCE OF COAT-COLOUR IN RABBITS. By R C. PUNNETT, M.A., F.R.S. CONTENTS. PAOE Introduction 221 General scheme of the experiments 222 The Fi generation 223 The F2 generation . 223 Agoati from black x black 225 The Hypothesis 227 The F3 generation 229 The synthesis of agoati- bearing blacks .... 233 The test of the coupling between D and E . . . 234 The chocolate series ........ 235 The Himalayan pattern 236 Introduction. When the following experiments were started in 1907 we were already familiar, through the work of Castle (1) and Hurst (5). with certain phenomena in the inheritance of coat-colour and pattern among rabbits. These investigators had shewn that the wild grey, or agouti, is dominant to black, and that these two full colours are respectively dominant to the two dilute forms, yellow and tortoise (= the " sooty- yellow " of Castle). Moreover with regard to pattern Hurst had published a brief statement (6) to the effect that Dutch marking is recessive to self-colour, the heterozygote being variably marked, and that the Himalayan pattern is recessive to the self-coloured form. Castle also(l) had given an account of a few experiments with the Himalayan in which this form was shewn to be dominant to the pure albino. More recently (1909) Castle has published a general account (3) of the colour varieties in the rabbit that have so far been analysed. With that account my own work is in general agreement, and in so far as is possible I have adopted his system of nomenclature for the various factors concerned*. ' Of the four colour varieties agouti, yellow, black, and tortoise, the two former are regarded as containing the agouti factor A which is absent from the black and the tortoise. The agouti and black again differ from the yellow and the tortoise in containing a factor E for the extension of the pigment. Melanic pigment occurs also in the tortoise and the yellow but in much smaller amount and is chiefly localised in the nose, ears, tail, and feet. Jonm. of Gen. 11 16 222 Coat-Colour in Rabbits General Scheme of the Experiments. My own experiments were started with the idea of investigating more fully the genetics of the Dutch and Himalayan patterns, but they had not proceeded far when it became evident that certain phenomena in connection with the inheritance of colour were unlike any hitherto met with, and promised results of unusual interest. It is with these colour phenomena. that the present paper is concerned. At the same time I can confirm Hurst's statement as to the recessive nature of the Himalayan pattern, though with regard to the nature of the Dutch marking and .its relation to the self-coloured form I am inclined to think that the matter is more complicated than his account appears to imply. Experiments on the inheritance of coat-pattern are still in progress and 1 hope to publish them when more complete. Pattern and colour appear however to be quite independent of one another and in the present paper the results will be treated solely from the standpoint of colour. The subjoined scheme provides a general view of the experiments. The qualitative result of the various matings is alone indicated. The quantitative results will be found in the various tables to which reference is given. Fi Fi^ Fz Tort. X Him. Him. x Yel. Yel. x Him. [?7] Blk. Table II {in part) \ Blk. Tort. Him. [c?5] Blk. X Blk. x Blk. [i 11 X >i 13 8 4 2 6 >> 12 X >> 13 8 3 — 5 >» 16 X >> 10 4 2 1 6 >> 27 X >> 12 3 2 1 3 Totals ... 57 27 15 5 29 Expectation 56-0 18-8 18-8 6-2 33-2 Fi $ 10 was also crossed with the Himalayan ^ and gave 2 agouti, 1 black, and 5 Himalayans, expectation here being 2 agouti, 2 black, and 4 Himalayans. The F2 generation from the F^ blacks. Though the F^ blacks originated from two different sorts of cross, viz. from yellow x Himalayan as well as from tortoise x Himalayan, they should nevertheless all behave similarly. For there is no reason to suppose that there is any difference between a tortoise gamete TABLE IL Male 35 Black F^ Black Fi ? ? ex Tortoise x Himalayan No. of ^ Female Blk. (22 8 23 — 24 — 25 — 26 — 45 — 47 — 53 — 56 — 106 — Tort. Him. 4 2 Male 20 Male 74 Male 116 Black f, Tortoise Himalayan Blk. Tort. Him. Blk. Tort. Blk. Him. — — 6 4 4 3 — 4 5 2 2 Black Fi ? 9 ex Yellow X Himalayan 17 37 39 43 V102 2 — Totals Expectation 10 16 26 : 8 : 12 27-7 : 6-7 : 12-<5 19 25 22 22 10 10 10 10 R. C. PUNNETT 225 produced by a tortoise and one produced by a heterozygous yellow. And with a single exception, which will subsequently be dealt with more fully, the breeding results indicate that all the blacks are similar. These results are set out in Table II which shews not only the matiugs between the F^ animals but also the effect of crossing them with pure tortoise (<^ 74) and pure Himalayan {^ 116). All the results accord closely with expectation. Agouti from Black x Black. So far the experiments had merely served to confirm the work of previous writers and offered nothing of novelty beyond the fact that the black of the Himalayan rabbit can behave like the black of the ordinary self-coloured. All the F^ blacks hitherto dealt with were the progeny of a yellow or a tortoise doe by a Himalayan buck (139 13 12 18 10 12 4 6 12 2 1 3 1 (5) (3) (-) (4) (3) (4) (-) (- (3 (- (1 (- (- (- Totals 73 29 B. The Agoutis. Of the Fi agoutis five does and two bucks were mated together in the way indicated in Table V. The results shew that all were hetero- TABLE V. Male B 104 > Male D12S ?£ 146 ?fil88 $£217 ?C 55 ?C102 Agouti 9 19 12 Yellow 5 2 4 Black 3 6 3 Hlma- Tortoise laran Agouti Yellow Black Hima- Tort191____ 421— — — — — 4 2 1 — D 237— — — — 555— — — — — 6 1 3 1 £35____ 563— — — — — 4 2 2 1 22 9 4 3 49 43 23 9 2 2 2 — 23 13 10 6 Black Agouti-Black ? « Agouti Totals 96 67 18 39 11 7 Amended totals ... 107 74 — 39 Expectation ... 96-25 82-5 41-25 In order to test the agouti-blacks further with respect to the factor E both does and bucks were crossed with F^ heterozygous yellows, and the does also with pure tortoise. Since the gametes of the agouti-black are DEA, DEa, dEA, dEa, and those of the F2 yellow deA, dea, the expected result of such a cross is given in Fig. 3. The expectation is blacks and agoutis only in the ratio 5 : 3. Tables VII and VIII shew that only blacks and agoutis resulted from such a cross, the actual numbers being 69 black and 33 agouti where the expectation was 64 black and 38 agouti. ^ Since the above was written an F3 black ? made in this way has been shewn to be of the constitution DDEEJIa.. Crossed with a buck of the constitution ddSSaa she has given only blacks and agouti-blacks. Among the 14 young so bred from her there was no agouti. * The rabbits placed in this category were blacks which died too early to determine whether they were true blacks or agouti-blacks. In making the amended total they have been distributed pro rata among these two classes. 232 Coat-Colour in Rabbits DBA DBa dEA dBa d«A deA deA deA Black Black Agouti Agouti DBA DBa dBA dBa dea dea dea dea Black Black Agouti Black Fig. 3. TABLE VII. Fz Agouti-black Males Male C 32 Male D 128 i^2 ? ¥ Yellow (heterozygous) Agouti Black 4 Agouti 2 1 3 Black 5 3 6 4 3 Male D 122 Agouti Black 3 6 7 21 Totals ... 24 Agouti 48 Black Expectation 27 ,, 45 ,, 14 21 TABLE VIII. Male E 31 (heterozygous yellow) Male 74 (pure tortoise) No. of Female Agouti Black Agouti Black .B 76 B 151 — — 2 6 — — 2 4 £191 — 5 1 6 C144 — — 2 10 D 20 4 4 — — Fi Agouti-black ? ? \ D 21 — 2 — — D133 3 4 — — D145 2 6 — — D156 — — 3 3 1)191 — — 1 14 D287 — — 2 4 \E 35 — — 2 4 Totals 9 21 15 51 Expectation 11-25 18-75 16-5 49-5 R C. PUNNKTT 233 Eight of the agouti-black does were also crossed with a pure tortoise buck {^ 74), and the numbers obtained, 51 black and 15 agouti, tally closely with the expected 3 : 1 ratio. It may therefore be fairly claimed that the constitution of the F, generation from <^ 28 as tested by further breeding from them is in accordance with the hypothesis framed above. The Synthesis of agouti-hearing blacks. So far the evidence that a rabbit, visibly pure black, can carry the agouti factor rests entirely upon a single individual — the ^i ^^ 28. But if our hypothesis is correct, there should be no difficulty in synthe- sising the agouti-bearing black from the material to hand, and as a matter of fact this has actually been done within the past year. Reference to Fig. .3 (p. 232) shews that of the five blacks resulting from the mating of agouti-black with yellow three carry the factor A, two being heterozygous and one homozygous for that factor. Neither of the two remaining blacks carries A. The simplest way to find these blacks would be to test them by crossing with a pure black of the constitution ddEEaa. As however I had not such an animal ready to hand, I used for testing purposes a chocolate ^^ which had previously been shewn to be homozygous for E, but to contain neither D nor A\ In this way 13 black does ex agouti-black x yellow have been tested, with the result that three proved to be homozygous for A, four proved to be heterozygous, while in the remaining six A was absent. The details are given in Table IX. TABLE IX. X Male « 31 X ChocoUte male (ddEEajt) (heterozygous yellow = See notes to the illustrations for evidence of a third pair of hair factors. 258 Inheritance in the Groundsel while others are either (2) wore, or (3) less hairy. Arranging the plants in three groups accordingly, and commencing our analysis with the F^ generations of three RR plants (probably the result of an accidental cross of lanuginosus by erectus, radiatus (see p. 247)), we get the following result : TABLE VIII. Cross Ray Character in Fi No. of Plants Types Exp. UH Hh hh Exp. 56 Lanuginosus x erectus, radiatus ? RR 108 21 51 35 Exp. 57 Lanuginosus x erectus, radiatus ? RR 166 45 79 42 Exp. 58 Lanuginosus X erectus, radiatus? RR 1-60 41 78 41 Totals ... J Found 434 107 208 118 Expected 434 108 217 108 Segregation apparently takes place normally in each of these three cases. If we now examine in the same way the F2 generations of the three NR types of known parentage, we get the result shewn in Exp. Cross Exp. 23 Lanuginosus x praecox Exp. 25 Lanuginosus x erectus Exp. 27 Lanuginosus x multicaulis TABLE IX. Ray No. of Plants lypes in Fi "HH Hh hh X Rr 84 22 47 15 i Rr 88 19 43 26 lulis Rr 203 61 79 63 The result of Exp. 27 is clearly aberrant. Deferring its con- sideration, we may summarize the results in the other five experiments, thus : — Types « Exp. No. of plants HH Hh hh Nos. 56, 57 and 58 434 108 208 118 Nos. 23 and 25 172 41 90 41 ^ , J ( Found 606 149 298 159 ^^^ \ Expected 606 151 303 151 No thoroughly satisfactory explanation can be given of the result in Exp. 27. Not only are the numbers very aberrant, but the mode of obtaining them is liable to criticism. All the HH plants were certainly more hairy than the original F^ heterozygote. There were certainly 61 HH plants present. But all the other plants had some hair, and at the first examination only 20 were marked as hh. Ex- tracting these and re-examining the remainder, this was found to A. H. Trow 259 include apparently two types more and less hairy respectively; the less hairy were added to the hh group, and the more hairy constituted the presumed Hh heterozygotes. Under these circumstances, it should be understood that the result shews little more than that segregation actually occurs in this case. The aberration is probably due to trans- gressive variability, which makes it impossible to fix exactly the limits of the three types. A nearer approximation is gained by applying the law of dominance, but this procedure reveals nothing new. All groundsels, even the most glabrous types, vary a little in hairiness, and in such sunny weather as that of last summer (1911) to a con- siderable extent. It is possible that the multicaulis strain introduced an unknown disturbing factor. We may now turn our attention to the relation of the factors for rays and hairiness when they act together. Lanuginosus {RRHH) X praecox (rrhh), erectus (rrhh) or multicaulis (rrhh) might be expected to produce nine types of plants in the F, generation. Table X shews both the expectation and what was actually found. TABLE X. 9 Types No. of plants flowered RR Rr rr Exp. Cton HH Hh ~hh ^HH Hh ~hh HH Hh hh Exp. 23 Lannginosns x | 83 Found 13! 10 3! 6 23 6 2! 14 6! praecox [ — Expected 5 10 5 10 21 10 5 10 5 Exp. 25 Lanuginosus X ) 88 Found 14! 8 1! 4 25 14 1! 10 111 erectus j" — Expected 5 11 5 11 22 11 5 11 5 Exp. 27 Lanuginosus X \ 202 Found 34! 6 3! 25 51 24 1! 22 261 multicaulis j — Expected 12 25 12 25 50 25 12 25 12 h Confining our explanations to Exps. 23 and 25, in which Hh and Rr taken separately segregate normally, we note that the heterozygotes for hair are distributed fairly, according to the usual law, among the three types RR, Rr and rr. There are, however, too many plants of HHRR type and too few of hhRR, and also too many plants of hhi^ and too few of HHrr. In other words, hairy radiate types and smooth non-radiate types are produced in greater relative abundance than glabrous radiate or hairy non-radiate types. In Exp. 27, which although anomalous, agrees with Exps. 23 and 25 in these respects, only one plant was HHrr and could therefore be of the type of a hairy multicaulis or non-radiate lanuginosus. In Exp. 25 there was only one HHrr plant, and in Exp. 23 only two. Hence it is clear that the production of a hairy praecox, erectus or multicaulis, or of a non -radiate lanuginosus^ 260 Inheritance in the Growidsel is a matter of difficulty, for the factors for hair and rays are by no means the only ones present. There are at least two possible explanations of this behaviour, viz. (1) that the four types of gamete RH, Rh, rH, rh are not produced in equal numbers — gametic coupling; and (2) that whether produced in equal numbers or not, certain unions are preferred to others, e.g. RH X RH and rh x rh to rH x rH or Rh x Rh. This is not an im- probable explanation, for certain combinations might easily have special advantages; e.g. more rapid growth of the pollen tube or quicker response to the chemotactic stimulation of the ovule. Let us assume that R and H are really dominant, and simplify Table X to TABLE XL Exp. Cross Exp. 23 Lannginosus x praecox Ratio 9:3:3:1 Ratio 22 : 5 : 5 : 4 No. of t j.)y plants raised HR Ht hR hr 83 Found 52 16 9 6 — Expected 45 15 15 5 — Expected 51 11 11 9 Exp. 25 Lanuginosus x erectus 88 Found 51 11 15 11 Ratio 9:3:3:1 — Expected 50 16 16 55 Ratio 22 : 5 : 5 : 4 — Expected 54 12 12 10 Exp. 27 Lanuginosus x multicaulis 202 Found 116 23 37 26 Ratio 9:3:3:1 — Expected 113 37 37 12 Ratio 22 : 5 : 5 : 4 — Expected 123 28 28 22 Totals 373 Found 219 50 61 43 Ratio 9:3:3:1 — Expected 210 70 70 23 Ratio 22 : 5 : 5 : 4 — Expected 228 52 52 41 The agreement with the latter ratio is remarkably close, and the inference may fairly be drawn that coupling takes place according to the system ^HR : IHr : IhR : 2hr, There is, however, more than this to be deduced from the experiments. The excess of HHRR plants suggests that the presence of R helps the factor H to assert itself more effectively in the development of the hair character. The analysis of the F^ generation of these lanuginosus hybrids made it desirable that further tests should be made by utilizing the ^3 generation. The following cultures were therefore undertaken in 1911, with seeds produced on selfed plants of 1910 {F^). The three objects already described were still in view — to secure new hairy and non-radiate types, to test the general transmissibility of hairiness, and to determine the relationships of the factors for hair and rays — but there was now a further one, viz. to test the accuracy of the A. H. Trow 261 analysis of the F^ generation. It is well known that the safest test (as well as the most troublesome) for a presumed heterozygote is to raise a colony of plants from it. Ten of the eleven colonies raised shewed that the constitution of the parent plants had been correctly estimated with respect to hairiness. TABLE XII. Exp. Culture Na for 19U CnMS Hair type of parent F^ Hair type of progeny /"s Ray type of parent No. of plants required Exp. 59 28 1 hh — RR 100 Exp. 60 29 Lanuginosus x praecox Hh Hh rr 100 Exp. 61 30 1 HH HH Rr 100 Exp. 62 Exp. 63 31 1 X „2 t Lanuginosus x erectus Hh HH HH HH rr Rr 100 100 Exp. 64 33 \ Hh Hh Rr 100 Exp. 65 34 Hh Hh rr 100 Exp. 66 35 Lanuginosus x multicanlis Hh Hh RR 100 Exp. 67 36 HH HH RR 50 Exp. 68 37 HH HH Rr 100 Exp. 69 38 HH HH rr 100 Exp. 70 39 > HH HH RR 50 The eleventh (Exp. 62) marked Hh, proved to be constant for hairi- ness, all the F3 plants conforming to the standard H\ The plants were all 7^ ; if we assume that the non-radiate condition has the effect of depressing hair development, we secure at least a provisional explanation of this case. The original analysis, however, becomes as a whole subject to an error of 10 7o» ^^^ ^^^ must admit that the results are to be accepted only with such reservations. Let us now note the result of the examination of the segregation of the hair in the Hh types of Table XII, and take in order the rr, RR and Rr groups. The results are presented in Table XIII. The numbers accord fairly well with the expectations. The divergences are partly due to the attempt to recognise the heterozygotes. The most hairy types in No. 29 were as hairy as lanuginosus — in the three cultures Nos. 34, 35 and 33 this was not the case. We may assume that typical multicaulis carries a factor Y which depresses the development of the hair character. In its absence (y) hair is fully developed if the factor for hair, H, is present. Pure praecox is a yy plant, multicaulis is YY. This hypothesis will also enable us to explain the interesting result of Experiment 36. Two types of hairy plants appeared in this case : (a) very hairy — H^ — like lanuginosus, and (6) hairy— i/2— like the HH plants of No. 35. The 13 very hairy 262 Inheritance m the Groundsel TABLE XIIL Types Exp. Exp. 60 Wo. 29 Exp. 65 34 Exp. 66 35 Exp. 67 36 Exp, 64 33 Cross Lanuginosas x praecox Lanuginosus x multicaulis Lanuginosus x multicaulis Lanuginosus x multicaulis Lanuginosus x multicaulis Hair type Hh Ray No. of type plants rr 97 HH 22 90 23 Hh hh 49 26 44 23 Hh RR 96 31 (H«) 33 32 Hh RR 49 13(H') 36 H^ Remarks Very hairy individuals present in HH The hair in HH plants relatively weak The hair in HH plants well developed The H' plants as hairy as lanuginosus Hh Rr 49 14 20 15 Hair fairly well veloped de- plants were apparently of the same physiological constitution as lanu- ginosus— the soil did not suit them — the 36 hairy individuals were fairly well adapted to their environment. It was doubtless no accident that caused one-fourth of the plants of this colony to behave in this way. The plant selected as the parent for this colony happened to be an individual with the constitution Yy. The thirteen very hairy plants were yy in constitution. This explanation may possibly suffice to account for the occurrence of a certain percentage of weakly plants in other cultures derived from the cross lanuginosus x multicaulis. These weaklings are recognisable in the seed-pans and boxes, and invariably perish when planted out. May they not have been to a great extent the yy plants of the different generations ? They are always so hairy that the possi- bility has always been kept in mind that ill health in itself, whatever its cause, promotes hair development by its indirect influence on certain regulating factors. These weak, hairy plants were carefully grown and watched in 1912, and proved to be invariably long-styled and relatively infertile. Fig. 17, PI. XVII shews the capitulum of one of them. TABLE XIV. Remarks No rr plant was very hairy, although 2 were placed under H^ Rather weak type of H^ hair. Unaf- fected by segre- gation of Rr Hair tending to dis- appear with age ; unaffected by the segregation of Rr Culture No. Cross Hair type Ray type VariabiUty in hairiness Exp. m m 1p Exp. 61 30 Lanuginosus x praecox HH Rr 34 (22RR lORr 2rr 52 4RR 21Rr 21rr 6 - iORR iRr 2rr Exp. 63 32 Lanuginosus x \ erectus j HH Rr 0 92 0 Exp. 68 37 Lanuginosus x ) multicaulis J HH Rr 0 98 0 A. H. Trow 263 Let us now examine the variability of the hair character in the six remaining groups of Table XII, in which the parents were all homo- zygous for hair (Hff). Table XIV gives the result for those groups whose parents were Rr, Table XV for those whose parents were RR or rr. In Exp, 61 the hairiness varies in degree, shewing the three grades H', H* and W, and the RR, Rr and rr types are irregularly distributed amongst these. There are so few H^ types and these are so loosely distinguished from J?' types that the colony may be arranged thus : — 92 plants I . I . 34fl» 58H« I ^ 1 I ^ 1 111 111 22iJJJ lOjRr 2rr 4BB 3LRr 23rr The numbers, although 58 : 34 might represent 9 : 7, shew that there is apparently no real segregation of hair factors in this case. The diflBculty of arranging the plants in definite groups according to their hairiness supports this view. We have therefore to explain as best we can the remarkable association of RR and H^, and rr and H^. We are perhaps justified in assuming that the greater type of hairiness — W — is due to the association of the hair factor H with R. On this assumption HHRR plants should be very hairy and of the type JT*, HHRr „ „ „ less hairy „ „ „ „ H\ and HHrr „ „ „ slightly hairy „ „ „ „ H\ The facts revealed by Exp. 61 agree very well with this assumption. Applying the hypothesis to the explanation of the results of Elxps. 63 and 68, we are at once met with a difficulty, the standard of hairiness is constant at the grade H^ and the segregation of the ray character does not affect it. This difficulty may be removed by amplifying the original hypo- thesis. Let Z be an unknown factor which when present nullifies the stimulating effect of R on H. Then, since the presence of ii promotes hairiness in Exp. Q\, praecox is probably a zz plant, and since the hair character is constant in Exps. 63 and 68, erectus and multicaulis are probably ZZ plants. Such a hypothesis is somewhat involved and perhaps unnecessary; but as it explains the results hitherto obtained and can readily be tested 264 Inheritance in the Groundsel by experiment, it deserves provisional acceptance. It is possible that the presence of the Y factor alone in Exps. 63 and 68 might serve to explain the result. Further experiments are necessary. The remaining three experiments are of greater interest, as they were intended to test the behaviour of the progeny of plants homo- zygous for hair and rays. An unexpected segregation occurred in one of them. TABLE XV. Variability in hairiness Culture No. Cross Hair type Ray type if3 H2 HI 31 Lanuginosusx 1 erectus J HH ?T — — 9BYY 38 Lanuginosus x ) multicaulis J HH rr 202/2/ 49 r?/ 27 FY 39 Lanuginosus x multicaulis HH RR — 49YY — Exp. No. Cross type type H^ H^ H' Eemarks Exp. 62 31 Lanuginosus x [ HH rr — — 93YY Hairiness of the lowest type and constant. Hairiness probably de- pressed by YY Exp. 69 38 Lanuginosus x | HH rr 20yy 49 F?/ 27YY All H^ plants unhealthy. Hairiness upon the whole less than in the praecox cross. Exp. 61 Exp. 70 39 Lanuginosus x ) HH RR — 49 FF — Hairiness not quite equal multicaulis [ toH'. Hence FFplants. R raises the grade of hairiness The plants of Exp. 62 were all of a very low standard of hairiness. This may be partly accounted for by the absence of R and the presence of F. Exp. 69 is comparable with Exp. 67. The Y factor is involved, and the heterozygotes are recognisable. The slight loss of hair may be due to the absence of R. The numbers found correspond very well to the expectation. Found :— YY : Yy : yy :: 27 : 49 : 20 Expected:— :: 24 : 48 : 24 In Exp. 70 the plants shewed constant hairiness ; they were probably of YY type, the presence of R raising the grade of hairiness from H^ The study of the inheritance of hair is no doubt incomplete ; but it is clear that (1) the transmissibility of hair from one type to another is possible ; (2) that two pairs of factors at least are involved, H, h and Y, y ; (3) that consequently there must be at least several grades of hairiness ; and (4) the presence {R) or absence (r) of the ray factors modifies the hairiness due to the proper hair factors. The difficulties of the study are accentuated by the transgressive variability of the segregating characters and by the direct influence of the environment in producing slight non-heritable modifications. A. H. Trow 265 Observations on Stem Colour. At an early date in the course of these experiments it was noted that some of the pure types were green-stemmed, viz. erectus, erectus, radiatus and rmdticaulis; others had stems which were more or less reddish in colour, viz. praecox, genevensis and lanuginosus. No attempt was made to study the behaviour of this pair of characters in segregation until 1911, when certain families of the F^ generation of the crosses lanuginosus x erectus and lanuginosus x midticaulis were found to consist either of green-stemmed individuals only, of red-stemmed individuals only, or of individuals of several types of stem-colour. It seemed that stem-colour varied independently of the other characters under investi- gation, and was therefore probably represented by definite factors in the constitution of the plants. There are very great difficulties in arranging the plants in definite categories according to stem-colour, for green-stemmed plants tend, under certain conditions (exposure), to develop some red colour, and red-stemmed plants lose some of their red colour in shade. Moreover, the red-stemmed types are not red-stemmed throughout — the upper internodes are apt to be green, although these parts are subject to the most intense illumination. Green-stemmed plants, if they develop red colour at all, do so most freely on the lowest internodes. Let GG denote red-stemmed and gg green-stemmed plants. When G is present (GG) and certain standard conditions are maintained, at least one-half of the main axis should be of a reddish colour; if G is absent (gg), there should be little more than traces of red colour except at the very base of the main stem ; the intermediate condition (Gg) is represented by varying degrees of redness, both as regards the intensity of the colour and its extent. The pure types grown in adjacent beds under similar conditions furnish a convenient standard of reference. Perhaps the simplest proof of the existence of factors for stem colour is furnished by the investigation of the F^ generation of genevensis X erectus, radiatus. Both parent plants are glabrous, but differ with respect to two pairs of characters. Genevensis is red-stemmed and non- radiate (GGrr) ; erectus, radiatus is green-stemmed and radiate (ggRR). Exp. 16. A colony of 234 plants was raised and brought to the flowering stage without the loss of a single plant. 266 Inheritance in the Groundsel Segregation took place as follows : — For stem colour:— GG : Gg : gg :: 53 : 124 : 57 For the ray character : — RR : Rr : rr :: 5S : 115 : 66 The expectation in each case was :: 58 : 117 : 58 But nine types should be present in the F^ generation. If these occur in the expected proportion, we have a fair proof that there are two pairs of factors and that they not only segregate normally, but that the method of investigation is fairly accurate. Table XVI gives the result of the analysis made from this point of view. TABLE XVI. RR Rr rr GG Gg gg GG Gg gg GG Gg gg Nos. found 10 30 13 28 57 SO 15 37 14 Nos. calculated 15 29 15 29 58 29 15 29 15 Differences -5 +1 -2 -1 -1 +1 0 +8 -1 The approximation is as close as could be reasonably expected. The greatest difference amounts to a divergence of 33 °/^. A similar result was obtained with the Fs generation of the cross lanuginosus x erectus, the F^ parent plant having apparently the con- stitution HHYYRrGg. The F^ generation in this case consisted of 92 plants, and the segregation proved to be as follows : — For the ray character: — RR : i2r : rr :: 23 : 52 : 17 For stem colour :— GG : Gg : gg :: 20 : ^8 : 24; For both characters, see Table XVII. TABLE XVII. RR Rr rr GG Gg ga GG Gg w GG Gg 09 Nos. found 5 9 9 11 29 13 4 10 2 Nos. calculated 6 12 6 12 23 12 6 12 6 Differences -1 -3 + 3 -1 + 6 + 1 -2 -2 -4 When we consider the small number of plants experimented with, the result agrees fairly well with the expectation. The deviations in the types RRgg and rrgg are however worthy of note. They may mean more than mere chance aberrations. Treating R and G as dominant, the segregation appears as follows Found:— RG : rG Calculated :—RG : rG Rg-.rg :: 54 : 14 : 22 : 2 Rg :r^:: 52 : 17 : 17 : 6 A. H. Trow 267 There appears to be a distinct impediment to the production of green- stemmed, non-radiate plants, and therefore a tendency to the formation of a 9 : 3 : 4 ratio. The examination of the ^3 generation of the cross lanuginosua X multicaulis for stem colour was attended with difficulties. It is difficult to arrange the plants under the three types GG, Gg and gg, owing possibly to the direct colour modifications due to the environ- ment. Much more work is required before a clear and complete explanation can be given of the results in the case of this cross. Six colonies were examined, all growing together in the open air, under conditions as similar as possible. The outside plants of each colony might have been expected to shew differences when compared with the inside ones, but none were noticed, although specially looked for. On the other hand, a certain efifect of shading was very obvious ; when the earlier plants were uprooted for examination from some of the beds, the later ones received considerably more light and air, and their stems quickly acquired a certain amount of red colouring matter. This kind of response is well known to botanists, and appears to be more or less independent of the colour factors for stems that we are considering here. The result of the examination of these six colonies for stem colour is presented in TABLE XVIII. Exp. Exp. 64 Culture No. Cross 33 Exp. 66 35 Exp. 67 36 Exp. 68 37 Exp. 69 38 Exp. 70 39 Lanuginosus x ) multicaulis ; Lanuginosus X | multicaulis ) Lanuginosus x ) multicaulis j Lanuginosus x ) multicaulis f Lanuginosus x ) multicaulis \ Lanuginosus x ) multicaulis j Gen. Estimated constitution of parent No. of plants Types of Stem Colour HhRrGg 49 HhRRGg 96 GG 13 Gg 20 77 16 19 Bemarks HHYyRRGG HHYYRrGg 49 97 49 — F3 HHYyrrGg 92 F3 HHYYRRGg 48 Difficult to distin- guish GG from Gg All plants, reddish green Colony difficult to sort. GG not sharply defined Possibly all plants were gg Possibly all plants were gg It does not seem desirable to spend much time in analysing further these incomplete results. Additional experiments are necessary. Never- theless all are of interest. Green is apparently recessive. It is certainly easier to pick out the pure greens (gg), for the pure reds are apt to pass over more or less gradually into the reddish greens. 268 Inheritance in the Groundsel Cultures 39 and 35 suggest the possibility of a single pair of factors; cultures 38 and 37, of two pairs of factors, giving the ratio 9 : 7. The difference between the numbers found and those calculated is consider- able. For 97 plants the ratio should be 55 : 42 and not 62 : 35 ; for 92 plants, 52 : 40 and not 57 : 35. Culture 36 did not segregate, although apparently of an inter- mediate type. It is probably to be regarded as a modified pure red (GO). The facts, as a whole, point unmistakably to segregation for stem colour taking place in the hybrid lanuginosus x multicaulis. Culture 33 is of special interest, as there were three pairs of characters involved. Theoretically, there should have been 27 different kinds of plants in this colony. Although there were less than 50 plants, 15 of these kinds were found. The segregation for rays, stem colour and hair separately considered was as follows : — RR : Rr: rr :: 11 : 25 : 12 GG :Gg :gg :: 13 : 20 : 16 HH'.Hh.hh :: 14:20: 15 The detailed analysis of this result suggests that there is possibly a gametic coupling of the form IHG : nHg : nhG : Ihg where ri is a fairly large number. The demonstration of such a coupling, including the determination of the value of n, involves further experiments on a larger scale, and the discussion of this result must be postponed until these have been completed. Leaf Colour Factors. A brief reference may be made to the factors for leaf colour, the detailed examination of which has scarcely commenced. It is certain that segregation of leaf colour characters takes place in these groundsel hybrids. The foliage leaves of erectus and erectus, radiatus are of a yellowish-green colour, those of multicaulis are of a dark green. Each pure type can be almost said to have its own distinctive shade of green, so that one pair of factors can scarcely suffice to explain the facts unless we assume that the leaf colour factors are affected by other factors. In the F2 generation of the hybrid multicaulis x erectus, radiatus the leaf colour was observed to segregate, but the apparatus necessary for esti- mating the types exactly was not available and therefore no definite analysis was made. It may be provisionally assumed that there is a factor for leaf colour L which determines the darker leaf colour, its A. H. Trow 269 absence (/) may then represent the yellow-green colour. It is note- worthy that in one culture of hybrids of Senecio sylvaticus x 8. viscosus — a species hybrid — a number of seedlings with white cotyledons were observed, which of course soon perished. It is probable that the occur- rence of these marked the segregation of another leaf colour character, one which is very likely to occur in Senecio vulgaris also*. The most exacting and perhaps most interesting work of the last five years, consisting of a long and very tedious series of measurements of the vegetative organs, still remains to be discussed, but must form ' the basis of a second contribution to our knowledge of the common groundsel. Summary. The common groundsel, Senecio vulgaris, Linn., is an aggregate species which has been found to include many segregate or ele- mentary species. Twelve of these elementary species have been cultivated and main- tained pure and true to type for at least several generations. Six of them have been studied in detail, and are distinguished by more or less descriptive names — praecox, erectus, multicaulis, latifolius, genevensis and lanuginosus. Of these lanuginosus is radiate, the other five non- radiate. Lanuginosus was obtained from the Channel Islands, ^enere/wi* was collected in the vineyards near Montreux, and the others were found in the neighbourhood of Cardifi". Another form, collected near Cardiff and equally well studied, is erectus, radiatus — the radiate variety of erectus. Five other forms have proved true to type in cultures, but have been so incompletely studied that they are for the present simply designated with their place of origin — all Glamorgan localities: (1) Car- diff, (2) Burry Green, (3) Horton, (4) Cross Common, and (5) St Bride's. It has been shewn by experiment that the radiate character of erectus, radiatus can be transferred by hybridization and subsequent segregation to praecox, multicaulis, latifolius and genevensis. A radiate variety of each of these elementary species has in fact been produced in this way, and is now being cultivated. In multicaulis there are at least three kinds of radiate varieties, with yellow, cream and fimbriate florets respectively. Hence, if we accept the usual terminology and persist in regarding S. vulgaris as the species, we have to make varieties of * This prophecy was fulfilled in 1912. Seedlings with white cotyledons were observed in the F^ and F4 generations of the cross lanu^inosut x praecox. Joorn. of Gen. 11 19 270 Inheritance in the Groundsel varieties of a variety; thus, Senecio vulgaris, Linn., var. multicaulis, var. radiatus, vax. fimhriatus. Hairy varieties, in contrast to radiate varieties, are not produced so readily. The non-radiate type of lanuginosus has not yet been certainly obtained, nor a typically hairy praecox, erectus or multicaulis. All, or nearly all, non-radiate types are less hairy than lanuginosus. Consequently it is still somewhat doubtful whether the hair of lanu- ginosus can be transmitted to the glabrous types quite unchanged. Analyses of the various segregations reveal the presence of eight or nine factors, which may be represented in the usual way, accepting provisionally, on account of its convenience, the theory of dominance and the presence and absence hypothesis ; thus (a) Flower-factors. 1. R, r. Factor for the radiate character. JJ = radiate. r= non-radiate. 2. C, c. Factor for flower colour. C = yellow. c = cream. 3. X, X. Factor for determining the development X inhibits the development of of the cream colour. cream colour. xc = cream, Xc = yellow. 4. F, f. Factor for the development of the fim- F= normal rays. /= fimbriate briate character. rays. (b) Hair-factors. 5. H, h. Main factor for hairiness. Jf= hairy. ^ = glabrous. 6. Y, y. Factor modifying hairiness. Y depresses hairiness, y allows full development of hairiness. 7. Z, z (?) Factor affecting the stimulation of If by Z reduces hairiness in R plants. R. z allows normal hairiness in R plants. (c) Stem Colour factor. 8. G, g. Factor for red and green colour in stems. G=:red. gr = green. {d) Leaf Colour factor. 9. L, I. Factor for leaf colour. i=dark green. Z= yellow green. These factors behave normally in segregation, except that in certain cases gametic coupling has been recognised, viz. a gametic coupling of the form IHU : \JIr : IhR : 2hr and possibly one of the form IHG : nHg : nhG : Ihg where w is a fairly large number. The latter type of coupling suggests that certain combinations of characters may either be unrealizable or realizable only with difficulty. A. H. Trow 271 The constitution of the pure types has not been adequately in- vestigated, but already the results are of interest. The constitution of the twelve already referred to is given below : — Praecox — rrCChhyyzzGGLL. Erectus—rrCChhXX YYZZggll. Multicaulis — nxchhxx YYZZggLL. Latifolius — rrhhgglL Genevensis — rrhhGG. Erectus, radiatus — RRCChhX X ZZggll. Lan uginosus — RRHHyyGGLL. Cardiff Groundsel \ Burry Green Groundsel I Horton Groundsel frrHHGG. Gross Covimon Groundsel St Bride's Grmindsel Crosses are apparently possible between any pair of these. The cross multicaidis x erectus, radiatus gives a hybrid of the constitution RrLlGcXx heterozygotic for four pairs of factors, and there should thus be at least 18 distinct types of plant in the ^2 generation, recognisable by inspection, without further experiment. The cross lanuginosus x multicaulis gives a hybrid of the constitution HhGgRrYy, and there should be at least 54 recognisable types in the F^ generation. Moreover several other factors affect the form and habit of the plants, producing still further diversity of type. Consequently there must be several hundreds of these groundsel forms still unidentified, but never- theless recognisable. Such facts shew how inadequate the older methods of study are for the investigation not only of hybrids, but of critical species also. After an investigation extending over six years, including the critical examination of about 10,000 groundsel plants, I still often find it very difficult to estimate, even provisionally, the constitution of a casual wild plant. Yet the methods of genetics, diligently applied, obviously give one the power to replace loose speculation and guesswork by irrefutable inductions, and so to lay down a foundation upon which the evolutionist and taxonomist can build with safety. AfrU 12tA, 19X2, 1»— 2 272 Inheritance in the Groundsel Additional Note {August 14, 1912). Two F^ colonies of the cross lanuginosus x praecox of 213 and 215 plants respectively were raised during the current year, and further careful observations made of the segregations for iiairiness and ray characters. It seemed desirable for several reasons to verify the type of coupling or reduplication of the form 2 : 1 :: 1 : 2 described at p. 260. A close study of the various types of coupling may give valuable clues as to the time, mode or place of segregation. The type represented by the above ratio has not hitherto been definitely recognised. Gregory (/. of Genetics, Vol. i, p. 128), however, publishes a result which he suggests may be due to coupling in one sex only, according to the ratio 7:1:1:7. While one must admit the possibility and even the probability of the occurrence of differences of this kind in the sexes, such an interpretation becomes doubtful when we find that the hypothesis of a coupling of the form 2:1:1:2 in each sex explains the result equally well; thus Nos. observed by Gregory Nos. expected when coupling is of the form 7:1:1:7, and) confined to one sex j Nos. expected when coupling is of the form 2:1:1:2, and| occurs in both sexes J The results obtained by me this year may be summarized as follows : — HR Br hR hr Culture 23. Nos. found 128 36 31 18 Culture 24, „ ... 132 32 25 26 MG Mg mG mg 411 98 97 78 417 96 96 75 418 95 95 76 Both cultures ,, 260 68 56 44 Expectation, on the basis of coupling of the form) „„„ ^0 cq a a 2:1:1:2 in both sexes. Eatio:— 22 : 5 : 5 :4i ^^^ ^^ ^^ ^*^ Expectation, on the basis of coupling of the form) .-,-5, „„ ^„ .„ 7:l:l:7inonesex. Katio:-39 : 9 : 9 : 7 | ^^^ ^^ ^^ ^^ Further experiments are necessary to determine which of these forms of coupling is actually operative in such cases. In Primula, such experiments would be comparatively easy ; in Senecio, they would be somewhat difficult If the Fi plants are crossed with the recessive parents, thus (1) F, ? (HRhr) xP^ (hr) and (2) P ? (hr) x F, ^ (HBhr), A. H. Trow 273 we should expect in each case (if the coupling occurs in both sexes in the form 2:1:1:2) families of the constitution 2HR : iHr : IhR : 2hr. If, however, coupling occurs in one sex only in the form 7:1:1:7, one family should have the constitution 7HR : IHr : IhR : Ihr, and the other the constitution \HR : \Hr : IhR : Ihr. In Senecio, the ^i $ plants would need isolation, and the bisexual disc-florets would have to be removed while in bud. The P*$ plants unfortunately have no $ flowers, and it is practically impossible to remove the stamens without injury to, or accidental pollination of, the stigmas. It would therefore be necessary to allow self- and cross- pollination to take place, and to allow for the additional hr types produced by self-pollination. The ratios to be looked for are : — 2:1:1:2; 7:1:1:7; 1:1:1:1; 2:1:1:2-1-0;; 7:1:1 :7+a;; and 1:1:1 :l-f- a:, where x represents the additional hr types produced by self-pollination. In Primula, where the necessary castrations are easily made, the three ratios 2:1:1:2, 7:1:1:7 and 1:1:1:1 should alone occur. Small families of 16 to 32 plants should suffice to determine the type of ratio. I regret to say, however, that the pure strain of lanuginosus has become so weak and unhealthy — perhaps by successive selfings — that it is no longer available for experiment. The experimental proof in the case oi Senecio cannot be furnished before the end of 1915 \ The assumption of dominance, although often of great assistance, as in this case, in giving a clear and condensed view of an experimental result, often tends to obscure facts of importance. The nine types which result from the interaction of two pairs of characters are, I think, recognisable in these cultures. The following table gives the numbers found and the expectation on the assumption of coupling of the form 2:1:1:2 in both sexes. Compare also Table X, p. 259. RR Rr rr Culture 23 Culture 2i HE Hh hh HH Hh hh HH Hh hh 35 11 1 15 67 30 3 33 18 47 5 0 17 63 25 2 30 26 82 16 1! 32 130 55 5! 63 44 48 48 12 48 119 48 12 48 48 Total (23 and 24) Expectation. Coupling of the form 2:1:1:2 ^ I find that Gregory (Proc. Roy. Soc. B., Vol. 84, p. 14) has made the necessary crosses and proved that coupling takes place in each sex. The exact form of the coupling, however, still remains somewhat uncertain. Primula sinensis appears to be a somewhat difficult type for such a determination. Upon the whole, the evidence clearly points to the occurrence in this case of the 2:1:1:2 ratio in each sex. 274 Inheritance in the Groundsel Some of the discrepancies may be due to the difficulty of recognising the various types. RRkh plants cannot however be mistaken, yet there was only one and not 12 as expected. Although five plants are marked rrHH, not one of these was a typical non-radiate lanugitiosus — the production of which was one of the chief objects of the experiment. The single RRhh plant was a perfect example of a radiate praecox. Treatirng // as dominant over h, the table becomes RR Rr rr andff/i hh HH and Hh hh HH&ndHh hh 98 1 162 55 68 44 96 12 167 48 60 48 Found Expected .. Treating H and R as dominant over h and r, we get (compare Table XI) : H h H h 260 56 68 44 263 60 60 48 Found Expected ... The process gets rid of all discrepancies of importance, but at the same time obscures the very important facts revealed by the fuller analysis — the great dearth oi RRhh and rrHH plants. This is a point of interest to which students of genetics might pay greater attention. Now that the heterozygotic forms are being more acutely studied, it is probable that many similar results will be brought to light and help to increase materially our knowledge of the actual mechanism of segregation. a Fig. 28. 7?Ex3 Fig. 29. J?rx3 Fig. 80. J?rx3 Fig. 31. rrx3 Enlarged drawings of individual flowers by Miss M. Brockington. Fig. 28 — a ray floret of erectus, radiatus, RR ; Figs. 29 and 30 — ray florets of the hybrids praecox x erectus, radiatus and lanuginosus x erectus respectively, Rr; Fig. 31 — a disc floret of erectus, rr. Note the presence of the five corolla lobes in Fig. 30. A. H. Trow 276 NOTES ON THE ILLUSTRATIONS (Plates XV-XVIII). All the photographs were prepared from the crops of 1912, which incladed colonies of all the types referred to in the paper. Certain new features of interest were observed for the first time in 1912, such as the unexpected appearance of the single specimen of a cream coloured, fimbriate, multicaulis represented in Fig. 20, PI. XVII. These are reserved for furtlier consideration and a later report. Two points, dealing with the ray character and hairiness, deserve a brief notice here. The examination of two large colonies of the F^ generation of the cross lanuginosus x praecox shewed that there were present in each, two distinct RR types (Figs. 7 and 10, PL XVI.) and three Rr types (Figs. 25, 26 and 27, PI. XV.). The difference in the degree of ray development is due doubtless to the presence or absence of the hair factors. Comparison of the figures 25, 26, 27, and of the figures 3, 4, 7, 8, 9, 10, and 11, will prove that the character due to the presence of the same factor may be a variable one. The examination of the Fi generations of the cross lanuginosus x multicaulis, in the current year (1912), led to the conclusion that the four types of hairiness referred to as Ho , Hi, H2, and H3 are mainly due to the existence of three factors for hairiness. In addition to H and 1' there is a third factor which made itself evident by certain colonies from a H^ parent segregating in the proportion H^ : H^::^ : 1. Ho, Hi, and H3 plants do not segregate apparently; Ho = hh, Hi = HHDD, where D represents the third hair 'factor, H3=HHyy. i/2 may be homozygotic and is then probably always HHrF. It is often heterozygotic, and may throw either Hi or H3 plants, but not both, in the proportions Hi : H-i '■: S : I or H-i'. H3 :: 1 : 3. It must not be forgotten that other factors apparently influence the development of hair (see p. 263). Fig. 1. Praecox. 45 days old, photographed as it commenced to flower. The neck of the flask seen in Figs. 1, 2, and 3 is 26 mm. in diameter. The whole plant is practically glabrous. Fig. 2. Lanuginosus. 103 days old, photographed as it commenced to flower. The whole plant is very hairy. The flowers are not easily recognisable in the illustration. Most botanists will, I think, be prepared ultimately to concede specific rank to the two types shewn in Figs. 1 and 2. Fig. 3. One of the H^ types of Exp. 67, specially remarkable for the peculiar character of the rays, which were originally recorded as Rr? but proved to be RR. A single head is shewn enlarged as Fig. 11. It will be noted that the plant differs in other respects trom praecox and lanuginosus. Figs. 4, 5, and 6. Microphotographs of single capitula of multicaulis, radiatus, multicaulis, radiatus x multicaulis, and multicaulis. Radiate and non-radiate forms, if glabrous, produce in all the generations one type of heterozygote only, that represented in Fig. 5. All the microphotographs in Plates XV — XVIII, are magnified about four diameters. Figs. 7, 8, and 9. The radiate forms of praecox, erectus, and genevensis. Fig. 10. Capitulum of lanuginosus. Fig. 11. Capitulum with peculiar RR rays, liable to be mistaken for the Rr type. Com- pare this with Fig. 25, which is that of a Rr head. Figs. 12, 13, 14, and 15. Side views of capitula of multicaulis, praecox, erectxu, and genevensis to shew size and shape of capitula and normal length of styles. Such types set seed freely when selfed. 276 Inheritance in the Groundsel Figs. 16, 17, and 18. Various long-styled types. These are very sterile, '\i selfed, but fertile to their own pollen, j'ielding seed quite freely when pollinated artificially with their own pollen or that of neighbouring long- or short-styled capitula. The capitulum represented in Fig. 16 was taken from an ordinai-y plant of genevensis. Fig. 17 illus- trates the type of head alone found in the sickly, hairy, individuals referred to on p. 262, and was taken from an individual of the F^ generation of lanuginosus xpraecox. Fig. 18 was taken from a plant which may almost be described as a non-radiate lanugi- nosus {Fi generation of lanuginosus x multicauHs). Fig. 19. MulticauHs Jimbriatus, yellow. Fig. 20. Multicatdis JimbriatiDt, cream. Both these are apparently ? , the anthers producing no good pollen, hence the difficulty in clearing up the mode of inheritance of the fimbriate character. Figs. 21, 22, 23, and 24. These figures illustrate the various types of hair development denominated Hq, Hi, H^ and H3 respectively. The portions were carefully selected from young stems at the same phase of development. 21 is from multicauHs ; 22 from the Burry Green groundsel (in 1911 this type had the higher grade of hairiness H2) ; 23 is from a very late flowering type (F4 generation of lanuginosus x multicauHs) ; and 24 is from lanuginosus. Figs. 25, 26, and 27. Three types of Rr rays produced in the F^ generation of lanugi- nosus X praecox. In the Fi generation there is one type only, that represented in Fig. 26. Compare Fig. 25 with Fig. 11. In such cases as these mistakes would be made unless the influence of the hair factors was considered and allowed for. Figs. 28, 29, 30, and 31 (in text, p. 274). Fig. 32 (cp. PI. XV). A nearly ripe head oi praecox, radiatus, shewing the revolute ligules of the old withered florets, a condition often erroneously credited to the flowers at the period of pollination. University College of South Wales and monmouthshiee, cardiff. July 22, 1912. JOURNAL OF GENETICS, VOL II. NO. 3 PLATE XV Fig. 1. Fig. 2. Fig. 25. Fig. 26. Fig. 27. Fig. 3. Fig. 32. JOURNAL OF GENETICS, VOL II. NO. 3 PUTE XYI Fig. 4. Molticaolis, radiatas. Fig 5. Hybrid. Fig. 6. Molticaalis. Fig. 7. Praecox, radiatns. Fig. 8. Erectus, radiatas. Fig. 9. Genevensis, radiatas. Fig. 10. Lanaginosas. Fig. 11. JOURNAL OF GENETICS, VOL II. NO. 3 PLATE XVII Fig. 12. Multicanlis. Fig. 13. Praecox. Fig. 14, Erectns. Fig. 15. Genevensis. Fig. 16. Genevensis. Fig. 17. Fig. 18. Fig. 19. Fimbriate type, Yellow. Fig. 20. Fimbriate type. Cream. JOURNAL OF GENETICS, VOL II. NO. 3 PLATE XVIII Fig. 21. Ho Multicaulis. Fig. 22. Hi Bnrry Green Groundsel. Fig. 23. Ho, F4 of Lanaginosos X multicaulis. Fig. 24. H3 Lanuginosus. THE r6lE of oxydases IN THE FORMATION OF THE ANTHOCYAN PIGMENTS OF PLANTS. Br FREDERICK KEEBLE, Sc.D, Professor of Botany, University College, Reading, Ain) E. FRANKLAND ARMSTRONG, D.Sc, Ph.D. — CONTENTS. PAOB I. Introduction and Methods 277 n. The Distribntion of Oxydases in Plant- tissues .... 283 A. The Oxydases in the Vegetative Members of Primula sinensis 283 B. The Oxydases in the Flower of P. sinensis . . 289 1. Self-coloured varieties 289 2. Recessive white varieties 290 3. Flaked (Ever-sporting) varieties .... 291 C. The Oxydases of Dianthus barbatus (Sweet William) 293 1. Ever-sporting varieties 293 2. White varieties of Sweet William ... 294 D. The Oxydases of Geranium sanguineum .... 295 E. The Localisation of Oxydases in the Tissues of the Flower (P. sinensis) 296 F. Dominant White Varieties 298 G. The Nature of the Oxydases in Flowers .... 304 1. The Oxydases and Peroxydases of P. sinensis . 304 2. The Oxydases and Peroxydases of other flowers . 305 3. The Influence of Light and Darkness on the Oxydase-content of plants 306 I. Introduction and Methods. The progress of discovery with respect to the genetics of colour has been so rapid of recent years as to outpace that of our knowledge of the chemistry of pigment-formation in plants and animals. Thanks to the labours of Bateson, Miss Saunders, Punnett, Gregory, Baur and many 278 Oxydases and Pigments of Plants other workers, the modes of inheritance of colouration are known with detailed accuracy in many species. This rapid 'advance has stimulated enquiry into the chemical aspect of the phenomena of pigmentation, and the investigations of Miss Durham (1904), Palladin (1908), Combes (1910), Miss Wheldale (1910) and Gortner (1910) have resulted in the production of a large body of evidence in favour of the current hypothesis that pigmentation is the outcome of the action of oxydase on chromogen. We propose in the present communication, some of the results of which have been published in abstract form elsewhere (Keeble and Armstrong, 1912), to present further and as we think convincing evidence in favour of this hypothesis, to describe the distribution of oxydases in Primula sinensis and certain other plants, to produce experimental evidence of the fact that the pigment-forming activity of oxydases may be inhibited in various races of plants (Dominant White races) and to show that the oxydase-content of a plant is modified by such external conditions as light and darkness. The Role of Oxydase in Pigment Formation. Inasmuch as the literature relating to oxydase is very extensive and has been summarised recently in monograph form by Kastle (1909) and by Clark (1911) we need do no more here than state briefly the hypothesis of Bach and Chodat, the acceptance of which is implied in the present communication. This hypothesis holds that an oxydase is of a dual nature. The constituents of an oxydase are a peroxydase and a peroxide. The peroxide behaves as an activator to peroxydase in the sense that it supplies the latter with oxygen which may then be transferred to an oxidisable body. This activating action may be effected by hydrogen peroxide. The nature of the peroxides of the plant is unknown. Peroxydases are classed generally with enzymes ; they differ, however, from hydrolytic enzymes in that they undergo destruction, at least in vitro, in consequence of the exercise of their oxidising function. There seems reason to believe that iron, manganese and other elements in combination with organic substances play some part as yet obscure in the oxidation processes of the plant. The hypothesis that oxydases are concerned in the formation of plant-pigments was first suggested by Pick (1883), and evidence in ftxvour of this hypothesis has been put forward by many observers — F. Keeble and E. F. Armstrong 279 particularly by Palladin and Miss Wheldale. The latter investigator (1910 loc. cit.) has formulated in a clear and schematic manner the following hypothesis as to the course of events which leads to the formation of pigment in plants. The colourless chromogen by the oxidation of which pigment is produced occurs in the plant as the constituent of a glucoside. In this combined form it resists oxidation. Enzymes of the emulsin type hydrolyse the glucoside and liberate chromogen which is then oxidised by atmospheric oxygen made active by the oxydase. Thus an anthocyan or soluble sap pigment is formed. Such anthocyan pigments, which are common in the flowers and other parts of plants, are of very different chemical constitution and origin from the plastid pigments which are also of common occurrence. We are concerned here only with the former. The anthocyan pigments are generally red, violet or blue. The chemical nature of the anthocyan pigments is obscure. They are regarded by Miss Wheldale as flavone derivatives which are known to be widely distributed in plants in the form of glucosides. An illuminating paper suggesting a mode of formation of organic pigment has been published recently by Chodat (1912). He shows that when tyrosinase acts on ^-cresol in the presence of one or other of the products of protein hydrolysis a series of diversely coloured pigments resembling the natural pigments is produced. Chodat believes that a pigment of a given kind is produced by the action (1) of an oxydase on (2) a phenolic compound in the presence of (3) an amino compound, and suggests that, as the compositions of the phenolic compound and the amino compound vary, so the composition and hence the colour of the pigment varies. Gortner (1910 and 1911) working independently has already pub- lished conclusions which in some respects are similar to those of Chodat. In a series of valuable papers Gortner has produced evidence in support of the view, put forward previously by Miss Durham (1904) and other observers, that the black or brown melanin pigments of animals are formed by the action of tyrosinase on a product of prot,ein hydrolysis, namely tyrosine. He holds further, as described in the text, p. 299, that inhibition of pigment formation may be brought about by the con- version of part of the tyrosine into a closely allied substance which is not only itself resistant to tyrosinase but also checks the action of that substance on tyrosine. Gortner applies this hypothesis to the elucidation of the suppression of pigment which is known on genet ical grounds to occur in certain white forms of plants and animals. 280 Oxydases and Pigmeiits of Plants Such forms have been investigated by Bateson and are described as dominant whites. Although they may be identical in appearance with true albinos they, differ fundamentally from the latter in gametic com- position and hence in genetical behaviour. The true albino lacks colour, the dominant white possesses the power of pigment formation, but that power is held in abeyance by an inhibitor. Inasmuch as the phenomenon of dominant whiteness is dealt with somewhat fully in the text, we need not devote attention to it here, beyond remarking that attempts to demonstrate by chemical means the truth of the Mendelian hypothesis with respect to dominant whites have not as yet been wholly successful. Miss Wheldale (1910), whose researches have contributed so much to our knowledge of this subject, regards inhibition as due to the action of deoxidising substances such as sugars, tannins, and the like, and has brought forward experimental evidence in favour of this opinion. In cases of partial inhibition, which is illustrated in P. sinensis by the dominance of pale over more deeply coloured varieties, it is suggested that the effect is due to a reductase. Methods. The methods in general use for detecting the presence of oxydase in plants depend on the addition of a colourless chromogen to the solution or extract obtained from the plant. If the result of the operation is to produce a pigment it is concluded that oxydase is present. If the pigment be produced only after the addition of hydrogen peroxide the oxidising substance is described as a peroxydase. The method has many disadvantages. Certain of the chromogens used as tests for oxydases undergo oxidation with more or less rapidity when exposed to the air. The extract or solution may contain reducing or inhibitory substances which interfere with the reaction, and in any case the localisation of the oxydase in the tissues of the plant is wellnigh impossible by this method. Clark (1911) has carried out recently an elaborate series of tests with different chromogens and finds that certain of them may be used for the microchemical determination of oxydase in plant tissues. Among the reao-ents used by Clark are guaiacum, phenolphthalein and a-naphthol. Pyrogallol in the presence of glucose is employed by Chodat. Schreiner has obtained valuable results with respect to the oxydases of living roots by the use of benzidine. F. Keeble and E. F. Armstrong 281 As the result of an extended trial of the many reagents which have been employed for the purpose of demonstrating oxydase we find that the most serviceable for our purpose are a-naphthol and benzidine. We make use of the fact demonstrated by one of us (Armstrong, 1910) that substances dissolved in alcohol penetrate rapidly into plant cells, and the method which we employ is as follows : In the case of benzidine, a 1 per cent, solution is made in 50 per cent, alcohol which is then diluted with so much water as not to cause a precipitate. The object to be examined, for example, an intact corolla or a section of a petal of P. sinensis, is taken from the plant and placed immediately in the reagent in a corked specimen tube. The tube is incubated at 37° C, and if no direct oxydase reaction is obtained the material is removed from the tube, washed lightly with water and treated with 1 to 2 drops of 10 vols, hydrogen peroxide, or the latter reagent may be added directly to the tube containing benzidine. In the case of o-naphthol, the solution may be used of even less alcoholic strength. Sections may be prepared either with a dry razor or with one moistened with a solution of cane sugar of suitable strength. No inhibitory action appears to be produced by this substance. The first effect of treating coloured flowers with either of these reagents is a decolourisation of the pigmented parts. This decolourisation appears to be due to the action of a catalase, but we have not made it a subject of special investigation. As we show in detail in Section II. the result of treating a coloured or recessive white flower of Primula sinensis is the demonstration of peroxydase in the epidermal layer of the petals and also in the bundle sheath which surrounds and accompanies the veins in their finest ramifications. Preparations of a considerable degree of permanence may be made for demonstration purposes in the following manner: — The flower in which the oxydase reaction has developed is washed well in running water, the corolla tube is cut away, the petals floated on water, whence they are transferred by means of filter paper to a glass lantern slide. The superfluous water is removed by means of filter paper, and dry filter papers are placed over the flattened flower, A glass slide is laid on the filter paper, the preparation is placed in the incubator at 37° C. and pressed down by means of a weight. After a day or two, when the preparation has become dry, it is treated as a lantern slide, bound and kept in a dark place. The degree of permanence seems to depend first, on the thoroughness with which the preparations are washed. 282 Oxydases mid Pigments of Plants and second, on the extent to which they are preserved from the access of light. Many of our preparations with benzidine retain their colour for months, but the colours produced by the use of a-naphthol may show signs of fading within a shorter period. Sections may be mounted either in glycerine jelly or may be taken up rapidly through the alcohols and mounted in Canada balsam. In the case of large objects such as fruits the reagent may be painted on to the surface to be studied and the reaction examined by reflected light. It is noteworthy that, as we show in the text and illustrate in Figs. 4 and 5, Plate XIX, the only reagent which gives a satisfactory reaction with the epidermal oxydase is benzidine, and that benzidine and a-naphthol discriminate as it were between epidermal and bundle oxydases. The former is picked out by benzidine but is untouched or almost untouched by a-naphthol. The latter gives strong reactions with a-naphthol and benzidine. The behaviour of a-naphthylamine is very similar to that of o-naphthol. The colour of the reaction with benzidine is a rich brown due to the deposition of the oxidised product in the cells. Under certain circum- stances, and in early stages, this reagent produces a blue or blue-green colour which however passes more or less rapidly into brown. The reaction with a-naphthol takes the form of a delicate lilac blue or lavender colour, and that with a-naphthylamine, a pink. It remains to mention that the reagents should be made up fresh for use and that the flowers or other subjects to be investigated should be fresh, uninjured, and in good growing condition. Late formed flowers, for example, are apt to give uncertain results and this is in conformity with the fact that such flowers show often considerable departure from their normal colouration. The length of time of exposure to the several reagents varies with the object and can be decided only by trial. In the case of sections, the reaction takes place almost immediately ; but solid objects such as whole corollas require to be incubated for 1, 2 or more hours. p. Keeble and E. F. Armstrong 283 II. The Distribution of Oxydases^ in Plant Tissues. A. The Oxydases in the Vegetative Members of P. sinensis. The methods described in the previous section permit of the mapping out of oxydases in the tissues of a plant with very considerable accuracy. Hence it should be possible, by determining the distribution of pigment and of oxydase in a given species, to obtain evidence as to the validity of the hypothesis that the formation of anthocyan pigment depends on the action of an oxydase on a chromogen. For. if oxydase is specifically concerned in pigment formation a certain parallelism is to be expected between the distributions of pigment and of oxydase. Before describing the results of our observations on the localisation of oxydase in the tissues of Primula sinensis it is necessary to point out that we must not expect to find an exact coincidence of pigment and oxydase in each and every variety of this species. For, as shown by Bateson, Gregory, and others, the range of pigmentation in both vegetative and floral members of P. sinensis is very considerable. Some varieties have white flowers and others have coloured flowers of widely diflfering depths and shades ; some varieties possess green stems, others reddish stems, and othei-s, again, deep red stems. Certain white-flowered, green-stemmed varieties, such for example as Snow Drift (see Gregory, 1911) are lacking altogether in anthocyan pigment. In other green-stemmed varieties the small amount of sap- pigment which they contain is confined to special parts of the plant, for example the roots, root-stocks, and bases of the petioles where it occurs either in the epidermal cells only or in the sub-epidermal cells as well. Of such minimally pigmented varieties, some possess no pigment in their other vegetative parts, for example the flower- peduncle, others, for instance. Sirdars, contain anthocyan pigment in isolated epidermal and sub-epidermal cells of the flower-peduncle. Between the pure green-stemmed varieties and those with reddish stems is a series of forms characterised by a progressively widening distribution of epidermal pigment in the vegetative members ; and in the reddish stemmed plants the pigmentation is wellnigh continuous throughout the epidermal layer. At the other end of the colour series are the dark red-stemmed varieties in which anthocyan pigment is ^ For the sake of brevity we ase the term oxydase to connote both peroxydase and oxydase and reserve for sub-section G (p. 304) the description of the distribution of these bodies in the several tissues. 284 Oxydases and Pigments of Plants much more widely distributed than in the green and reddish stemmed forms. For example the flower-peduncle of the dark red-stemmed variety Mt. Blanc Star contains pigment not only in the epidermis and in three or more outer layers of the cortex but also in the tissues of the stele (vascular cylinder). If the investigator has these facts of variety of pigment distribution in mind he will not be surprised to discover when he proceeds to determine the distribution of oxydase that, although oxydase is present in every pigmented cell, it is not necessarily confined to these cells. This fact might appear at first sight to militate against the hypothesis which associates pigment formation with oxydase action; but further consideration shows that it does not. For it is evident that oxydase and chromogen, though they interact to form pigment, may be produced independently one of the other and therefore the failure of certain tissues, rich in oxydase, to produce pigment may well be due to the absence of chromogen. Or, to express the idea in other terms, chromogen and not oxydase is the factor the lack of which limits pigmentation in P. sinensis. The evidence offered by the known facts of the Genetics of this species lends weighty support to this conclusion and indeed leads directly to the expectation that one of the two agents concerned in the production of anthocyan pigment is more generally distributed than the other. For if the two agents, chromogen and oxydase, were equally sus- ceptible of restriction of distribution in the plant then it would be reasonable to expect that two types of green-stemmed plant exist; one, green-stemmed because it lacks chromogen, the other, because it lacks oxydase. In point of fact, and in spite of the large amount of breeding work which has been carried out with this species, only one type of green-stemmed plant is known in P. sinensis. The evidence from Genetics in support of this statement is clear. Some plants, for example Stocks and Sweet Peas, show by their genetical behaviour that either of two factors for pigment production may be lacking from a variety. In such plants, as Bateson, Miss Saunders and Punnett (1906) have proved, it is possible, by mating white-flowered individuals of the two types, to bring the complementary colour factors together and thus to produce a reversionary, coloured F^ generation. But with P. sinensis a like result has never been obtained. A pure green-stemmed variety mated with any other similar variety gives rise to a green-stemmed Fi generation, and pure breeding white-flowered varieties, each of which F. Keeble and E. F. Armstrong 285 lacks a factor for colour, give rise when crossed with one another to white-flowered offspring only. The simultaneous study of the genetical behaviour of P. sinensis and of the distribution of oxydases in the tissues of the plant provides an explanation of this behaviour. For the latter line of inquiry demon- strates that all varieties of P. sinensis, including the pure green-stemmed varieties, contain oxydase in certain of their tissues. Whence it follows that at least enough oxydase for a definite amount of pigment pro- duction is a possession common to all Chinese primroses. We have in this fact the explanation of the failure of green- stemmed varieties to yield a red-stemmed Fi when crossed with one another. It is also clear why the cross, green-stemmed variety by reddish-stemmed variety, yields a reddish-stemmed Fi generation. Thus if we represent the chromogen producing factor by C, its absence by c, and the oxydase producing factor by O, and its absence by o, then green stem x reddish stem = Oc X OC, and Fi = OOCc, and since both chromogen and oxydase are present in Fi the plants of that generation are pigmented. Fi plants of this kind produce gametes OC and Oc, and hence when such plants are self-fertilised the P, generation is of the typical mono- hybrid kind and consists of 30C : lOc, i.e. 3 reddish-stemmed plants: 1 green-stemmed plant. As we shall show immediately there is a similar agreement between the results of genetical inquiry and those obtained from the study of oxydase distribution in the case of red-stemmed varieties. But the broad features of the relation between oxydase and pigment may be seen in an equally striking manner when they are viewed from another standpoint. If, instead of fixing our attention on one variety only, we consider the distribution of pigment and oxydase in the different coloured varieties we obtain clear evidence of the truth of the hypothesis that oxydases play a part in pigment formation. For when we adopt this procedure we discover that, as we pass stage by stage from the less pigmented to the more pigmented forms, the new additions of pigment occur in those tissues which in unpigmented and pigmented forms alike are most generally rich in oxydase. In other words, the cells which in general contain most oxydase are prepotentially pigment forming cells. Thus, although in pure green-stemmed varieties neither the ordinary Jonm. of Gen. ii * 20 286 Oxydases and Pigments of Plants epidermal cells nor those of the epidermal hairs contain pigment, these elements give pronounced oxydase-reactions and they constitute the tissue to which pigment is confined in those pale reddish-stemmed varieties which have a minimum of pigment. Next to the epidermis, the sub-epidermal layer is, of all tissues outside the stele, richest in oxydase, and it is in the sub-epidermal layer that pigment occurs in the varieties somewhat more pigmented than those which belong to the category of pale reddish-stemmed forms. In the case of green- stemmed and reddish-stemmed varieties, the distribution of oxydase in a member of the series is indicative of the distribution of pigment in the next higher member of that series. It is only in dark red-stemmed varieties, e.g. Mt. Blanc Star, that the pigment extends to any considerable depth into the cortex. These varieties are remarkable also for the fact that their anthocyan pig- ments are not limited to the epidermal and cortical tissues systems, but occur also in those of the stele (vascular cylinder). As is the case with all pigmented tissues of the epidermis and cortex, the pigmented tissues of the stele give clear evidence of a high oxydase content. Therefore we have a further means of testing the hypothesis; for by analogy with the distribution of oxydase in the progressively less pigmented members of the colour series we may expect to find that the localisation of oxydase in those forms which are without pigment in their stelar tissues coincides with that of the pigment in the corre- sponding tissues of the dark red forms. Appropriate examination shows that this anticipation is realised. For example, the pigment in the stelar tissues of the flower-peduncle of a dark red-stemmed plant occurs in the pericycle and in patches of pith cells which lie against the protoxylems of the wood ; and these, together with the phloem, are the tissues which are rich in oxydase. In the root of a dark red-stemmed plant, large quantities of pigment occur in the pericycle, in the medullary rays which extend from near the periphery of the stele almost to the centre, and also in the phloem. The distribution of oxydase in the stele of the green and reddish varieties is very similar to that of the pigment of the red varieties. Thus the flower-peduncle of a green-stemmed plant contains oxydase in considerable quantity in the pericycle, phloem, and in the pith cells which abut on the protoxylems. We conclude therefore that those tissues which in non-pigmented forms are richest in oxydase are the tissues in which in coloured forms pigment makes its appearance. F. Keeble and E. F. Armstrong 287 Although there can be no doubt but that this broad relation exists between oxydase and potentiality of pigment formation, it is not to be concluded that all varieties of P. sinensis have identical oxydase content. Simple observations and experiments suffice to show that the extent of distribution of oxydase diflFers in different varieties. For example, in the superficial tissues of the flower-peduncle of Sirdar, oxydase is limited exclusively or almost exclusively to the epidermal layer; whereas the corresponding tissues of other green-stemmed varieties give a well marked oxydase reaction in both epidermal and sub-epidermal layers. There is good reason to believe that chloroplasts act as inhibitors of oxydase -formation in a cell and it is noteworthy that the sub-epidermal layer of Sirdar is specially rich in chloroplasts. Again in red-stemmed varieties, the petioles, which together with the roots appear to be the members richest in oxydase, are so rich in both chromogen and oxydase that pigment occurs in practically every cell ; in green-stemmed plants, though the roots are rich in oxydase, the cortical cells of the petioles give no oxydase reaction. These phenomena and others to be referred to immediately lead us to the opinion that such localisation of oxydase as occurs in P. sinensis is at least in large measure a phenomenon of inhibition. We shall see in a later .section, that, as postulated by Mendelians, flower colour may be inhibited. We know that the result of crossing a dark red- and a reddish-stemmed plant is the production of a reddish ^i, and that the F^ generation from this cross consists of 3 reddish : 1 dark red-stemmed ; that is, dark red stem behaves as a simple recessive to reddish stem. Nevertheless, as we have just seen, red stem differs from reddish stem in possessing pigment in certain cortical and stelar tissues which are not pigmented in the reddish-stemmed plants. This being so, it would be expected that the presence of pigment would be dominant to the absence and hence that red stem would be dominant to reddish stem. Since the reverse is the case we can scarcely escape the conclusions that reddish stems contain an inhibitor and that this inhibitor is not powerful enough to suppress pigment formation altogether though it suffices to suppress it in certain tissues. It remains to ask whether partial inhibition of pigment formation is to be attributed to inhibition of the action of oxydase on the chromogen or to inhibition of the processes which lead to the formation of chromogen. We are not in a position to answer this question with respect to stem colour though, as we show in the following section, we can answer it 20-2 288 Oxydases and Pigments of Plants with respect to inhibition of pigment formation in the flower (see page 301). Some of the evidence which we possess as to the nature of the inhibition of pigment formation in the vegetative parts leads us to favour the view that this inhibition applies to the production of chromogen rather than to the activity of oxydase. This evidence is derived from the oxydase reaction given by the outermost cortical layer and various tissues of the stele of certain green-stemmed varieties. Instead of no, or at most a weak, reaction, which is to be expected if substances inhibiting oxj'dase action are present, the tissues just enumerated give a particularly strong oxydase reaction and hence it would appear probable that these cortical and stelar tissues contain an inhibitor which is capable of preventing the production of chromogen. On the other hand the cortical tissues of the petioles of certain green-stemmed varieties give no reaction for oxydase although the corresponding tissues of red-stemmed forms are extraordinarily rich in that substance, and in this instance the failure to give the reaction for oxydase may be due to the inhibition of the latter substance. Whether inhibition be of the nature of the suppression of oxydase activity, as we know it to be in the dominant white flowers of P. sinensis, or whether it consist in the prevention of chromogen production, it follows that negative results with respect to the oxydase content of a given tissue must be accepted with caution : this applies, of course, not only to qualitative results such as those with which we are dealing, but also and with even greater force to those obtained by quantitative estimates of the oxydase content of plant juices. Elaborate methods are in use for this purpose (cf. Bunzel, 1912) but, unless they are asso- ciated with methods for removing any inhibitors which may be present, the results which they yield must be accepted with reserve. To sum up our observations on the distribution of oxydase in the vegetative parts of P. sinensis : We find that the methods described in Section I serve for the faithful mapping out of oxydases in the several tissues of the plant : that this mapping out leads to the conclusion, that although oxydase is more widely distributed than chromogen, the distri- bution is in conformity with that required by the oxydase-chromogen hypothesis and that, owing to the existence of inhibiting substances, caution must be exercised in interpreting the negative results obtained by the use of oxydase reagents as proof of the absence of oxydases. F. Keeblb and E. F. Armstrong 289 B. The Oxydases of the Flower of P. sinensis. (1) Self-Coloured Varieties. The oxydases of the vegetative parts of Primula sinensis are located in two groups of tissues ; one group — the epidermal — ^is superficial, the other — the stelar — is deep seated. The "epidermal" oxydase is con- fined to the epidermis in certain green-stemmed varieties but extends to the sub-epidermal layer in reddish-stemmed varieties and reaches its widest distribution in the dark red-stemmed races in the peduncles of which oxydase occurs not only in the epidermis but also in the two or three outer layers of the cortex. Hence it follows that the epidermal oxydase is separated widely by the intervening cortical cells from the bundle oxydase. The epidermal and bundle oxydases of the vegetative members have their counterparts in the flower ; but, inasmuch as the cortical tissues of the corolla consist only of some two layers of flattish cells, the epidermal and bundle oxydases of the petals lie in close proximity with one another. Nevertheless, and in spite of their proximity, it is possible to demonstrate macroscopically the presence of both oxydases in the petals. The discrimination between the two oxydases is aided by the fact that they do not react in precisely the same way to our reagents — a-naphthol and benzidine. The former reacts much more quickly and in most cases exclusively with the bundle oxydase to produce a lavender blue colour which picks out the veins in exquisite detail and for the most part leaves the epidermal oxydase unaffected. The selective action of benzidine is less precise. This reagent reacts with the epidermal oxydase to produce a rich brown colouration of the superficial layer of the petals and also produces a similar though darker colouration in the veins. Examples of the colour reactions which are obtained by the use of these reagents are given in Figs. 2, 4, 5, 11, 12, 14, Plate XIX. The coloured (blue) flower shown in Fig. 1 yields with benzidine the reaction illustrated in Fig. 2. Recessive whites give a precisely similar reaction (see Fig. 5) with this reagent. Unlike benzidine, which reacts with epidermal and bundle oxydase, o-naphthol reacts with the latter only both in the case of coloured and of recessive white flowers. Hence as shown in Fig. 4 the veins stand out prominently on an almost unstained ground. The central part of the corolla of these varieties is characterised, as are many other varieties of P. sinensis, by a yellow eye, the colour of which 290 Oxydases and Pigments of Plants is due, not to anthocyan, but to plastid pigments ; and it is a note- worthy and general fact that the region of the yellow eye shows no oxydase reaction except in the epidermal hairs which give with benzidine a deep brown-black colouration. The failure of the oxydases of this region of the eye to react with a-naphthol or benzidine is to be attributed to the inhibition of oxydase by the chloroplasts. We have investigated the oxydase contents of the corollas of many other colour varieties of P. sinensis, e.g. Crimson King, Coral Pink and Giant Red among the reds ; various magentas and lavenders, e.g. Giant Lavender ; and Czar, Cambridge Blue, etc. among the blues, and though the extent of the reaction of epidermal and of bundle oxydases varies considerably in the several varieties it is characteristic of them all. Of white-flowered varieties of P. sinensis, genetical research has shown that there are two kinds, which are known respectively as Recessive Whites and Dormant Whites. (2) Recessive White Varieties. The Recessive Whites show by their behaviour when crossed with coloured varieties that they lack a factor for colour. When crossed with a coloured variety they yield a coloured F^ which on self-fertilization gives rise to an F^ generation composed of 3 coloured : 1 white. The usual and evidently proper interpretation of this result is that recessive whites lack a factor for colour which is possessed by the pig- mented varieties. The cross is therefore to be represented thus: c X C F, = Cc i^2 = 3C:lc = 3 coloured : 1 white. As we should expect from our study of the oxydases of the vegetative members of P. sinensis that which is lacking from recessive white flowers is not the oxydase forming factor but the factor for chromogen produc- tion. Treatment of the petals with benzidine demonstrates that this expectation is correct, for, as indicated already, the result of the treat- ment is a well marked oxydase reaction in both epidermis and veins (Plate XIX, Fig. 5). Whence we conclude that since the corollas of recessive whites contain both epidermal and bundle oxydases their lack of colour is due to the absence of the factor for chromogen pro- duction. Of all the varieties the flowers of which we have examined only those belonging to three categories show any departure from the F. Keeble and E. F. Armstrong 291 general rule that epidermal and bundle oxydases are present in the corollas. These three categories are the dominant white, blue with white inhibitory patches and the flaked varieties. (3) Flaked {Ever-sporting) Varieties. We will deal first with the flaked varieties. Snow King and Mt. Blanc Star, the varieties of P. sinensis which we have investigated, bear white flowers marked more or less prominently by splashes of magenta. Sometimes a whole petal is magenta coloured and sometimes a magenta flower appears among neighbouring magenta flaked flowers. Although the amount of flaking varies very considerably the races breed wellnigh true to this habit. Thus Mt. Blanc Star produces oflfspring the great majority of which are flaked ; but it throws occasionally a plant all the flowers of which are magenta coloured. These magenta flowered plants may bear darker flakes of magenta on a lighter coloured ground and the numbers in which they appear are said to be about two per cent. The genetics of these flaked forms which is in course of investigation by one of the present writers (see Keeble, 1910 a) need not concern us here except in so far that it provides evidence of the exist- ence in the petals of a partial inhibitor of pigment formation. Magenta-flaked white flowers of Mt. Blanc Star give much fainter and far less regular oxydase reactions than are exhibited by any coloured or recessive white varieties. They provide also a good illustration of the relation between oxydase and pigmentation. For, as is exemplified in the text-figure (Fig. 1), if, before treating the flower with the oxydase reagent, the pigmented areas are recorded it is found that the distribution of oxydase coincides very closely with that of the pigment. In the case depicted in the text-figure a flower was chosen which had, in addition to certain irregular magenta flakes, one petal of a uniform magenta colour. A comparison of the distribution of pigment with that of the oxydase shows that the magenta petal gave a well marked oxydase reaction, that the magenta patches on other petals were also the seat of a fair amount of oxydase and that the white areas gave no reaction. The absence of oxydase-reaction from the unpigmented parts of the flower is noteworthy because it is the first piece of evidence which we have been able to produce to show that failure to form pigment may be connected with failure to yield the reaction for oxydase. Flowers of this kind, characterised by a considerable degree of fluctua- tion of colour and by the localisation of such colour as they may have 292 Oxydases and Pigments of Plants in flakes or spots, are very common among cultivated plants, for example. Azaleas, Sweet Williams, Stocks and Carnations. They are known as ever-sporting and their genetical behaviour is difficult of interpretation. The observations of which we have just given a brief account indicate quite definitely that in the case of the white magenta-flaked flowers of Mt. Blanc Star the ever-sporting habit is associated with irregularities in amount or activity of oxydase in the tissues of the petals. Fig. 1. The coincidence of peroxydase with pigment in the white magenta-flaked flower of Mt. Blanc Star. Diagrammatic. The distribution of pigment in the several petals was recorded with reference to the incision indicated atZ) (a). Petal A was uniformly magenta coloured, B and C indicate the position of magenta flakes. The peroxydase distribution is shown in fig. 6. The copious wound peroxydase is indicated by the black patch, the position of which coincides with D which marks the place of the wound. (See text and Text-figure 5.) Inasmuch as the problems presented by ever-sporting varieties are of considerable interest, we have extended our inquiry with respect to them to other species of plants. We were the more anxious to do so because, except for the difficult case of white magenta flaked varieties, in which, as we have seen, the amount of oxydase-reaction is small, we had discovered no race of P. sinensis the flowers of which lack oxydase. Now as pointed out already, this fact, though it might be predicated from our knowledge of the genetics of P. sinensis, is less likely to obtain with respect to certain other plants. Thus, the genetical behaviour of Sweet Peas is such that two distinct factors for colour must be assumed if this behaviour is to be accounted for in terms of Mendelian hypothesis. Hence the suggestion is bound to present itself that the two factors in question are an oxydase-producing and a chromogen-liberating factor. We have therefore an added reason in seeking among plants other than P. sinensis, for an example of a flower which lacks oxydase. F. Keeble and E. F. Armstrong 293 C. The Oxydases of Dianthus barbatus (Sweet William). (1) Ever-sporting Varieties. The nearest approach to a flower of this description which we have found so far is that of the Sweet William (Dianthus barbatus). This species presents a wide range of variety of flower colour. Pure white- flowered races as well as races with purple, red and salmon colour are to be found in almost every garden. In addition to races with pure white flowers, others occur in which the flower is white except for a ring of fine pink dots or lines across the middle of the petals. Ever-sporting varieties are likewise common. They bear on one and the same inflor- escence flowers of very dififerent colours. Thus in the race with which we have experimented the colours of the flowers of a single plant were deep magenta, strawberry, pale and streaked pink on a white ground and white. In some of the white flowers a small amount of rose coloured pigment occurs a little below the middle of the limb of each petal, and in others the amount of pink colouration is so small that the flowers are almost pure white. If this series of dififerently coloured flowers of an ever-sporting variety of Sweet Williams be examined for oxydase it is discovered that the amounts of oxydase present in the petals of the several members of the series are strictly proportional to the amounts of pigment in those members. As illustrated in Text-figure 2, all the coloured forms contain both A Fig. 2 B CD The oxydase (benzidine) reactions of the flowers of an ever-sporting variety of Dianthus barbatus (Sweet William), illustrating the parallelism between pigmentation and oxydase content. Flower-colour : A Deep red magenta. B. Light red magenta (strawberry colour). C. Pale rose, blotched. D. White with trace of rose in centre. [The limbs of the petals contain no pigment and no oxydase. ] 294 Oxydases and Pigments of Plants epidermal and bundle oxydases; the most deeply coloured flowers contain most, those of intermediate tint, less, and those which approach most nearly to whiteness, contain the least amount of oxydase. The coincidence of oxydase with pigment in the all but white forms of flower is most impressive. In such forms the pigment is confined to three short rosy lines at the base of the limb of each petal and it is only along these lines, which mark the position of the main veins, that oxydase is found. These approximately white forms give but a light brown reaction with benzidine ; the pink pigmented forms give a rich brown and the fully pigmented forms give a brown-black reaction. Hence we conclude that in this example of an ever-sporting flower the extent and depth of pigmentation are determined by the quantitative distribution of oxydase. (2) White Varieties of Sweet William. Among the various, non-sporting varieties of Sweet Williams which we have examined are races with fully coloured, white and nearly white flowers. The fully coloured varieties all give pronounced oxydase reactions ; the white and approximately white varieties behave to our reagents in one of two ways. The pure white race gives a very definite, albeit limited, oxydase reaction which is most pronounced in the central region towards the base of the petal limb. This variety is therefore colourless because it lacks chromogen. The all but white race, the petals of which bear a ring of rosy dots or lines about one-third of the way from the point of junction of limb and claw, contains no oxydase except in the region which in the fresh flower is occupied by the rosy ring. Hence this variety, unless indeed it prove to be a dominant white, owes its whiteness to lack of oxydase. The final establishment of these conclusions must await the results of breeding-experiments which are being carried out. Should these experiments lead to the production of a coloured F^ generation, we shall have the proof of the hypothesis suggested first by Miss Wheldale, that, where two colour factors are involved in the production of pigment, one is a chromogen producing factor and the other an oxydase producing factor. But apart from these experiments we have in the cases of white, magenta flaked P. sinensis and of the various races of Sweet William clear evidence that pigment formation depends on oxydase action, that depth of colour is determined by amount of oxydase and that a lack of this substance results in an absence of pigmentation. F. Kbkble and E. F. Armstrong 295 D. The Oxydases of Geranium sanguineum. We may cite further evidence in favour of the major proposition that the two factors for pigment formation are chromogen and oxydase. As has been shown by genetical research, with such plants as Sweet Peas (Bateson, 1906) and Orchids (Hurst, 1909), so we show by means of chemical methods that albino races are of two kinds. The white flowered races of Sweet Peas {Lathyi-us odoratus) and of culinary peas {Pisum sativum), so far as we have examined them, all contain oxydase of the type present in the coloured forms and hence the lack of colour in these races of the two species is due to lack of chromogen. With respect to the former species, however, it should be remarked that we have not yet had an opportunity of investigating thoroughly the two types of white flowered plants which yield when crossed with one another a coloured ^i. So far as our observations go at present we have found, neither in round-pollened nor in long-pollened albino Sweet Peas, no flower from which oxydase is absent. It maybe that further search will discover such flowers or it may of course be that the factors determining pigmentation are other than those suggested above. In any case the evidence provided by the albino of another species of plants. Geranium sanguineum, is favourable to our hypothesis. The petals of the flower of the purple type of G. sanguineum give with benzidine a definite epidermal oxydase reaction and a yet more marked reaction for bundle oxydase. The pure white petals of the albino variety give with the same reagent a distinct bundle reaction, but no epidermal reaction. Hence this white form is either a dominant white or a true albino of the second type, namely one which owes its whiteness to deficiency of oxydase in the epidermal cells. Although we have been unable, owing to lack of material, to determine absolutely to which of these categories it belongs, the fact that the bundles give a definite oxydase reaction appears to indicate that the white G. sanguineum is not a dominant white. The intermediate form, Geranium, lancastriense, which is characterised by a flower of pale flesh colour with darker pink veins stands midway between the type and the albino ; for although it gives no or at most a very slight oxydase reaction in the epidermis of the flower, it gives a very distinct (violet-brown) bundle reaction. It would appear therefore from the observations on ever-sporting races Primula sinensis and Dianthus harhatus and on the albino forms of Sweet Peas and of Geranium sanguineum that two types of albino 296 Oxydases and Pigments of Plants exist. In the one type the white flowers are rich in oxydase and from the other, as judged by the reactions with benzidine and a-naphthol, oxydase is absent. We hope by the application of these methods to the flowers of the many albinos which occur in cultivation and in the wild state to ascertain whether one or the other type is of more frequent occurrence in nature. E. The Localisation of Oxydases in the Tissues of the Flower {Primula sinensis). We return now to Primula sinensis. In order to complete our survey of the oxydases of the flower, and before proceeding to discuss the oxydase content of dominant white flowers, we will describe briefly the localisation of the oxydases present in coloured and recessive white forms. The distribution of the epidermal oxydase in the petals is definitely circumscribed. The epidermal oxydase is confined to the epi- dermal cells and the hairs which are outgrowths from these cells. The distribution of the bundle oxydase is less sharply defined. By treat- ing tangential sections of the petals with the reagent, oxydase may be seen to follow the veins throughout their whole course and to extend to their finest ramifications. If a section which exposes one of the finer veins be treated with benzidine and then examined micro- scopically the brown granular or needle-like products of the interaction of the reagent and the oxydase are seen as dense masses in the elongated cells of the bundle sheath — which cells are more deeply stained than any others. These elongated cells are wrapped around and also extend beyond the tracheids of the veins. They give off short branches which make contact with corresponding branches of the stellate parenchyma of which the body of the petal is composed (see Text-figure 3). In petals rich in oxydase the stellate parenchyma cells, the branches of which abut on those of the bundle sheath cells, also give oxydase reactions, sometimes as marked but generally less marked than those of the cells of the bundle sheath. The reaction in the cells of the stellate parenchyma becomes more faint as the cells of that tissue are traced further from the veins (see Text-figure 3). The appearance presented by the stellate parenchyma and the bundle sheath is as though oxydase were passing from the cells of the latter to those of the former. We have given reason in a former communication (Keeble and Armstrong, 1912 b) for our belief that oxydase may be translated from cell to cell, and have offered the suggestion that the frequency with F. Kkkblb and E. F. Armstrong 297 which lines of colour are seen to demark the veins of petals is attributable to the action of oxydases which pass from the bundle to the epidermis. Fig. 3. The oxydase content of the cells of the handle sheath of the terminal portions of a vein in a petal of P. sinensis. b.s. Bundle sheath — rich in oxydase. s.p. Stellate parenchyma, the cells in contact with those of the bundle sheath are rich in oxydase : those more distant contain less. tr. Tracheid. i.ct. Intercellular spaces. If preparations of sections treated Mrith the benzidine reagent are examined microscopically, the localisation of the brown product of oxydase action may be seen in the cytoplasm of the tissues rich in 298 Oxydases and Pigments of Plants oxydase. It happens very frequently that the walls of these cells take on also a brown stain. Although this staining of the wall may be due to diffusion of the oxidized benzidine derivative from the cell, yet the possibility is not precluded that in the living plant oxydase may occur in the walls of the cells. We are not in a position to make a positive statement on this subject but may point out that changes, often of a remarkable nature, go on in cell walls and in some cases, for example, in the macrospores of species of Selaginella very considerable changes of structure and dimensions occur in cell walls which are far removed from contact with cytoplasm. Therefore it would seem not unreasonable to suppose that oxydases may play a part in the growth processes of cell walls, and that the occurrence of wall staining after the treatment of a tissue with an oxydase I'eagent may be due to the formation in situ of the colour substance indicative of oxydase and not to the passage of that substance from the cytoplasm to the wall. F. Dominant White Varieties. We have now to consider the oxydase content of those varieties of P. sinensis known to students of Genetics as Dominant Whites. As the term implies. Dominant Whites present the appearance of albinism but, as it also suggests, the albino-like appearance masks a distinct and remarkable genetical constitution. Hitherto the only test serving to discriminate between Recessive and Dominant Whites has been the breeding test. Recessive Whites crossed with coloured varieties give a coloured F^^ generation ; Dominant Whites when mated with similar coloured varieties yield a white F^ generation. The F^ plants from the cross, recessive white x colour, give on self-fertilization or on inter- breeding an F2 consisting of 3 coloured : 1 white. The F^ generation derived from self- fertilized or inter-bred F^ plants from a cross between coloured and dominant white consists of 3 white : 1 coloured. The phenomenon of dominant whiteness does not appear to be very common among plants. It is exemplified by the flowers of Primula sinensis (Bateson, 1906), and the Foxglove (Digitalis purpureus, Keeble, 1910 b). Among animals a similar phenomenon occurs with respect to coat colour in certain breeds of fowls, for example White Leghorns (Bateson, 1906). The interpretation which meets the facts of the genetical behaviour of Dominant Whites is well known and involves the existence of a factor which, even though the factor or factors for colour be present, prevents the development of pigment. In other F. Keeble and E. F. Armstrong 299 words the genetical hypothesis of the nature of dominant whites holds that the latter are due to the inhibition of pigment formation. This Mendelian interpretation is so convincing that it scarcely needs the collateral support to be derived from chemical investigation ; but, although this be the case, it is at once evident that the hypothesis suggests a promising line of inquiry into the physiology of pigmentation. Investigations on this subject made by Miss Wheldale (1910) supply confirmation of the hypothesis by indicating that an inhibitor of pigment formation exists in flowers the pale shades of which are dominant to deeper shades. A further advance has been made by Gortner (loc. dt. 1911) who has produced experimental evidence confirming the conclusion reached by previous investigators, that the black pigment (melanin) of various insects and other animals is produced by the interaction of the oxydase, tyrosinase with a chromogen, tyrosin, and has shown that, when these substances are allowed to act in vitro, the addition of certain dihydroxy phenols such as phloroglucinol, orcinol and resorcinol prevents the reaction from taking place. These substances which exercise an in- hibitory action on pigment formation Gortner terms antioxydases. He suggests further that the inhibition of melanin formation in animals may be due to a chemical change in the chromogen (tyrosin). Gortner's views may be expressed schematically thus : Let T = Tyrosin, and T"** = Tyrosinase. Then T+T«««= Melanin, and T+ derivative of T4- T*"* = Inhibition (Dominant White). We proceed now to consider dominant white flowers of P. sinensis with respect to their behaviour with oxydase reagents. The addition of a-naphthol or benzidine and the subsequent addition of hydrogen- peroxide to intact corollas of dominant white plants results in no oxydase reaction whatever. Even after prolonged action of the reagents both epidermis and veins fail to give the colouration characteristic of oxydase. It follows therefore that either oxydase is absent from the flower or it is inhibited from oxidizing the reagents. It so happens that among the Primulas which one of us has been breeding at Reading is a strain derived from a cross between a pure blue-flowered variety and the white, magenta-flaked Snow King to which reference has been made already. Among the descendants of this cross are certain plants illustrated in Plate XIX, Figs. 10 and 13, the flowers of which are 300 Oxydases and Pigments of Plants characterised by fairly regular and symmetrically placed white areas on an otherwise uniformly blue ground. We have gi'ound for believing, both on account of the origin and of the genetical behaviour of these plants, that the white areas represent what may be called inhibitory patches. It is further to be mentioned that these white-zoned blue flowers exist in two forms. In the one form, all the flowers of the plant are marked symmetrically with the white areas and remain in this state as long as they last. In the other form, certain of the flowers and par- ticularly those which are produced late in the season show a blurred, pale blue colour extending from the blue perimeter into the white areas. The plants which exhibit the white areas sharply and per- manently are homozygous, that is breed true to the character; those in which the areas tend to be blurred are heterozygous for the character ; that is to say, they throw plants which bear uniformly blue flowers as well as others with blotched blue flowers. The results of an investigation of the distribution of oxydase in the corolla of the true breeding, white-zoned blue flowers are depicted in Plate XIX, Figs. 11 and 12, and show in most striking and definite manner that whereas both the epidermis and vascular bundles of the blue areas give well marked oxydase reactions, no such reactions are given by the white areas. Careful observation of the preparations indicates (see Figs. 11 and 12, Plate XIX) that, although they exhibit no sign of epidermal reaction, a faint bundle reaction may be traced in some cases along the veins of the white areas. From the eye of the corolla to the edge of each white patch, the bundle reaction is very distinct; but as soon as the bundle enters the white area the reaction becomes either imperceptible or at most very faint ; and when the veins pass from the white area into the blue region of the petals, the bundle reaction resumes its distinctness. A comparison of Figs. 11 and 12, Plate XIX with the illustration (Fig. 2, Plate XIX) of the oxydases of a uniformly coloured blue flower shows how remarkable is the definition of the white areas of the blue, white-patched petals. The curious attenuation of the bundle oxydase reaction over the white area recalls that which occurs over the yellow eye of the flowers of P. sinensis. Dominant whites and the white areas of the white-zoned blue flowers are alike in that they yield no oxydase reactions, and they stand in this respect in marked contrast with Recessive whites which as we have shown give pronounced oxydase reactions. Recessive whites F. Keeble and E. F. Armstrong 301 contain oxydase but lack chromogen. Dominant whites would appear from our biochemical investigation to lack oxydase. Such a conclusion however is not compatible with the known results of crossing Dominant and Recessive whites, and we are driven therefore to seek the interpre- tation of the behaviour of the dominant white races in terms of inhibition of oxydase action. It seemed possible that — if inhibition occur — the suppression of oxydase reaction over the white areas might be due to the presence of sugar ; but the application of appropriate tests indicates that reducing sugars are present very generally in the petals of those varieties, coloured and recessive whites, which give good oxydase reactions. We conclude therefore that inhibition is not due to the presence of reducing sugars. Those (heterozygous) plants in which the white areas are ill defined, show the same readiness to give a colour reaction with benzidine or o-naphthol as to develop a blue colouration when in the living state: cf Figs. 13 and 14, Plate XIX, Just as the blue ground colour of the petal may spread into the inhibitory area in the heterozygous plants, so the benzidine reaction may extend over the boundary be- tween the blue, where it is well marked, and the white from which, at first, it is absent. Here again the suggestion of inhibition forces itself on the observer ; for the phenomena are paralleled by those exhibited by heterozygous dominant whites. For whereas pure bred dominant whites have pure white flowers, plants which are heterozygous for the inhibition factor may show a faint tinge or flush of colour over the surface of their petals. The results obtained by the investigation of the oxydase contents in dominant white and white-zoned blue flowers, though they suggest the presence of an inhibitor of pigment formation do not, of course, supply actual proof thereof, nor do they indicate the mode of action of the supposititious inhibitor. If the inhibitory substance exist, it should be possible either to remove or destroy it. If we assume the existence of an inhibitor of pigment formation, then we must suppose that it acts in one of two ways. It either inhibits chromogen formation or it checks the action of oxydase. The evidence which we have just produced points to the latter mode of action, for dominant white and the white- zoned blue flowers are unique among varieties of P. sinensis in not giving an oxydase reaction. We will assume therefore that the inhi- bitor exercises its influence on oxydase. That influence may be brought to bear in one of two ways : either the oxydase may be destroyed or it may be prevented from doing the work of pigment formation. If the Joarn. of Gen. ii 21 302 Oxydases mid Pigments of Plants oxydase be destroyed, the only method by which inhibition may be demonstrated must consist in the isolation of the inhibitor, and the addition to it of oxydase. If the latter be destroyed, the evidence for inhibition is forthcoming. If, on the other hand, the inhibitor does not destroy, but only checks the action of oxydase, it should, perhaps, be possible to discover a reagent which, whilst acting destructively on the inhibitor, leaves the oxydase unharmed. Could this be done, the addition of the appropriate reagents should demonstrate the existence of oxydase in the dominant white flower. We have pursued our inquiry along the lines suggested by these reflections and have succeeded in demonstrating experimentally the existence of an inhibitor of oxydase in the flowers of Dominant whites. As the result of experimenting with various substances, we have found that dominant white flowers, after preliminary treatment with certain reagents, are no longer refractory to benzidine or a-naphthol, but give with them well marked reactions for epidermal and bundle oxydases. The reagent which is most efficient in producing this result is hydrogen cyanide. Thus, if a dominant white flower be treated for 24 hours with an aqueous solution of hydrogen cyanide and then washed with water and treated with the oxydase reagent, it gives a uni- form and marked oxydase reaction. A comparative examination of Figs. 6 and 8, Plate XIX, shows how striking is the difference in behaviour with respect to oxydase reaction between a dominant white flower placed directly in the oxydase reagent and a similar flower treated previously with hydrogen cyanide. Whereas the former shows scarcely a sign of oxydase, the latter proves itself by the reaction to be rich in that substance. The most satisfactory results are obtained when the following method is practised : Immersion of the dominant white flower in a 0*4 per cent, solution of hydrogen cyanide for 24 hours, washing with water, addition of benzidine (alcoholic solution), washing and then adding to the water a drop of hydrogen peroxide. Instead of hydrogen cyanide, a saturated solution of carbon dioxide may be employed, but the removal or the destruction of the inhibitor takes place more gradually with this reagent than with hydrogen cyanide (see Fig. 7, Plate XIX). Our white-zoned blue flowers provide us with material for verifying the conclusion that hydrogen cyanide brings about the removal or destruction of the inhibitor. If a blue flower with inhibitory patches be treated with hydrogen cyanide in the manner described above, F. Kkkhlk and E. F. Armstrong 303 the subsequent addition of the oxydase reagent brings about a uniform colouration of the petals. The originally white areas are now as deeply stained as the blue regions, and the veins in the former, which as we have seen yielded at most a faint oxydase reaction give after this treat- ment as marked a reaction as those in the blue areas of the flower. The results confirm the Mendelian hypothesis that dominant white flowers owe their lack of pigment to the presence of an inhibitor of pigment formation. They show moreover that the inhibition is exercised, not on the process which results in the liberation of chromogen, but on the oxydase and that the inhibitor acts not by destroying oxydase but by effective interference with its action. Attention may be drawn to the fact that, as shown by the results obtained with dominant white flowers and contrary to the general opinion, hydrogen cyanide does not destroy oxydase. We may summarise our observations on the oxydases of the flower (sections B — F) thus : In all coloured and recessive white varieties of P. sinensis oxydase is present in the petals. It occurs in two situations, namely in the epidermis and in the bundle sheath of the veins. The epidermal and bundle oxydases react differently with our reagents (a-naphthol and benzidine). Dominant white varieties of P. sinensis contain an inhibitor of oxydase. On the removal of the inhibitor from the petals of dominant white flowers a strong oxydase reaction is obtained. The white areas in the petals of certain races of blue-flowered Primulas also contain an inhibitor which prevents the oxydase contained in those areas from reacting with oxydase reagents. When this inhibitor is removed the white areas give well marked oxydase reactions. Ever-sporting varieties of P. sinensis and of Dianthus barbatus show most epidermal oxydase in the most deeply pigmented flowers, less in the less pigmented and none in the white flowers. The albino forms of P. sinensis, Pisum sativum, Latht/Jtis odoratus (in the forms as yet examined) all contain oxydase and their floral albinism is attributable to lack of chromogen. The white-flowered Geranium sanguineum lacks oxydase and we are of opinion that it owes its albinism to lack of oxydase. Albino or approximately albino forms of Dianthus barbatus are of two forms ; in one form oxydase is present and from the other it is absent. 304 Oxydases and Pigments of Plants G. The Nature of the Oxydases in Flovjers. (1) The Oxydases and Per oxydases of Primula sinensis. As indicated in the footnote at the beginning of the section the term oxydase is used in the foregoing sub-sections (A — F) to include both oxydase and peroxy(iase, that is to say those oxidising substances which react directly with benzidine or a-naphthol (direct oxydase) and those which react with these reagents only after the addition of hydrogen peroxide (peroxydases). Our reasons for the practice which we have adopted are threefold : first to simplify description ; second because a given tissue may yield, in one variety, an oxydase reaction and, in another variety, a peroxydase reaction ; and third because there is evidence that a tissue of a given variety at one time and under one set of circumstances may contain peroxydase, and at another time and under other circumstances may contain a complete oxydase — namely peroxydase plus organic peroxide — and so give a direct reaction. It is necessary therefore to describe the results of our observations on the nature of the oxydase content of the various tissues of P. sinens'is and of the other plants which we have investigated. In the case of P. sinensis direct oxydases are relatively rare and peroxydases are the rule. Thus the oxydase reactions of the petals of coloured and recessive white flowers are invariably or all but invariably indirect ; in other words peroxydases are present in the flowers both in the epidermis and in the tissues of the veins : hence we may speak of epidermal peroxydase and bundle peroxydase as present in the petals of the flower. The only exception to this rule occurred in the case of the flower of a Sirdar, one petal of which gave on one occasion a well marked direct oxydase reaction. We have no evidence, except in the single example just recorded, of the existence of direct oxydase in the flower of P. sinensis. We have satisfied ourselves, however, that it occurs in the vegetative parts of normal plants. Thus sections of flower-peduncles of dark red-stemmed varieties, of green-stemmed Sirdars and of reddish varieties, may all possess direct oxydase in the phloem. In Sirdars and in dark red-stemmed varieties there is evidence that the pericycle contains a direct oxydase. On the other hand it is very rare indeed for the epidermal tissues of plants grown under normal conditions to exhibit a direct oxydase reaction. We have obtained such reactions occasionally in certain red -stemmed varieties; for example sections of flower-peduncles treated with benzidine may give the colouration characteristic of the oxydase reaction in the epidermal cells as well as in the phloem. F. Kkkble and E. F. Armstrong 305 For reasons which will be apparent immediately there is no need to attempt to discriminate more closely between the distribution of per- oxydase and that of oxydase in the normal flower. What is evident and of importance is the fact pointed out already by Miss Wheldale and by Clark that peroxydase is more widely distributed than the organic peroxide which activates it. Even in the case of the phloem which, as we have seen, gives a direct reaction there appears to be more per- oxydase present than the organic peroxide is capable of activating ; for the result of adding hydrogen peroxide to sections which have already given a direct reaction, is to increase that reaction. Hence it would appear that the eflSciency of the system, peroxydase and organic per- oxide, is determined by the extent to which the organic peroxide is present, and that whereas peroxydase is relatively stable, the organic peroxide is unstable. We have satisfied ourselves not only that this is the case but that external conditions play an important part in determining the amount of peroxydase present in the tissues of a plant. (2) The Oxydases and Peroxydases of Other Flowers. Before proceeding to justify this statement we will describe briefly the nature of the oxydases in the various plants to which reference is made in the preceding pages and in certain other species which we have had occasion to examine. Flowers of the Sweet Pea and of the Culinary Pea contain no direct oxydase in their epidermis and either none or a very small amount in the veins. They give, however, well marked peroxydase reactions. Geranium sanguineum possesses direct oxydase both in the epidermis and bundles of the petals. The pale flesh-coloured variety G. lancastnense contains direct oxydase in its veins and peroxydase in its epidermis. The white variety contains in its epidermis neither oxydase nor per- oxydase, but its bundles give a distinct peroxydase reaction. Coloured varieties of Sweet William give good oxydase reactions ia both bundles and epidermis. Species of Prunus and Pyrus are of interest in this connection. Those species the flowers of which are white, and remain white on fading, contain peroxydase only; those species which turn brown on fading contain direct oxydase. The very general phenomenon of browning presented by dried plants is to be regarded in all probability as an indication of the presence of a direct oxydase, and in order to prevent the discolouration of herbarium 21—3 306 Oxydases and Pigments of Plants specimens some method must be devised whereby the oxydase is destroyed. It is probable that the method of preserving the colours of flowers by drying them in sawdust or sand at moderately high tem- perature owes its efficiency to a destruction of oxydase. (3) The Infliience of Light and Darkness on the Oxydase Content of Plants. We now turn to the consideration of the influence of external conditions on the oxydase content of plants. From our experiments with P. sinensis it would appear that light and darkness play an all- important part in determining the amount of oxydase present in plant tissues. As demonstrated by the facts which we record below, light exercises a destructive influence on the organic peroxide constituent of oxydase. Thus under normal conditions of illumination a tissue may give no reaction with a-naphthol or benzidine, though it yields a well marked peroxydase reaction when hydrogen peroxide is added. When, however, a plant is maintained in darkness for 24 hours or longer, the tissue corresponding to that previously tested gives a pronounced and direct reaction with the oxydase reagents. In illustration of this fact we may cite the following examples : — Sections of the flower-peduncle of a reddish -stemmed plant which had been maintained in normal con- ditions of illumination gave no direct reaction with benzidine. Similar sections from a sister plant, which had been kept in the dark for 48 hours, gave a distinct oxydase reaction in the epidermal hairs. Other sections of the same two plants treated with a-naphthol gave the follow- ing results : the illuminated plant — a direct reaction (faint lilac) only in the phloem ; the dark kept plant — a much deeper reaction in the phloem (rose-colour to almost black) and a fair reaction in the epidermis and hairs. The same experiment shows further that the effect of dark- ness is also to increase the peroxydase content of the tissues. Thus, if flowers of sister plants, one exposed to normal illumination and the other maintained in the dark, be tested for peroxydase the reaction of the latter is found to be so definitely more considerable as to be appreciated by macroscopic examination. Text-figure 4 is typical of the results which were obtained in numerous experiments. Whatever be the interpretation of these facts the phenomena them- selves are definite. Darkness leads to the formation of peroxide and to an increase of peroxydase. We cannot say whether the latter result is to be interpreted as being due to a destructive action of light on oxydase or whether it is to be regarded as a consequence of the F. Keeble and E. F. Armstrong 307 continuous using up of oxydase in the production of new pigment to replace that which, for all we know, may be continuously destroyed when the plant is exposed to light. It is a well known fact that conditions of illumination influence the amount of anthocyan pigment which occurs in a plant. For example, it is a common practice among horticulturists to enclose choice fruits such as grapes and apples in translucent paper bags and it is claimed that this expedient, beside protecting the fruit from insects, improves its colour. This, if true, would point to the con- clusion that light of high intensity exercises a destructive influence on anthocyan pigment. ^n Ai A. Bi B2 4A. 4B. Fig. 4. (From photographs.) The effects of Light and Darkness on the Peroxydase of Primula sinensis. Ai and Bi, normally illuminated plants. A2 and B-i, plants kept in darkness for 48 hours. A. A double, light magenta plant (8/2/2/11) : flowers treated with benzidine and hydrogen peroxide. B. A white magenta-flaked plant of Mt JBlanc Star : flowers treated with a-naphthol and hydrogen peroxide. The most casual observations in the garden show that the depth of colouration of flowers varies considerably during the course of develop- ment of the flower. Thus the petals of many varieties which are white in the mature stage are pink or pinkish in the bud stages and it is possible that we have in such instances an example of the pigment destroying action of light. Although an adequate discussion of the facts just recorded would be out of place in the present communication, it will be evident that if they are shown to be true generally these facts may have an important bearing on many phenomena other than those connected with the pig- mentation of plants. For if we ascribe to oxydases a general rdle in the metabolism of the plant as well as a special function in pigment pro- duction, the fact that the oxydase content of plant tissues waxes in darkness and wanes in the light may have bearings on the phenomena of periodicity which are at once such puzzling and general attributes of the Vegetable Kingdom. The work of Palladin (1911) and others 308 Oxydases and Pigments of Plants indicates that oxydases have such a general function as is postulated above and that they are part of the respiratory mechanism of the plant. Hence it is reasonable to suppose that the rate of respiration of a given cell is determined by the amount of oxydase present in that cell. If light exercise directly or indirectly a destructive action on the oxydase content of the cell the rate of respiration of the latter will fall off during the day and will rise again after a sufficient exposure to dark- ness has set going the oxydase secreting apparatus of the cell and allowed of the accumulation of peroxydase and organic peroxide. It may be that, beside the phenomena of periodicity referred to above, the remarkable respiratory phenomena presented by succulent plants are attributable to their diurnal rhythm of oxydase destruction. During the night the respiration of such succulent plants as Mesemhry- anthemum results in a smaller output of carbon dioxide than that which takes place during the day. Instead of the respiratory substances becoming completely oxidised, they produce incompletely oxidised bodies, namely, organic acids. During the day the normal respiration is resumed and the organic acids which have accumulated at night disappear. We propose to investigate this phenomenon in the light of our knowledge of the diurnal variation of oxydase. Lastly a brief reference must be made tb the influence of wounding on the liberation of oxydase. As is well known the effect of mutilating a tissue is to produce a speedy and copious liberation of oxydase. The benzidine reagent serves to indicate wound oxydase. That this is so is illustrated in Text-figure 5, which represents white corollas of Fig. 5. (From a photograph.) The wound peroxydase of Primula sinejisis. Each of the petals of a dominant white flower of P. sinensis (5 A) was stabbed in three places with a needle. The wounded corolla and also one from an uninjured flower (5 B) were subjected to the action of the benzidine reagent, then washed with water and treated with hydrogen peroxide. The unwounded corolla gave no peroxydase reaction : the mutilated corolla gave an intense reaction in the neighbourhood of each wound (see text). F. Kekblk and E. F. Armstrong 309 P. sinensis the surface of one of which was wounded in several places by means of the point of a needle. The flowers were transferred at once to the benzidine reagent and treated subsequently with hydrogen peroxide with the result that the wounded areas gave an intense peroxydase reaction. It is therefore evident that the wound oxydase is in this case of the nature of peroxydase. Experiments of a similar nature made with other flowers lead to the conclusion that when a peroxydase only is present in the coloured or white parts of the petals the effect of wounding is to bring about copious liberation of peroxydase, but when a direct oxydase is present wounding results in the liberation of direct oxydase. The main facts dealt with in the foregoing section may be sum- marised as follows : Cells in which anthocyan pigment is present contain oxydase in one of two forms, namely peroxydase or complete (direct) oxydase. The latter is found in the flowers of Sweet William (Dianthus harhatus), of Geranium sanguineum and certain species of Pyrus and Prunus. The former, of more general distribution, occurs in P. sinensis, Lathyrus odoratus, Pisum sativum and many other plants. The oxydase content of a plant varies with external conditions. A tissue of a normally illuminated plant contains less peroxydase than is contained in the corresponding tissue of a plant kept in darkness : and the organic peroxide constituent of the complete oxydase, though it may be absent from the normal plant, makes its appearance after that plant has been maintained for some time in darkness. The wound oxydases of plants resemble those which are concerned in the work of pigment production. Those plants which contain per- oxydase only, liberate, when lacerated, wound peroxydase and those which contain both peroxydase and organic peroxide show in their wounds the complete oxydase. For summaries of other subsections see pages 288 and 303. In conclusion we wish to express our thanks to Miss D. Richardson for her kindness in preparing the coloured drawings from which the figures of Plate XIX are reproduced ; to Messrs Sutton and Son and to Mr Macdonald, the Primula expert of that firm, for providing us with some of the material used in the coui-se of the experiments, and to Mr G. Coombs, Assistant Lecturer in Botany, University College, Reading, for the drawings represented in Text-figures 1, 2 and 3. 310 Oxydases and Pigments of Plants DESCRIPTION OF PLATE XIX. The illustrations are reproduced from water-colour drawings by Miss D. Richardson from natural flowers (Nos. 1, 3, 10 and 13) and from preparations made by treating the flowers with reagents for oxydases (see text). 1. A uniformly (self) coloured blue flower (natural colour). 2. The same after decolourisation with the benzidine reagent and subsequent treatment with hydrogen peroxide ; showing epidermal and bundle peroxydases. 3. A white-flowered form (natural). 4. A recessive white (a-naphthol and H2O2); showing bundle peroxydase. 5. A recessive white (benzidine and H2O2) ; showing epidermal and bundle peroxydases. 6. A dominant white (benzidine and H2O2). 7. A dominant white (treated first with carbon dioxide, then with benzidine, and sub- sequently with H2O2). Cf. with 6. 8. A dominant white treated with hydrogen cyanide — benzidine — H2O2 . Cf . with 6. 9. A dominant white treated with HCN, a-naphthol H2O2. 10. A blue with white inhibitory patches (54/2/1). F^ of cross between Cambridge blue and Snow King (natural colour). 11. The same after decolourisation with the a-naphthol reagent and subsequent treatment with H2O2 . 12. Another flower of 54/2/1 after decolourisation with the benzidine reagent and subse- quent treatment with H2O2 . 13. A heterozygous plant of the same race (54/2/1) natural colour. 14. The same after decolourisation with the benzidine reagent and subsequent treatment with H2O2. REFERENCES. 1883. Pick, H. "Ueber die Bedeutung des rothen Farbstoffes bei den Phanero- gamen u. die Beziehungen desselben zur Starkewanderuug." Bot. Centhl. 1883, Vol. XVI. 1904. Durham, Florence M. " On the Presence of Tyrosinases in the Skins of some Pigmented Vertebrates — Preliminary Note." Proc. Roy. Soc. 1904, Vol. Lxxiv. p. 310. 1906. Bateson, W., Miss E. R. Saunders and R. C. Punnett. " Reports to the Evolution Committee, Royal Society, London." Report III, p. 18. (See also Report /, 1902.) 1908. Palladin, W. " Die Verbreitung der Atmungschromogene bei der Pflanzen." Ber. d. Bot. Oes. 1908, Vol. xxvi. p. 378. 1908. . " Die Bildung der Atmungschromogene in der Pflanzen." Ihid. p. 389. 1909. Hurst, C. C. " Inheritance of Albinism in Orchids." Gard. Chron. 1909, Vol. XLV. p. 81. 1909. Kastle, J. H. " The Oxidases." U. S. Treasury Dept. Hygienic Laboratory. Bulletin, No. 59, December 1909. JOURNAL OF GENETICS. Vbt.ll. N? 3. PLATE XIX ^^^^^ E Vilaon.C&mbruigo F. Kkeblk and E. F. Armstrong 311 1910. Armstrong, H. E. and E. F. " The Function of Hormones in stimulating enzjmic change" Proc. Roy. Soc. 1910, VoL Lxxxil. B, p. 349. See also: — "Differential Septa in Plants with reference to the translocation of nutritive materials." Proc. Roy. Soc. 1911, Vol. lxxxiv. b, p. 226. 1910. Combes, R. "Du r61e de I'oxygfene dans la formation et la destruction des pigments rouges anthocyaniques chez les v^etaux." C. It Acad. In these formalae n is a power of 2 and is equal to one-half the number of gametes in a series. Joam. of Gen. u 99 314 Forms of Reduplication and if so, of what type must it be ? We need only state now that the answer is that there is reduplication between B and C of the type nm +1 : n + m : ?i + «i : w wi + 1. The mode by which this ratio is found is given below. We have however to note that this type of ratio also does not conform to the Bateson-Punnett formula. Certain experimental results will, I believe, in view of these con- clusions, repay further study. All the gold hjis not yet been extracted from the ore. Reduplication clearly depends upon peculiarities in the mode of formation of the gametic series. As however it, so far as we know at present, affects pairs of factors only, it is convenient to ignore such possible cases of reduplication as might occur between, say, triplets or quartets. With this limitation and adopting the Bateson-Punnett hypothesis of reduplication {Jourti. of Genetics, Vol. i. No. 4, p. 293). it is quite easy to construct the gametic series for any set of re- duplications. Let us consider first the simple case in which three factors A, B and C are involved, with reduplication between A and B only, and in the form n : 1 : 1 : n. The gametic series if A and B are alone considered would be wAB 4- lAb + laB + nab. To include the factor C, the series must consist of eight terms and be arranged so that each member of the above will be associated with C and c, without disturbing the established reduplication; thus nkBC + wABc + lAbC -I- lAbc + laBC + laBc 4- nabC + na.bc. By extracting the pairs separately from this series, we get AB : Ab : aB : ab :: 2n : 2 2 : 2n or /i : 1 : 1 : n. AC : Ac : aC : ac :: w + 1 : n + 1 : w + 1 : w + 1 or 1:1:1:1. BC:Bc:bC:bc::n-\-l:n+l:n+l:n + l or 1:1:1:1. Clearly a reduplication between two factors A and B does not alter the ratios for A and C and B and C. An experimental illustration of this is furnished by Gregory's work on Primula sinensis, in what may be called the MSD group of experi- ments; where M = magenta dominant over m = red ; and S = short style „ „ s = long style ; and D = single flower „ „ d = double flower. A. H. Trow 315 In these experiments there is reduplication between M and S of the form 7:1:1:7; but M and D and S and D show no reduplication and give each the normal ratio 1:1:1:1. We may now consider the more important case, where there are three factors A, B and C and reduplication between A and B in the form n : 1 : 1 : 71, and between A and C in the form m : I : I : m. The gametic series when A and B are alone considered would be nABC + nABc + lAbC + 1 Abe + laBC + laBc + wabC + nabc. To secure reduplication between A and C as well, and of the form m : 1 : 1 : w, the terms involving AC and ac must be multiplied by m ; the series thus becomes w/nABC 4- nABc + /?iAbC + 1 Abe + laBC + maBe + nabC + wmabc. Extracting the three pairs separately from this series, we get AB : Ab : aB : ab :: nm 4 n : m + 1 : 1 + m : m + rnn AC : Ae : aC : ac BC : Be : bC : be From this procedure, it is clear that reduplication between A and B and betw^een A and C involves reduplication between B and C. It is worthy of note that this derived or secondary type of reduplication has apparently been entirely overlooked, especially as there is good reason to suppose that it has already been observed experimentally. Moreover, it belongs to a fundamentally different series, — of the form p • q ' g '■ p. Gregory's interesting results on Primula sinensis illustrate this case. In the MSG group of experiments, where M and S have the same significance as above, G represents green stigma, dominant over g — red stigma. The best numerical results are given by the crosses in which the Fi — MSGmsg was crossed by the triple recessive msgmsg. In such cases it is clear that the ratios of the zygotic series coincide with those of the Fi gametic series. The results may be grouped as follows : — MS Ms mS ms Nob. found 53 3 6 40 Expectation on ratio of 7:1:1:7 45 6 6 45 : n : 1 1 : «. : nm + ni : : w + 1 :l+ n : m + nm : ni 1 1 TO. : nm + 1 : n + m : : m + n : 1 + nm. MG Mg mG mg Nob. found 39 17 18 28 Expectation on ratio of 2 : 1 : 1 : 2 34 17 17 34 22—2 316 Forms of Rediqdication Nos. found Expectation on ratio of 5 : 3 : 3 : 5 derived from SG 64 Sg 35 (7:1:1 I2 : 1 : 1 54 32 sG 30 32 8g 44 5i The suggestion of the 2:1:1:2 ratio in the second case is made on my own responsibility — Gregory does not assign one. The main interest lies in the fact that the derivative ratio 5:3:3:5 explains fairly well the facts of the case. The following diagram will serve to illustrate the hypothetical course of the segregations and the cell-divisions in this case. liMSG ^MsG P'msG 7MSg 5mSg i^msg The sign x signifies increase in the number of gametes, or gameto- genic, segregating cells, and the following number the relative amount of increase along the different axes. The primary and secondary reduplications, three in number, are notable in that each represents a case of coupling. Let us therefore consider the case in which the primary reduplications are of the form 1 : w : n : 1 for A and B and 1 : m : m : 1 for A and C. Under these conditions the gametic series will be ABC + mABc + ?iAbC + nmkbc + nmaBC + 7?aBc + rwabC + abc, and the reduplication between B and C will be found (by extracting) to be of the form BC : Be : bC : be :: 1 + nm : m-{- n : n-\- m : nm + 1. We thus get the result that the reduplication between B and C is of the same form whether the ratios between A and B and A and C axe 1 : n : n : 1 and I : m : m : 1 or n : 1 : 1 : n and m : 1 : 1 : m. A. H. Trow This case may be represented diagrammatically ; thus 317 ;i//( Abe «AbC abC naBc /imaBC Since n and m are each greater than one, it may be shewn that 1 "J- Tiin whether n is equal to, greater than or less than m, is greater ^ ° m + n '^ than one, and therefore that the type of reduplication between B and C is of the nature of a coupling. We have therefore established the rule that reduplications between A and B and between A and C whether of the form of couplings or of repulsions, give rise to a secondary reduplication between B and C of the form of a coupling. We may now consider the case in which the types of reduplication between A and B and between A and C belong to the series n : 1 : 1 : n and 1 : wi : m : 1 respectively. The gametic series in this case will be 7iABC + nmABc + AbC + mAbc + maBC + aBc + mnSibC + nabc, and the reduplication between B and C will be necessarily of the form BC : Be : bC : be :: n + 7/1 : nm + 1 : 1 -\- nm : m + n, 318 Forms of Redtiplication and be graphically represented thus :- mAbc ttmabC nabc nmABc maBC Since n + m is less than nm + 1 this type of reduplication is of the nature of a repulsion. The EBL group of experiments conducted by Bateson and Punnett and described in Proc. Roy. Soc. B, Vol. 84, p. 7, illustrates this case. The cross Ebl x eBL shews repulsion between E and B and coupling between B and L. It will simplify comparison to write the factors in the order BLE. It has been found that bl ::7 be :: 1 1 : 1 m : m 7, 1. BL : Bl : bL and BE : Be : bE Hence it may be deduced that LE : Le : IE : le :: 7 + m : 7m + 1 : 7m + 1 : 7 + m, which indicates a repulsion. This appears to have been observed, but it is not clear to me from the description of the results whether these three types of reduplication have been observed in the same cross. They should certainly be looked for. yAbC gAbc pABC pahC ABCc xp abCc pABc pahc (jaBC aBc A. H. Trow 319 We have found that certain derivative reduplications are of the form p :q : q : p. It seems probable that there may be primary reduplications also of this type. When such a reduplication is confined to one pair of factors A and B, a third factor C being unaffected, the gametic series would be j)ABC +/)ABc + q^bC + ^Abc + ^aBC + q, we get coupling; if q is > p, we have repulsion. But we may have reduplication between A and B of the form p : q : q :p and between A and C of the form r : s : s : r. In this event there will be a derivative reduplication between B and C, the form of which may be ascertained as follows: — The gametic series will be pr^BC +psfiiBc + qrfiibC + qs^bc + qsa.BC + qra.Bc + psabC + prabc, and, by extracting, the derivative reduplication is found to be BC : Be : bC : be :: pr + qs : ps + qr : qr + ps : qs + pr, or more simply :: jyr + qs : ps + qr : ps + qr : pr + qs. This is the most general formula for a derivative reduplication and is of course applicable to all the preceding simpler cases. The following diagram illustrates the course of the assumed segre- gations and cell-divisions : — ^rAbC 93 Abe prABC p* ABc ABCc xp abCc prahc 9«aBC qrzBc 320 Forms of Reduplication These tjonsiderations shew that the reduplication hypothesis adequately explains the occurrence of all the ratios hitherto determined. We perceive too how segregation and cell-division may he associated, and that these appear to be carried out symmetrically. In the complete absence of reduplication there is a typical radial symmetry of the segregation apparatus. When reduplication sets in, a bilateral structure is developed, and this may ultimately assume quite a complex form. There seems some reason, moreover, to believe that the development of cells at the two ends of the same axis may be unequal. This would produce a new form of symmetry — the structure becoming two-ended, enabling one to distinguish not only between different axes, but between the two ends of the same axis. In such a case, with a single pair of allelomorphs A, a, we should get a divergence from the gametic ratio, — 1 : 1 and the normal zygotic ratio, — 1 : 2 : 1. Such divergences are not infrequently met with in the literature of genetics. We may, therefore, ultimately find cases of asymmetrical types of reduplication, such as are represented by the ratios (1) w '. X : y : lu, (2) w : X : X : z, (3) w : X : y : z. The experimental determination of such ratios would of course be difficult. Finally let us consider the case of four or more factors A, B, C, D . . . with reduplication between A and B, A and C, A and D ... of the form n : \ : \ : n, m : 1 : 1 : m, p : \ : 1 : p, ... . In addition to the derivative reduplication between B and C, there will be now reduplications also between B and D and C and D .... The gametic series for four factors A, B, C and D would be wmpABCD -I- nwABCd + wpABcD -l- wABcd -\- mpfKbCD -f- 7nAbCd 4- ^AbcD + Abed + aBCD + jt)aBCd + ?naBcD + ?7ipaBcd + wabCD -I- npahC6 +ww*abcD + wm^abcd and it can be shewn, by extracting, that the reduplications between B and D and C and D have the form : — BD : Bd : bD : bd :: w^ + 1 : n -I- ;) : n -H p : «i> + 1, CD : Cd : cD : cd :: mp + 1 : m + p : m-\-p : mp + \. A. H. Trow 321 This result may be reached more easily by making use of the general formula on p. 319. p. np + 1. For if AB : Ab : aB : ab :: n 1 1 and AD : Ad : aD : ad :: p 1 1 then BD : Bd : bD : bd :: n;) + 1 : n+p : n+/> Any number of derivative ratios may be ascertained in the same way by this method. This more complex case may be represented graphically thus : — nmpABCD paBCd m aBcD mp aBcd ntnp ahcd According to this scheme of segregation (which however must not be regarded as the only possible one), each additional factor (or pair of allelomorphs) Ee, Ff, Gg, etc. will necessitate a further dichotomy of each branch. If these additional branches are equally developed it can readily be shewn that reduplication does not take place. We have the important rule that equal dichotomies produce normal segregation; unequal dichotomies produce reduplications. These two types of behaviour may occur in any order or at any stage in the phylogeny, but as Bateson and Punnett have already stated, they cannot occur simul- taneously. 322 Forms of Reduplication There seems to be no reason why the most various types of re- duplication should not occur together in the same plant as the result of the same cross. The hypothesis of reduplication seems adequate to explain the occurrence of any type of ratio. The most suggestive point which emerges from the analysis is the importance of the product nmp . . . and of its constituent factors. From these, when all the factors and all the ratios in any one cross have been ascertained, it should be possible to compute the minimum number of successive cell-divisions needed to produce the complete system of segregation. It ought to be possible to determine also, in sweet-peas for example, the number of successive cell-divisions which normally intervene between the first division of the zygote and the last of the gametogenic divisions, and the distribution of these in the ontogeny. Comparison of the two results might serve to fix the stage at which segregation takes place. It is then advisable to distinguish between primary and secondary reduplications. A ratio of reduplication ascertained by experiment may belong to either series. The gametic series is based upon the primary reduplications alone. Every observed type of reduplication must be assigned to its proper position. It is comparatively easy, as we have seen, to calculate the secondary from the primary reduplications. The schemes on p. 323 will illustrate the relationships of primary and secondary reduplications. It is perhaps advisable to add that systems of segregation will probably be seldom found in which all the primary reduplications take place between one factor A and a number of others B, C, D, E, etc. Primary reduplications may occur between any pair of factors, and the consequent secondary reduplications will undergo corresponding modi- fications. The construction of the gametic series, when the ratios of primary reduplication are known, is easy, and from these any secondary re- duplication is ascertainable. The following scheme illustrates such a system of reduplications : — Primary reduplications Secondary reduplications A and B = n : 1 BandC = m:l AandC= nwi + 1 : n+m CandD=p:l ^BLnAD = nmp + n + m-\-p:nin + np-\-mp-\-l B and D = »ip + 1 : J7i+^ A. H. Tbow 323 08 08 13 ■4.3 (D uT uT q" d ocT < £ o -<^ o 08 + It b. Ill + + o, =>- i-l r-t + + II II ■ u u. 1 Q Q a. Oi w + + + g i£ S «-i i-l »H + + + a, *« t. S !£ ft II II II O u li. •o r— Tl s a a £1 1 ei c£ « o o o a a, &. K + + + + 8 8 e a + +-(- + 8 8* 8 5 II II II II O Q 111 U. '9 'O 'O •^ a a c a cS <8 eS ei CQ CD OQ (Q 8 6 a, <»' V. II II II II II CQ O O tiJ U. r^ r^ r^ '^ "^ c c s c c cs c: cs c: ci < < < < < r H » II U. •o s u a « 8 + + g e* s Sj «• + + ;i H 11 II UJ U. TS TS a 08 c o o ^ S, S, + + ■*- S, a, =». • S a-. + + + *• s s a, a, a, II II II O 111 ti. 73 TS TS a a a eg oS cS o o o a, v ft ft + + 5 8 8s + + a 2> s s + + 9 2> ft a + + ii I! II II UJ li. o o bl n b. a « 111 c- 1-1 a> « 9* Jh II II II O bJ b. a § cS s o o o lO 5« t» XI t» W ^ 90 ■-< 1-1 II II II II O Q bJ b. 'O 'O 'O 'O = 229 c3 c3 e3 e3 00 CD CQ 00 O Q bl b. S S 'O S "V rs g a a c d eS c« cS CQ CQ CO CQ O O II II II O bl b. "O "O -o a a a eS cS cS O O O •O 00 li lO ■* t» I-l t^ II II II II O O bJ b. 'a 'O 'O -^ a a a a gg c3 d a CQ CO OQ CQ ^ « » "1 »H c« OS •^ X3 II II II II II OQ o o bl b. a ■ OQ O O bl b. -o 1— • n rs "13 •c T! TS r3 t3 -n c a c a a c^ a C a eS s! eS a a i a — :£ d a o to o< » 3 8 a. «. » H II II II II II CD O O bl b. < < < < < X — V < < < < < 3J o < < < < < 324 Forms of Reduplication It is also noteworthy, that complex reduplications may arise owing to the combination of a primary and a secondary reduplication in the same gametic series; e.g. the primary reduplications may be of the following types : A and B = ?? : 1 : 1 : n A and C =m -.1 :1 : m, B and C = p : \ : \ : p. The first two alone involve a secondary reduplication between B and C of the type nm + 1 : n + m : n + m : nm + 1, and this combined with the primary reduplication between B and C gives the complex reduplication for B and C of BC : Be : bC : bc :: ^ {nm + 1) : n + ni : n + m : p (nm + 1). A careful study of systems of segregation will therefore, as soon as two or more reduplications have been discovered, furnish the student of genetics with data which will enable him on the one hand to test his hypotheses by further experiment, and on the other hand to extend and facilitate his researches. It must be borne in mind however that cell-divisions, if they do really set up the phenomena of reduplication, must themselves depend upon the structure of the protoplasm. It may be that the systems of segi'egation will prove of some value in the analysis of this structure. University College of South Wales AND Monmouthshire, Cardiff. SOME RECENT WORK ON MUTATION IN MICRO-ORGANISMS. II. Mutations in Bacteria'. By CLIFFORD DOBELL. In the following article I shall try to give a coherent summary of some recent work in bacteriology in so far as it is of interest to the student of genetics. This work is very extensive and scattered through numerous papers dealing with medical or purely bacteriological matters. It is, moreover, to some extent beyond the reach of the general biologist on account of the phraseology in which it is couched. I shall try, therefore, to present the facts in such a way that they may be seen stripped of irrelevant detail and in language intelligible to the average reader. Accordingly, the facts given in the following pages are to be regarded as a selection from a vast array recorded by many different workers, and not as a complete review of even those works given in the bibliography (p. 349). They represent, rather, certain personally chosen facts arranged in an orderly manner so as to interest workers in genetics. It must be remembered, therefore, that many additional facts — some of them perhaps of fundamental importance — are to be found not only in the works to which reference is made, but also in works which I have not considered*. I should like to point out that I employ the word "mutation" throughout in the sense of Wolf (1909), who follows Baur in this ' In a prerious article (this Journal, Vol. n. p. 201) I have given some accoant of recent work on mutation in Trypanosomes. 2 Those desiring an account of the work hitherto done on variation in bacteria, will find it admirably analysed in the recent monograph by Pringsheim (1910). 326 Mutation in Micro- Organisms matter. By mutation, accordingly, I mean a permanent change — however small it may be — which takes place in a bacterium and is then transmitted to subsequent generations. The word does not imply anything concerning the magnitude of the change, its suddenness, or the manner of its acquisition. The term denotes a change in genetic constitution. All other changes which are impermanent — depending generally upon changes of the environment — and not hereditarily fixed, are called modifications. The word " mutation " has been used with such different meanings by so many bacteriologists and others, that the foregoing statement seems called for. Indeed, discussions as to whether such and such a change is or is not a " mutation " might have been avoided in many cases if the opponents had defined their use of the word precisely. I do not wish to assert, however, that my usage is the correct one : I wish merely to state what I mean when I use the word in the following article. It is not appropriate to discuss here, I think, the applicability of the word " mutation " to the Bacteria. I am well aware of the difficulties involved in applying the word — generally applied to certain changes in sexual multicellular organisms — to the Protista. I am also well aware of the difficulties involved in the above definition. Perhaps an imaginary concrete instance will serve to make its meaning clear. Let us suppose that a given Bacillus is coloured red under normal conditions. By growing it and its offspring upon a new medium they become — let us suppose — colourless. If the organisms and their descendants when transplanted again into the original medium are again found to be red, then the change (loss of colour) is a modifi- cation : if, on the other hand, they are found to be now permanently colourless, then the change is a mutation. By far the greater number of variations described in Bacteria are of the former type. The mutations observed in Bacteria may be conveniently grouped into two classes — those in which the change is functional (e.g. changes in the power of producing ferments or pigments) and those in which the change is structural. Most of the mutations about to be recorded are of the former type — or physiological mutations, as I shall call them. I will therefore begin with a description of these, and con- sider some recent work on morphological mutations in a later section (p. 344). C. DOBELL 327 A. Physiological Mutations. The first series of experiments which I shall describe in this section concerns changes in the powers of fermentation observed in the Bacteria belonging to the coli-typhosus group'. This is a very large group containing mostly gut-inhabiting organisms. It contains a large number of different "species" or "races," ranging from the common "harmless" Bacillus colt to the parasitic Bacillus typhosus of typhoid fever. It is convenient to regard these two organisms as the extreme limits of the group, and to place the dozens of other members in various intermediate positions according to their properties. Morphological differences between the members of the group are so slight and inconstant that a physiological classification is at present the only one possible. For our present purposes it should be remem- bered that — in addition to the difference in pathogenicity — Bacillus coli differs from B. typhosus in the following features among others : it is able to ferment lactose and glucose, to produce indol when grown in broth, and to clot milk. B. typhosus, on the other hand, can do none of these things. Owing to the differences in their powers of splitting sugars, we find that these two organisms shew characteristic differences when grown in certain test media. On the medium of Drigalski and Conradi, B. coli forms red colonies, whilst the colonies of B. typhosus are blue. On Endo's medium, similarly, B. coli forms ' For those unversed in bacteriology a few additional remarks concerning the naming of these organisms will perhaps be necessary. The common colon bacillus is variously known as Bacillus coli and Bacterium coli commuTte. The typhoid organism is called Bacillus typhosus or Bacterium typhi abdominalis. The other members are in part named on the binominal system, but frequently also on a trinominal or even quadrinominal or quinquenominal system (e.g. Bacillus faecalis alcaligenes liquefaciens). The strings of Latin names thus used are as a rule descriptive terms rather than ordinary specific or varietal names. In part, however, the members of the group are known by the names of their describers (e.g. Gartner's bacillus, Flexner's bacillus). In part also they are designated by numbers or letters of the alphabet (e.g. paratyphoid bacillus B, etc.). Combinations of these methods are also resorted to. The vast numbers of these organisms, and the extreme difficulty of deciding upon the systematic status of the various "species,"' "varieties," "strains," etc. have thrown the nomenclature into a state of well-nigh hopeless chaos. Throughout this paper I shall call the common colon organism Bacillus coli and the typhoid organism Bacillus typhosus. In other cases I shall use the terms employed by the writers whose work I am considering. 328 Mutation in Micro- Organisms red, B. typhosus colourless colonies \ In the absence of extraneous colouring matters, both organisms form colourless or whitish colonies. We may begin our description of the mutations observed in the coli-typhosus group with a consideration of the fundamentally im- portant work of Massini (1907)2. From a case of enteritis, this worker obtained an organism which at first grew as whitish colonies (like typhosus) on Endo's medium. After the third day of growth, however, minute red nodules appeared in the whitish colonies — the nodules increasing in number in the course of time until certain colonies contained as many as 200. These nodules were found to be solid masses of bacteria, constituting daughter-colonies within the parent-colony : and, as their colour indicated, they consisted of organisms which possessed the power of fermenting lactose (like coli). One would naturally suppose that this peculiar phenomenon was due to the impurity of the original culture — its two different constituents having simply separated out in the older cultures. By carefully plating out^ the original culture, and by other tests, Massini convinced himself ^ To understand what follows it is necessary to understand the principles involved in these methods, which are in everyday use for the identification of members of the coli-typhosus group. The principles upon which the Drigalski-Conradi and Endo mediums are compounded are similar. Both consist essentially of an agar medium containing lactose and a colour indicator. The medium of v. Drigalski and Conradi is slightly alkaline and contains litmus as an indicator. B. typhosus does not attack lactose, and therefore grows as a blue colony on the medium. B. coli, on the contrary, splits the lactose with the formation of lactic acid and gas. The acid produced by a growing colony turns the litmus red, the coli colonies thus being distinguished by assuming this colour. Endo's medium contains fuchsin, reduced by Na2S03 to a colourless leuco-compound. B. typhosus accordingly grows on the medium in the form of colourless colonies : whereas B. coli, by splitting the lactose and forming acid, converts the reduced fuchsin back to its characteristic red colour. Colonies of B. coli on this medium are therefore dis- tinguished by possessing the deep red colour with a green shimmer characteristic of fuchsin. It will be understood, therefore, that when in the following account a bacillus is said to form — let us say — red colonies on Drigalski-Conradi agar, what is implied is not that the bacteria themselves are red organisms, but that they possess the power of fermenting lactose, i.e. produce a lactase. ^ Massini's investigations were carried out in the laboratory of Neisser. According to Pringsheim (1910) similar observations had been previously made upon yeasts by Hartmann, whose records I have not been able to consult {Wochenschr. f. Brauerei, 1903). ' "Plating out" is a method frequently used for testing the purity of cultures. It consists essentially in "diluting" a suspension of organisms with a large volume of culture medium, and then spreading it out on glass plates. The individuals are thus separated as far as possible, and the colonies which develop are known to be derived from single, or at most a few, individuals. By repeating the process a number of times a mixed culture can be detected and pure races isolated. C. DOBELL 329 that this was not the case. Absolute proof of the matter could only have been obtained, of course, by isolating a single individual from the original culture and cultivating it further — a course which Massini could not follow owing to the great technical difficulties involved. The assumption that the original culture was pure was, however, rendered extremely probable. The organisms, when grown in media containing lactose (1 °/J, always produced red daughter-colonies, but they never behaved in this way when grown in other media containing dextrose, mannite, or other similar substances instead. Further cultivation of the colonies yielded remarkable results. The organisms in the red nodules had permanently acquired the power of fermenting lactose. They always produced red colonies on Endo's medium — never whiter Even after transplantation on to other media free from lactose, they never lost this power. Subsequent colonies again grown on Endo's medium were invaiiably pure red. The typhosii3-\ike original, therefore, had given rise to a number of new individuals which closely resembled coli, having acquired the power of fermenting lactose. Now the white parts of the colonies containing the red nodules behaved exactly like the original organisms. When transplanted, they produced colonies at first white but subsequently developing red nodules, the individuals in which bred true. White colonies, if transplanted every 24 hours, remained white : but they always pro- duced red nodules if left for several days. The "white" race was therefore to all appearances constantly undergoing a partial mutation into a pure " red " race. Or, to put it in another way, the original non- lactose- fermenting race constantly split up into two daughter races — a pure lactose-fermenting race and a non-lactose-fermenting race of the same nature as itself (i.e. with the power of splitting into these two components again). The non-lactose-fermenting race might therefore be called an " ever-sporting variety," and it was called by Massini, in consequence, Bacterium coli mutabile. He regarded it as a typhosus-Vike organism which was constantly undergoing mutations into coli races'. The other characters of his remarkable race were carefully studied by Massini. He also tried to discover whether other coli-typhosus 1 In one instance, under peculiar conditions, a single white colony was obtained from a red race. The red colonies never developed nodules. ' Massini regarded the "red" races as typical races of B. coli. See however the observations of Thaysen recorded on p. 333, footnote (infra). Jonm. of Gen. a 23 330 Mutation in Micro- Organising organisms^ behaved in a similar manner, but he could find no evidence that this was so. It may be mentioned that as regards their virulence and agglutination reactions the " white " and " red " races appeared to be identical. It is somewhat remarkable that all the numerous workers who have subsequently studied these organisms have confirmed Massini's observations. But later workers have been able to amplify his work in several directions, and have discovered similar phenomena in many other races of coli-typhosus organisms. Twort (1907) recorded independently that certain coli-typhosus organisms were able to acquire the power of fermenting certain sugars if grown in them for a sufiiciently long time. The change took place slowly. In this manner he modified dysentery bacteria (Kruse and Flexner strains) so that they were able to ferment saccha- rose: and he was able to train B. typhosus to ferment lactose and dulcite. The organism which had acquired the power of splitting dulcite is stated to have retained this power permanently — even after passage through a guinea-pig and cultivation in a dulcite-free medium. Burk (1908) isolated an organism almost exactly like that studied by Massini. He made similar observations upon it. Sauerbeck (1909) records closely similar results. Benecke (1909) and Kowalenko (1910) were able to confirm Massini's results and to make an important addition to them. They succeeded in isolating and testing individual organisms. Both ob- servers adopted Burri's Indian ink method^, and obtained the same results, which completely agreed with Massini's original results. As Kowalenko's results are recorded in greater detail, they may be given here, though Benecke's were recorded first. Kowalenko (1910) obtained both the races studied by Massini and 1 Amongst the forms investigated were typical B. coli, B. typhosus, Flexner bacilli, Shiga bacilli, and paratyphosus races. '^ The introduction of this ingenious method has made it possible to isolate and cultivate individual organisms from a colony of bacteria with comparative ease. (See Burri, Das Tuscheverfahren, Jena 1909.) In principle the method is as follows. A drop of fluid containing a few bacteria is mixed with a sterile solution of Indian ink. A minute drop of the mixture is then placed under a cover-glass and examined under the microscope. The bacteria can be seen, under a comparatively low magnification, as white dots in a black field : and the smaU number of individuals in each preparation can be counted. Those preparations containing but a single individual are then selected, their contents transferred to a suitable culture medium, and cultivated further. In this way it may be known with certainty that the culture produced has sprung from one original individual. C. DOBELL 331 Burk. Whet) sub-cultures were made from them by isolating indi- vidual organisms, precisely the same results were obtained. That is to say, it was found that a single individual (belonging to either race) which at the outset was unable to split lactose, produced in presence of this substance offspring which were in part like itself and in part able to split lactose. The only possible objection to Massini's results was therefore removed. Single individuals from " white " colonies gave rise to mutating races : single individuals from " red " colonies bred true. Their properties were unaltered by passage through animals, by changes of temperature, or by phenol and other drugs. A series of investigations parallel to those which have so far been mentioned has been conducted by Reiner Muller (1909, 1911), whose work appears to have been extremely thorough. Altogether he has studied several hundred races of organisms of the coli-typhosus group, especially as regards their powers of fermenting 18 different kinds of carbohydrates. The most important outcome of Muller's work has been the demonstration that all typical races of B. typhosus behave towards rhamnose exawtly as Massini's B. coli mutabile behaves towards lactose. That is to say, typical pure races of B. typhosus are unable to ferment rhamnose. When grown in a medium containing this sugar, however, the colonies develop daughter-colonies (nodules) consisting of indi- viduals which have permanently acquired the power of splitting rhamnose. He claims that this is the most typical cultui^al character of the typhoid bacillus*. Mutations of this sort invariably occur, and under no conditions do the rhamuose-splitting organisms lose their power. Muller has confirmed these observations in the case of indi- viduals isolated by Burri's method. A single non-rhamnose-splitting individual gives rise in the presence of this substance to offspring which are partly like itself and partly able to ferment rhamnose. In the case of certain paratyphoid organisms {paratyphosus B., Schottmiiller) Muller has found that rhamnose and lactose do not produce any corresponding change. But the organisms behave to- wards raffinose exactly as mutabile behaves towards lactose, or typhosus behaves towards rhamnose. In other words, paratyphosus B. grown in raffinose-containing media produces daughter-colonies which can ferment this sugar — though the mother-colony cannot do so. These raffinose-fermenting organisms never revert to the non-raffinose- fermenting t3rpe. I No less than 120 races of B. typhottu were sttidied. 23—2 332 Mutation in Mict'o- Organisms A case of a similar nature has been recorded by Jacobsen (1910). From a typhoid epidemic he obtained a bacillus (A) which differed from a typical B. typhosus in that it refused to grow properly on the Drigalski-Conradi medium^ used in the laboratory. Finally, however, nodules appeared in these weakly colonies, and were found to consist of individuals which grew strongly in the orthodox manner. On culti- vation the latter organisms were proved to be typical B. typhosus. Experiments showed that the Drigalski-Conradi medium had been altered in some way by repeated autoclaving^. The altered medium retarded the growth of the strain A^, which was able, however, to undergo a partial mutation into strains which could grow easily in the medium (B. typhosus). Jacobsen — on analogy with Massini's results — ^proposes to call his original strain A by the name B. typhi mutabile. The mutated form — B. typhosus — never reverted to the original type (A). Jacobsen's results have been in the main con- firmed by R. Muller (1911). Some other cognate facts may be briefly considered here. Schroter and Gutjahr (1911) record the following observations. A race Y of coU-typhosus-grou-p organisms cannot ferment maltose. After culti- vation, however, it partially acquires the power — thus coming to resemble Flexner bacilli. The Y organisms appear to behave towards maltose as B. coli mutabile behaves towards lactose. They found similar changes in other related organisms. For example, Shiga-Kruse bacilli may acquire the power of splitting both maltose and saccharose when grown in media containing these sugars. The change appears to be permanent — that is, once the property is acquired by the race, it is never lost. Sobernheim and Seligmann (1911) have isolated an organism which behaves exactly like Massini's B. coli mutabile. They have confirmed their results by isolating individuals by Burri's method. (See p. 330, footnote.) In addition, they record that they have obtained four dif- ferent pure races from one pure original race. From a coli-typhosus- group organism (Haustedt) they have obtained in pure culture (1) a true Gartner strain, (2) a similar strain, but differing in agglutinating power, (3) a typical typhosus, and (4) a strain almost identical with B. coli mutabile (Massini). » See p. 328 footnote. ^ An autoclave is an apparatus used for sterilizing culture media, etc. Sterilization is effected by steam under pressure. 3 But not, of course, of the typical strains of B. typhosus. C. DOBKLL 333 In a preliminary note Thaysen (1911) announces that he has iso- lated and studied eight different races of coli-typhosus-gronp organisms. These have the following properties : Four races ferment dextrose, maltose and lactose, but not saccharose. By cultivation, however, they acquire the power of splitting this sugar also. One race splits dextrose and maltose, but not saccharose or lactose. But it can acquire the power of splitting the last named sugar. It appears to be identical with B. coll mutabile^. Two races are similar to the preceding, but can acquire the power of splitting saccharose. They appear to be similar to B. imperfectum^ of Burri. Finally, one race splits dextrose, maltose, and saccharose, but not lactose. It can, however, acquire the power of fermenting this substance ^ Up to this point I have mentioned only those results which are essentially similar to Massiui's. I have purposely avoided referring to the work of Burri and his collaborators. The work so far is at the same stage, but Burri's work represents a step forward. Let us now consider it. The papers of Burri and Diiggeli (1909) and Burri and Andrejew (1910) may be considered — for the present purposes — as parts of the admirable work of Burri (1910), which I shall now endeavour to chronicle. Burri has isolated* a race of organisms of the coli-typhosus group which are unable to ferment saccharose and lactose. He calls this race Bacterium imperfectum. It never acquired the power of splitting lactose : but on the other hand, when grown in media containing sac- charose, some colonies acquired the power of splitting this sugar. The saccharose-splitting mutant Buni names B. perfectum\ (Both organisms belong to the paratyphosus sub-division of coli-typhosus organisms.) Burri's observations were originally made upon organisms grown in "shake-cultures*" — not on Endo's medium. The cultures were made ^ The lactose-splitting matant from B. coli miitahile is, according to Thaysen, unable to produce indol. It is therefore not a typical B. coli. ^ Vide infra. * Thaysen's full paper has appeared whilst this article is in the press. See C. B. Baku I. Abt. (Orig.), Vol. Lxvn. 1912, p. 1. * From fermenting grass. '^ On analogy with this, he proposes to call the lactose-fermenting mutant derived by Massini from the non-lactose-fennenting B. coli mutabile by the name B. coli mutatum. * In a shake-culture the organisms are distributed, by shaking, through a liquefied jelly — in this case containing saccharose. When the jelly has set, the isolated organisms produce colonies. Those which ferment the sugar produce bubbles of gas in the jelly and can therefore be readily distinguished. 334 Mutation in Micro- Organisms both in the ordinary way and also from individuals isolated by his Indian ink method (p. 330, footnote). Each non-saccharose-splitting initial organism was found to give rise to only " a very low percentage " of saccharose-splitting colonies. The latter, however, had acquired the power permanently, and always bred true to this character — after " any number " of sub-cultures in many different media. Burri then obtained Massini's strain from Neisser, and tested it by shake cultures. He made, conversely, test cultures of B. imperfectum on a modified Endo medium (containing saccharose instead of lactose). He concluded from all his experiments that B. imperfectum behaves towards saccharose exactly as B. coli mutahile behaves towards lactose^ Experiments were then made to determine what percentage of the offspring of B. imperfectum (non-saccharose-splitting) underwent mutation into B. perfectum (saccharose-splitting). The results of some ingenious work were surprising. An initial culture begun with some 10,000,000 germs of imperfectum, produced about 50 colonies of per- fectum, but the majority were of the imperfectum type. Beginning with about 100,000 imperfectum (only ji^ of the first number), how- ever, he again obtained about 50 colonies of perfectum. And by beginning with a very few germs, he found that practically every colony was of the mutated form (perfectum). " In a series of cultures with a diminishing number of initial germs, of which the greatest is a million times greater than the least, each member of the series shows approximately the same number of mutated colonies^" Burri was thus able to demonstrate that all the individuals of the imper- fectum race are — under suitable cultural conditions — able to mutate completely into perfectum forms. In other words, all imperfectum individuals are, as regards their power of mutation into 'perfectum races, equipotential. The apparent partial mutation of the race is dependent upon the different environmental conditions to which the different individuals of a large colony are subjected. When the indi- viduals are sufficiently separated so that every one is offered the same favourable opportunities, they all behave in the same way — that is, all mutate. To borrow a term from Driesch, every individual has the same prospective potency. ^ Massini, p. 328 supra. 2 In other words, suppose we begin by sowing 100 million individuals — we get then nearly all the colonies like the initial germs, and only x colonies with the new character. If we sow only x individuals of the initial germs, however, we obtain x colonies with the new character. Similarly, any number of initial germs between 100 million and x, always produced x mutations. C. DOBELL 335 By a further set of nice experiments which cannot here be con- sidered in detail, Burri has tried to shew that the power to split saccharose does not appear all at once\ Between the non-saccha- rose-splitting impei'fectum and the saccharose-splitting per/edum, intervene many generations of individuals shewing every transitional stage. The power of fermenting saccharose is gradually acquired in small successive steps, until it manifests its full development in the actively fermenting form B. perfectum. The power once acquired is never lost — it always persists in the ofiFspring. Moreover, individuals which are in a transitional stage do not lose such partial activity as they have acquired. Individuals which have only "half-acquired" the power of splitting saccharose^ may be transplanted to a saccharose-free medium. If their offspring subsequently come in contact with the sugar they then acquire complete power of fermenting it in half the time necessary for an ordinary imper/ectum race to do so. Burri supposes that the power to ferment saccharose is latent in every imper/ectum individual, probably in the form of a zymogen or pro-ferment of some sort. The enzyme is produced gradually by the constant action of the sugar on successive generations. He thinks it probable that the newly acquired power of attacking saccharose does not represent a " regeneration " of a power originally present in the race but temporarily lost*. All workers whose records we have so far considered have been unanimous on one point — namely, that when a race has once acquired the power of fermenting a certain sugar, it remains constant in this respect (i.e. " breeds true "). A few observations have been made, however, which shew that a reversion to the original state may occur in some races — that is, the acquired power may be subsequently lost under certain conditions. An instance of this sort is to be found in the work of Bernhardt and Markoflf (1912). They isolated* a coli- typhosus organism (" No. 459 ") which behaved exactly like B. coli mutahile. It grew as a blue colony on Drigalski-Conradi agar*, and con- stantly gave rise to red nodules in the mother-colony. The organisms ' The change is said to be "relatively quick, but not sadden." ' That they have partially acquired the power is inferred from this and other experiments. There is no outward and visible sign (gas production, etc.) of this "half-acquisition." ' Burri will not call the change a "mutation," because it is a gradually acquired adaptation. According to him the change is really not the acquirement of a new character, but the realization of a faculty already existing potentially. This appears to me, however, to be applicable to all variations — of whatever sort — in all organisms. •• From a patient suffering from an intestinal complaint. * See p. 328, footnote. 836 Mutation in Micro- Organisms from these nodules were found to breed true (i,e. as regards power of fermenting lactose). Attempts to obtain blue colonies from the " red " individuals almost always gave negative results. However, by passing the organisms from the red nodules through mice and rabbits they succeeded in obtaining a race like the original — one, that is, which produced blue colonies with red nodules on Drigalski-Conradi agar\ It seems, therefore, that the acquired power of fermenting lactose — which usually breeds true — may, under certain conditions, be lost. Results somewhat like those of Bernhardt and Markoff, but different from those of most other investigators, have been just recorded by Baerthlein (1912). He has studied 13 different races of coli-typhosus organisms from the guts of healthy and diseased persons. All these races behave like Massini's B. coli mutahile ; producing blue colonies on Drigalski-Conradi agar, colourless colonies on Endo agar, but shewing subsequently the characteristic mutations in the form of red nodules. These lactose-splitting organisms were found to retain this character after cultivation for a long time on Drigalski-Conradi and various lactose-free media. Nevertheless, Baerthlein claims that if the lactose- splitting organisms are cultivated continuously on ordinary agar, they revert in part to the non-lactose-splitting form. After only 6-7 days on agar, organisms transplanted back on to Drigalski-Conradi medium develop into both red and blue colonies. It follows, therefore, that after even so short an interval, the organisms may go back to their original form. These observations do not seem to square with those of most other workers. Perhaps the explanation is to be sought in the fact that different observers have studied different races, which do not all behave alike. Baerthlein, it may be noted, has supplemented his work by a morphological, cultural, and serological study of his races*. Confirmation of his results — by experiments on isolated individuals and on related races — is much to be desired. 1 The original "red" race was first tested and found to be pure. Bernhardt and Markoff state that they succeeded "often" in performing this experiment, but "not always." With the original strain of Massini, however, they obtained only negative results. 2 From Baerthlein's paper it is to be gathered that the behaviour of B. coli mutahile is even more complicated than at first appeared. He says that races of this organism (and also of ordinary B. coli) undergo mutation when grown on ordinary agar. The original race splits up into two constant daughter-races — (1) forming transparent colonies consisting of long, slender, individuals ; (2) forming opaque yellowish colonies of shorter and stouter individuals. Both these races when transplanted to a lactose-containing medium mutate into a lactose-spUtting and a non-lactose-splitting race — though both these retain the morphological characters of their originals. Four different races are thus produced from the initial race. All may "revert" under suitable conditions. C. DOBBLL 337 The changes undergone by organisms of the coli-typhosus group appear to be in all cases of the same nature — so fiir as they have been considered up to this point. All the changes appear to be direct adaptations. An organism comes in contact with a new sugar which it is unable to use for its own growth. It then changes itself so that it can split the sugar — the change being thus definite, towards a definite end, and apparently purposive. There appears to be a defi- nite relation between the sugar and the mutation. In some other cases, however, this relation is not obvious. Two instances of this may be briefiy mentioned. B. typhosus when grown on glycerin-agar produces acid. A closely related form (B. metatyphi), however, according to Mandelbaum (1912) produces alkali instead. But if metatyphi is grown for a long time on this medium, it forms nodules in the parent colonies — exactly like those produced in colonies of B. coli viutabile. The individuals in these nodules are typical acid -producing forms indistinguishable from ordinary B. typhosus. They do not revert to the alkali-producing form. In other words, metaiyphi mutates partially into typhosus^. Revis (1911) has studied certain coli-typhosus organisms which produce both acid and gas when grown in peptone broth containing certain sugars or polyhydric alcohols (e.g. lactose). He was able gradually to acclimatize these organisms to a medium containing 0*1 7o o^ malachite green. The organisms after this, however, had permanently lost the power of producing gas in the original media, though they could still form acid. The dye appears therefore to have made a lasting change in their method of attacking certain food-substances'. It will be seen, I think, that in these two cases just quoted the changes produced are not obviously of an adaptive nature — though possibly a greater knowledge of the chemistry of the matter might bring them into line with the preceding observations. But in the cases which we are now about to consider the changes produced seem to be quite definitely not adaptive. The mutations concern pigment- production — not ferment-production \ 1 Mandelbanm calls it a " remntation " or "atavistic throw-back" becaose he believes that metatyphi has arisen by mutation from typhosus. ' Bevis (1911) recorded only a single instance in which this change was observable. But in a subsequent paper (Revis, 1912) he states that he has been able to effect the same change in two other races of similar organisms. ' For earlier work on both these subjects consult Pringsheim (1910). Here also will be found an aecoont of similar changes in ferment-production among Fungi. 338 Miitation in Micro- Organisms There is one important paper to be considered in this connexion. It is a record of some extensive experiments performed by Franz Wolf (1909)^, who has studied Bacillus prodigiosus, Staphylococcus pyogenes^ Sarcina lutea, and numerous Myxobacteria. The aim of the experi- ments was to induce mutations, in respect of colour, by chemical or physical means. It may be stated at once that S. lutea gave only negative results, and that most of the changes observed in the other species were modifications — that is, transitory changes, not mutations (cf. p. 326). Many interesting observations were made concerning the effects of temperature changes and of a large number of chemical substances. But as we are now concerned with the mutations, further mention of the modifications will be omitted^. It should also be men- tioned that all the observations were carefully controlled by means of plate cultures ^ The first series of experiments was made with the historically interesting B. prodigiosus*. The initial colony on gelatine was coloured bright red®. Its purity was carefully tested, and as a further control the organisms were propagated under " normal " conditions for a long time. The strain was many times transplanted (altogether more than 50 times), being tested by a series of plate cultures each ^ It should be noted that Wolf's results are not always easy to comprehend from his account of them. Certain statements, for example, in his summary contradict others in the body of the paper. Dr Wolf has, however, kindly elucidated (in correspondence) all the doubtful passages, so that I trust my statements in the following lines coincide accurately with the facts. The discrepancies arose chiefly in the following way. The paper originally contained a summary — in graphic form — of all his experiments. This was eliminated by the editor, as he thought it superfluous. Without it, however, the paper is not always easy to understand, as in the text details are not given of all the cultures. 2 These modifications may, however, have some significance for the interpretation of the mutations. The original should be consulted, ^ No cultures were made from isolated individuals, as the organisms were found to be too small for this to be done. The plating out was done so carefully and. frequently, however, that this is probably a matter of small moment. * This is Ehrenberg's Monas prodigiosa — the organism which is supposed to have given rise to the legend of the Bloody Host. It occasionally appears in large masses on bread, etc. , *> The bright blood-red pigment has been named " prodigiosin." Though it has been much studied, its chemical composition is still practically unknown. It possesses several remarkable properties. It is formed only in presence of oxygen. It is bleached by sunlight — both from colonies of the bacteria and from solutions. Like most similar pigments, it is not present in the organisms themselves, but is excreted into the surrounding medium — thus giving the colonies their characteristic colour. It is not known what part it plays in the economy of the bacteria. C. DOBELL 339 time. As some of the colonies isolated in this way were found to be slightly paler or darker than -the original, attempts were made to obtain new varieties by selecting these. All such attempts gave negative results. It seems certain therefore that Wolf was not dealing with mixed cultures or with a race which was spontaneously mutating. By growing B. prodigiosus in media containing minute quantities of various salts, Wolf succeeded in obtaining a few races with per- manently altered colour. He obtained one white race and four dark red races. The white race appeared after 14 successive transplant- ations on a medium containing 0*01 % of corrosive sublimate (at 37"5° C). The dark red races arose in a similar manner, after a varying number of transplantations, on media containing minute quantities of potassium permanganate, cadmium nitrate, corrosive sublimate, and potassium bichromate'. All these changes were abso- lutely permanent — the colonies never reverting to the original colour when grown for a long period in normal media'. In some other experiments the mutations were of a dififerent type. They occurred in four cases in organisms grown upon media containing potassium bichromate, copper acetate, cadmium nitrate, and nickel nitrate. All these mutated races were white, but they proved to be inconstant. When plated out, the white races gave rise both to white colonies and to colonies of the original red colour. The white colQnies behaved again in the same way — giving when plated out some red and some white colonies. A white race could be con- stantly maintained in this manner by selection, but it was, obviously, differently constituted from the permanent white race obtained by the action of HgCIj. These impermanent white forms Wolf calls "reverting mutants." One of the most striking features in these experiments is, I think, the fact that the same chemical substance may produce several quite different mutations. Thus, HgClj produced a pure dark red race, and a pure white race — that is, mutations in two different directions (in- tensification of colour and loss of colour). Similarly, Cd(N03)2 produced a constant dark red mutation and a reverting white. The action of K,Cr207 is even more remarkable. Under the influence of this salt, 1 These also were in cultures kept at 37o° C. But with KjCrjO^ the mutation was obtained at room temperature also. ^ In Wolfs experiments raising the temperature produced only a white modification in colonies grown on ordinary media. But it is stated by Migula (Sy$t. d. Baet., Vol. ii., p. 845) that permanent white races can be obtained by this procedure. 340 Mutation in Micro-Organisms the original bright red colony gave rise to a dark red mutation, and a white reverting mutation. In addition to these, it produced a white modification. And when the dark red form was subjected to the continued action of the bichromate, it produced in turn another white reverting mutation. Certain substances seem, therefore, undoubtedly to cause profound changes in the pigment-metabolism of B. prodigiosus. But it seems that the changes themselves are purely a matter of chance : for they may be permanent, partly permanent, or imper- manent, and in either of two opposite directions. Another note- worthy point is that two quite different poisons may produce the same sort of mutation. For example, HgOla and KgCraOy may both give rise to dark red races. In his investigation of Staphylococcus pyogenes aureus, Wolf obtained less definite results'. In some cultures temporary modifications of colour appeared, but he was never able to produce a permanent change — or mutation — by chemical means. Only one mutation occurred, and that was in the control series of cultures on ordinary media. On plating out the organisms at the 22nd transplantation. Wolf found three white colonies among the remaining typical dark-yellow aureus colonies. These white races subsequently bred true. They never reverted to yellow. Moreover, they were found to answer all the cultural and other tests of the organism known as Staph, pyogenes alhus\ It would seem, therefore, that alhus may arise under " normal " condi- tions from aureus : though — as the author says — the mutation arises " from unknown causes." Wolf did not obtain any other colour mutations in the case of true Bacteria. But he made some curious observations upon the Myxo- bacteria^ which he also investigated. (He observed a number of transient colour changes (modifications) in Myxococcus races, but these need not be considered here.) It had already been shewn by Quehl that physiological differ- ences exist between different races of Myxobacteria. If two swarms* 1 Neumann had previously stated that he was able to select races of several different colours from this form. Wolf, working upon a pure line, was unable to do this. 2 Several similar alhus mutations occurred in the course of plating out subsequent sub-cultures of this series. 3 The Myxobacteria constitute a remarkable group of the Protista. They differ in many ways from ordinary bacteria, and their systematic position is still very doubtful, * The swarms consist of large numbers of separate individuals (like ordinary bacteria in some respects) invested with a common covering of slime. The swarms give rise to fructifications (I'esembling those of the Mycetozoa) in which spores are formed in a peculiar manner. C. DOBKLL 341 belonging to the same race are brought side by side on the same culture medium, they are seen to fuse as soon as their edges come in contact. In the case of different races, however, no fusion occurs between the contiguous edges under such conditions. Wolf endeavoured by various means to obtain from a pure race (in which the swarms readily fused) physiologically different races which would not coalesce when brought in contact. He easily succeeded in doing so in races belonging to two different species — Myxococcus rubescens and M. virescens. He found that the same race could produce swarms which would not fuse, if they were cultivated for a sufficiently long time. The change occurred even when the organisms were grown on the same medium and under the same conditions {M. rubesceiis). But the change took place more rapidly if the swarms were cultivated at different temperatures, on different media, or on media containing various salts (KjCrjOy, KNO3, etc.). The peculiar feature of the change was that it was irreversible. When once the swarms had lost the power of fusing, no amount of further cultivation on normal media would induce them to revert to their original condition. It seems obvious, therefore, that under certain ill-defined conditions, a permanent physiological change may occur. But it is by no means obvious what the nature of the change is — its only manifestation being an inability to fuse, which might be due to several different causes. It seems impossible at present to draw any further conclusion from these observations. A place must here be given to some statements just made by Baerthlein (1912 a), as they concern the colour mutations of B. prodigiosus and Staph, pyogenes. This worker states that by merely plating out on agar old cultures (in agar or broth) of B. prodigiosus, he can obtain numerous mutated colonies. These may be dai-k red, pink, white, white with red spots, red with white sectors, etc. The individual organisms in these different colonies also differ structurally from one another. These mutated races are said to breed true. They preserve all their characteristics when cultivated further on various media or after passage through animals. "Atavistic phenomena in the form of reversions" occur, however, if the mutated colonies are left for a long time in the same medium, and then transferred to fresh media. Similar changes are said to occur in the case of B. pyocyaneus. Baerthlein's brief statements are not easy to reconcile with Wolf's careful work on the same organism. It seems possible that he has not been dealing with pure lines. 342 Mutation in Mia^o- Organisms Baerthlein states, further, that by a similar procedure he is able to make Staph, pyogenes aureus mutate into Staph, pyogenes albus^. The albus race breeds true. But by treating it similarly, it will "mutate back" into aureus races. Albus and aureus races are, moreover, said to differ as regards the structure of the individual organisms. Baerthlein records similar observations on a number of other bacteria. But until further details of his work are available, it seems useless to attempt to criticize his results or to correlate them with those of Wolf and other workers in the same field. A case of an altogether different sort may now be mentioned. Every bacteriologist is familiar with the fact that the power to form spores is a very variable character in some races of bacteria. Sporo- genic and asporogenic races have been observed in many different species. In a recent paper, Eisenberg (1912) has given some inter- esting facts concerning this matter in the case of anthrax bacilli. His results are particularly noteworthy : for, although very many similar observations have been made, the interpretation is not ex- cluded that the results were due merely to a selection of certain pure lines from an originally mixed population. Laboratory cultures of Bacillus anthracis may consist — according to Eisenberg — of mixtures of sporogenic and asporogenic races (most frequently), of pure asporogenic races (less often), or of pure sporogenic races (seldom). Pure sporogenic and asporogenic races may be obtained from a mixed culture ; the former by the action of heat, the latter by constantly transplanting young cultures on to fresh media. A pure sporogenic race may thus be obtained. And Eisenberg has found that if it is constantly grown (5-20 transplantations) on glycerin-agar, it completely loses its power of forming spores. These asporogenic races bred true for a considerable time. They never reacquired the power of forming spores. The conclusion appears to be justified, therefore, that in B. anthracis a sporogenic race may mutate into an asporogenic race under certain conditions (in the present case under the influence of glycerin ?). In his ingenious experiments with bacteria. Barber (1907)^^ en- deavoured to obtain asporogenic races by the isolation of individual organisms. A culture of organisms isolated from the juice of the sugar-cane {B. megatherium ?) was found to contain some individuals ^ Compare Wolf, p. 340 supra. 2 vide infra, p. 345. C. DOBKLL 343 which formed spores and others which did not. Barber isolated a number of individuals of the latter type, and cultivated them further. In only one instance, however, did he succeed in thus obtaining a permanently asporogenic race. In all other cases the isolated individuals gave rise to spore-forming colonies. The mutation in this case may be called spontaneous. There is nothing to indicate that it was due to any particular external conditions. Some remarkable statements concerning "mutations" in the cholera vibrio have been made recently by Baerthlein (1911), and a further contribution to the same subject has just been made by Eisenberg (1912 a). It is said that pure cultures of this organism when plated out on agar constantly give rise to several different kinds of colonies — produced by " mutations " from the original race. The " mutated " forms are said to be in some cases constant, in others inconstant or capable of undergoing further " mutations " or " reversions." The facts at present available seem so confusing, and sometimes so contradictory, that I think no very satisfactory conclusions can yet be drawn from them. I will therefore merely refer the reader to this work without attempting to discuss it here. It seems legitimate to conclude from the foregoing facts that some races of bacteria are able permanently to acquire new characters under certain conditions : and also that they may in a similar manner lose these characters subsequently. So far we have considered the behaviour of bacteria under experimental conditions only, but one is naturally led to inquire whether similar changes occur in nature. The answer to this question is obviously of the greatest importance not only to the biologist but also to the medical man : and it has frequently been given both in the negative and in the affirmative. A little reflection on the problems involved in the question will suffice to make this easily explicable. It is, in fact, at present impossible to answer the question in other than a most tentative manner. Nevertheless, a few cases have been recorded which seem to throw some light on the matter. With one of the best and most recent of these I will conclude the present section. S^rensen (1 91 2) has just recorded the following remarkable facts. A patient suffering from glycosuria, developed in addition pneumaturia. This was found to be caused by a peculiar bacillus (B. pneumaturiae) which had gained access to the bladder, and by fermenting the sugar there produced large quantities of gas. The organisms were isolated 844 Mutation in Micro- Organisms and carefully studied in cultures, in which their behaviour was similar to that observed in the bladder. They fermented glucose, lactose, and saccharose with the production of much gas. After about two years, the patient recovered spontaneously from the pneumaturia, but the bacteria were still present in the bladder. Both here and in the cultures, however, they were found to have lost completely the power of forming gas by the fermentation of sugars. The cultures were kept for a long time subsequently, and repeatedly tested to see whether they would regain their gas-forming power. After about a year, these organisms suddenly reacquired the power of forming gas from lactose and glucose ; and shortly after, the patient began to suffer once more from pneumaturia. Examination of the bacteria from the bladder, showed that they had — like the organisms in the cultures — reacquired their gas-forming powers The behaviour of the organisms in the patient was therefore closely parallel to their behaviour in the artificial cultures. Moreover, the experiments seem to shew that the same bacteria were present through- out— that is, it was not a case of mixed cultures or a reinfection. It seems justifiable, therefore, to conclude that in this case at least a physiological character was lost and reacquired by the bacilli not only in artificial cultures, but also under *' natural " conditions in the living organism. B. Morphological Mutations. Concerning structural variation in bacteria very little indeed is known. This is not because structural differences between members of the same species are uncommon, but because the normal structure and life-history of nearly every species is still largely a matter of conjecture. Many species are probably polymorphic, the various forms depending partly upon the particular stage which has been reached in the specific life-cycle and partly upon external conditions. Many bacteria, moreover, continue to exist and multiply after they have assumed a degenerate or abnormal form (so-called involution forms). The modifications dependent upon these various factors are not variations in the ordinary sense of the word. One would be equally justified in classifying a spermatozoon, a foetus, a leper, and a man with both his legs cut off, as structural variations of the human species. ^ Both the organisms in the cultures and those in the bladder had also acquired the new power of clotting milk. C. DoBKLi. 345 There is little to record concerning structural variations which are permanent — mutations, that is, which when they have once appeared breed true in subsequent generations. Only two cases of this sort will be noted here. First, the work of Barber (1907) on Bacillus coli^ must be mentioned. This observer began with pure cultures of bacteria, and grew all his sub-cultures on the same media and under the same conditions. The environment was therefore alike for all individuals — as far as possible. Barber noticed that there were constantly present, among the typical individuals in his cultures, a small number distinguished by their greater length. By a special method which he devised*, he was able to select these long individuals and propagate them further. No less than 140 such individuals were so isolated in one series of experiments. With a single exception they all gave rise to colonies consisting of individuals with the normal dimensions'. The variations were, in other words, modifications — not permanent changes. In the one exceptional case, however, he succeeded in obtaining a new race of long individuals. This race bred true. It was kept for 32 months — being frequently transplanted — without undergoing any change. Selection of maximal and minimal sized individuals of this race was also without effect: neither the original nor a new race could be obtained. The mutated race had partially lost its motility, and differed also in certain cultural characters from the original race. In another series of experiments, Barber isolated 50 long individuals from another strain of B. coli. In this instance he succeeded in obtaining a similar long race which proved to be constant as regards this character. It was found necessary sometimes to make several successive selections from the colonies of mutating organisms in order to attain a pure fixed race. Altogether three new long races were finally established. Similar experiments were only partially successful in the case of B. typhosus. No constant long race was obtained. The mutations, it will be seen, occurred spontaneously in all these cases. New races were established by the selection of individuals which had already varied. It seems that the long individuals which occurred in the original cultures were of two classes — though outwardly ^ Barber made a number of similar experiments with yeasts. ' The method is fully described in the original. It consists essentially in a direct selection, under the microscope, of a desired individual by means of a very fine glass capillary tube. ' That is, in those cases in which snccessfol cnltnres resulted. In many cases, the isolated individuals grew badly or not at all. Joom. of Gen. n 24 346 Mutation in Micro- Organising more or less alike. The majority represented merely temporarily modified individuals : only a very few were permanently mutated organisms. Thei-e is, of course, nothing to indicate what factors may have been concerned in the production of these forms in the first place. I will end this section by referring to some work which has just been published by Revis (1912). His observations also concern B. coli. A typical strain of this organism was grown in peptone broth to which malachite green had been added'. The effect of this dye was to produce a new race of organisms which differed both structurally and culturally from typical B. coli. When the organisms were grown subsequently at 20° C. on ordinary gelatin or agar, they formed " large viscous, circular masses," consisting of "a mixture of very long fila- ments^ and short bacilli, together with a gummy cementing substance." Presumably the race breeds true to these new characters. The organisms were not propagated by the isolation of individuals, but the cultures were very carefully plated out. The purity of the original culture is guaranteed. Revis therefore claims to have produced — by tneans of malachite green — from a typical B. coli a new race " which is neither physiologically, morphologically, nor culturally a colon bacillus." Further details and confirmation of these observations are to be desired. Concluding Remarks. To epitomize in few words the numerous facts given in the fore- going pages is hardly possible, for what has been written is itself one long epitome of facts and their interpretations. A few general remarks may, however, be permissible fco a writer who has travelled thus far over very stony ground. They may serve, moreover, to call attention to certain facts which are undoubtedly important — though the magni- tude of their importance and their significance may well be appraised in very different terms by different individuals. It must be understood, ^ See also p. 337 supra. 2 It was already known that organisms of the coli-typhosus group — and others also — assume a remarkable filamentar form when grown on media containing certain dyes. The first observations in this connexion were made in 1904 by Walker and Murray (British Med. Journ., Vol. ii. p. 16). Similar results have since been obtained by Vay (G. B. Bakt., I. Orig. 55, 1910, p. 193) and others. C. DoBKi.L 347 then, that the following remarks embody merely my own conclusions drawn from the facts given in the body of this paper and from many other related facts of which no mention has been made. They lay claim to no finality, for the subject is not one upon which any final judgment can yet be passed. If it be assumed that the statements made by various workers, whose observations we have been considering, are correct, then the following conclusions are justifiable. First, it seems established that the Bacteria are subject to mutation — that is to say, in a given race individuals may occur which differ from their fellows in their genetic constitution. Individuals frequently occur which possess new structural or functional features; and these features, though often the transient peculiarities of the individual only, are in some cases transmitted to the offspring for many successive generations. There is reason to suppose that this phenomenon occurs in nature as well as in laboratory cultures. The progeny of an organism which varies may thus constitute a new race, in which every individual possesses the new character. We might anticipate this, indeed, by consideration of the fact that the Bacteria are non-sexual organisms. For a change in the genetic constitution of the parent — where there is but one — appears likely to find expression in all its offspring. There is no additional complication — in the trans- mission of characters — introduced by a second parent. In sexually- producing organisms, the genetic constitution of two parents must always be considered, and there is not, therefore, such an obviously direct relation between any one parent and its offspring as is seen in non-sexual forms. It seems impossible to gauge the permanency of new races which arise in this fashion. For there are indications that a new race may give rise to other new races or to one indistinguishable from the old race — all races arising in the same way. A race A may produce an abnormal individual, which becomes the ancestor of a new race B. In the same way, the race B may produce abnormal individuals giving rise to races C, D...etc., of which one may be identical with A. There is at present little to indicate the extent to which " reversion " of this sort may occur. The factors which determine changes in genetic constitution are in most cases obscure. It is impossible to say how most mutations have been "caused." In Barber's experiments, the environment was the same for all individuals — or at least he tried to make it so. The factor which determined the appearance of individuals with an altered 24—2 348 Mutation in Micro-Organismg genetic constitution can therefore hardly be sought — for the moment — anywhere but in the organisms themselves. In Wolf's experiments, on the other hand, there appears to be evidence that the variations depended in some way upon the environment ; for they occurred most frequently in organisms subjected to the action of poisons. But the relation between the mutation and the chemical is not apparent. The same substance seems to be able to produce two opposite results ; and two different substances seem capable of producing the same result. The remarkable phenomena so well studied by Massini, Burri, and others seem at first sight more illuminating. It appears that certain bacteria, which cannot ferment a certain substance, can acquire the power of fermenting it if kept in contact with it for a sufficient time. At first the organisms cannot avail themselves of the new food around them, but they then undergo a change which enables them to do so. It seems at first sight that the new power is the result of necessity — the "mutation" being a direct and indispensable adaptation to a definite end. But to necessity alone the change can hardly be ascribed — even by a confirmed Aristotelian. For it is apparent (e.g. from the work of Massini) that the ability to attack a certain substance (in this case lactose) is not a necessary condition for the survival of the race. It is, rather, a luxury. The lactose-splitting individuals arise and flourish in a non-lactose-splitting colony ; but the latter can survive for a very long period, and there is nothing to prove that the new race would supplant the old as a result of natural selection. If every individual — as Burri supposes — possesses the power {in posse or in esse) of fermenting the sugar ; and if, under ordinary cultural conditions, only a minority avails itself of this power : then surely it seems absurd to suppose that the splitting of the sugar is necessary for the survival of the race. These considerations do not affect the fact, however, which seems to be established that there is a direct relation of some sort between the sugar and the change which it produces in the organism. The action of the sugar is specific. Lactose, and lactose alone, makes B. coli mutabile able to ferment lactose, but does not enable it to ferment saccharose or any other sugar. We are not dealing here with a stimulus which may produce one of two opposite reactions, or with a reaction which may be produced by another stimulus. Variations of this sort seem to stand in a class by themselves. Pringsheim (1910) calls them "functional adaptations" or "accom- modations." It has been maintained that all variations in Protista are C. DOBELL 349 really of this sort : but there is little ground for the foundation of such a hypothesis. Many established facts appear to be flatly contradictory. It should not be forgotten, however, that acquired heritable adaptations of this sort are described not only among the Bacteria, but also among the yeasts and other Fungi ; and the observations have been made by many independent and competent workers. These few remarks by no means exhaust the significant inferences to be drawn from the works which have been under consideration. But it is hoped that they may serve to call more attention to certain facts which have hitherto been left in the obscure by-ways of bacteriology. It will be admitted, I think, that the outcome of the work fragmentarily recorded in the foregoing pages will be not merely of interest, but probably of very great importance, to every student of genetics. LITERATURE. 1911. Baerthlein. "Ueber Mutationserscheinungen bei Bakterien." (Ber. ub. v. Toff. d./r. Ver. f. Mikrohiol Dresden, in : C. B. Bakt. i. Abt (Ret) Vol L. Beiheft, p. 128.) 1912. . " Untersuchungen iiber Bact. coli mutabile." (C. B. Bakt. i. Abt. (Orig.) Vol. LXVL p. 21.) 1912 A. . "Weitere Untersuchungen iiber Mutationserscheiuungen bei Bakterien." {Ber. ub. d. vi. Tag. d. fr. Ver. f. Mikrobiol. Berlin, in : C. B. Bakt. L Abt. (Ref.) Vol i.iv. Beiheft.) 1907. Barber, M. A. " On heredity in certain micro-organisms." {Kansas Univ. Sci. BvU. Vol IV. No. 3, p. 1.) 1909. Bexecke, W. Review of the work of R. Miiller, published in iS'. B. physiol. Ver. Kiel, 1909 : and in " Die Umschau," 1909. {Zeitsch. f. indttkt. Abstam- mungs- u. Vererbungslehre, YoL n. p. 215.) 1912. Bershardt, G. and Markoff, W. N. "Ueber Modifikationen bei Bak- terien." (C B. Bakt. L Abt. (Orig.) VoL Lxv. p. 1.) 1908. BcRK, A. " Mutation bei einem der Koligruppe verwandten Bakterium." (Arch. f. Hyg. Vol. lxv. p. 235.) 1910. BcRRi, R. "Ueber scheinbar plotzliche Neuerwerbung eines bestimmten Garungsvermogens durch Bakterien der Coligruppe." (C. B. Bakt. n. Abt. VoL xxvni. p. 321.) 1910. and Andrejkw, P. " Vergleichende Untersuchung einiger Coli und Paratyphusstamme." {C. B. Bakt. L Abt. (Orig.) Vol. lvi. p. 217.) 1909. and Duggeli. "Beitrag zur Systematik der Coli-aerc^nes-Gruppe, u. s. w." {C. B. Bakt. I. Abt. (Orig.) Vol. xlix. p. 145.) 350 Mutation in Micro-Organisms 1912. EiSENBERQ, P. " Untersuchungen iiber die Variabilitat der Bakterien. I. trber sporogene und asporogene Rassen des Milzbrandbacillus." (C. B. Bakt. I. Abt. (Orig.) Vol. LXiii. p. 305.) 1912 a. . "Untersuchungen iiber die Variabilitat der Bakterien. II. t)l)er sogenannte Mutationsvorgange bei Choleravibrionen." (C. B. Bakt. i. Abt. (Orig.) Vol. Lxvi. p. 1.) 1910. Jacobsen, K. a. " Mitteilungen iiber einen variablen Typhusstamm {B. typhi mutabile), sowie iiber eine eigentiimliclie hemmende Wirkung des gewohnlichen Agar, verursacht durch Autoklavierung." (C B. Bakt. i. Abt. (Orig.) Vol. LVi. p. 208.) 1910. KowALENKO, A. "Studien iiber sogenannte Mutationserscheinungen bei Bakterien unter besonderer Beriicksichtigung der EinzeUenkultur." {Zeitschr. f. Hyg. Vol. lxvi. p. 277.) 1912. Mandelbaum, M. " Ueber das Bacterium metatyphij' {C. B. Bakt. i. Abt. (Orig.) Vol. LXIII. p. 46.) 1907. Massini, R. "Ueber einen in biologischer Beziehung interessanten Koli- stamm {Bacterium coli mutabile)." {Arch. f. Hyg. Vol. lxi. p. 250.) 1909. MiJLLER, R. See Benecke (1909). 1911. . " Mutationen bei Typhus und Ruhrbakterien." {C. B. Bakt. i. Abt. (Orig.) Vol. LViii. p. 97.) 1910. Pringsheim, H. " Die Variabilitat niederer Organismen." Berlin. 1911. Revis, C. "Note on the artificial production of a permanently atypical B. coli." {C. B. Bakt. ii. Abt. Vol. xxxL p. 1.) 1912. . " The production of variation in the physiological activity of Bacillus coli by the use of malachite green." {Proc. Roy. Soc. B, Vol. lxxxv. p. 192.) 1909. Sauerbeck, E. " Ueber das Bacterium coli mutabile (Massini) und Coli- Varietaten uberhaupt." (C. B. Bakt. i. Abt. (Orig.) Vol L. p. 572.) 1911. ScHROETER and Gutjahr. " Vergleichende Studien der Typhus-coli- Dysenteriebakterien." {C. B. Bakt. i. Abt. (Orig.) Vol. lviil p. 577.) 1911. SoBERNHEiM, G. and Seligmann. "Weitere Beitriige zur Biologie der Enteritisbakterien." {Ber. d. v. Tag. d. fr. Ver. f. Mikrobiol. Dresden, in : C. B. Bakt. I. Abt. (Ref.) Vol. L. Beiheft, p. 134.) 1912. S0RENSEN, E. "Eine Untersuchungsreihe iiber die Veranderung einer Urinbakterie in den menschlichen Harnwegen." {C. B. Bakt. i. Abt. (Orig.) Vol. Lxil. p. 582.) 1911. Thaysen, a. C. "Studien iiber funktionelle Anpassungen bei Bakterien. (VorL Mitteil.)." {C. B. Bakt. i. Abt. (Orig.) Vol. lx. p. 1.) 1907. Twort, F. W. " The fermentation of glucosides by bacteria of the Typhoid- coli group and the acquisition of new fermenting powers by Bacillus dysenteriae and other micro-organisms." {Proc. Roy. Soc. B, Vol. lxxix. p. 329.) 1909. Wolf, F. " Ueber Modifikationen und experimentelle ausgeloste Muta- tionen von Bacillus prodigiosus und anderen Schizophyten." {Zeitsch. ind. Abstammit7igslehre, Vol. il. p. 90.) MATERNAL INHERITANCE AND MENDELISM. (FIRST CONTRIBUTION.) By K. TOYAMA {Zoological^ Institute^ College of Agriculture, Tokyo Imperial University.) (With Plate XX.) CONTENTS. PAOB I. Certain egg-characteristics of the silk-worm 352 n. The origin of those characteristics 353 m. Besolts of line breeding of certain variants 353 Series 1. The reddish-brown eggs fonnd in the normal divoltine " Shinkawachi " 353 Series 2. The blae-egged variant found in the normal divoltine •• Kuni-nishiki " 357 Series 3. The whitish-grey egged variant 360 Series 4. The spindle-shaped eggs found in the normal univol- tine breed 364 IV. Crossing of various breeds or variants possessing different egg- characteristics 367 Series 1. Crossing between Theophila mandarina and Bombyx mori 367 Series 2. Crossing between normal egged breeds and the whitish- grey egged variant 370 (a) Crossing between tetravoltine normal egged breeds and the variant 370 (6) ^^ossing between divoltine cross-bred normal egged w breeds and the variant 374 (c) Crossing between divoltine normal egged breeds and the variant 374 Series 3. Crossing between tetravoltine yellow and white. . 376 Series 4. Crossing between various breeds possessing special colour -characteristics of the ^^ 378 Series 5. Inheritance of the crimson-egged breed . . 381 v. General considerations 391 VL Summary 401 352 Maternal Inheritance and Mendelism In my first contribution to the study of hybridology of insects published in 1906 (10), it was shown that certain colour-characteristics of the egg of the Siamese silk-worm follow Mendel's law of heredity. Thanks to the kind suggestion of Prof. Bateson of London, we again undertook a similar series of experiments with various breeds of the silk-worm. Some of the results obtained during the last five years which seem to us to be not without interest to students of heredity will be described in the following pages. I, Certain Egg-characteristics of the Silk-worm. Before going further, we shall first enumerate certain egg-character- istics which are the subject of the present paper. Colour. The ordinary colour of the Japanese silk-worm eggs is a light greenish white when newly laid. With the formation of the blastoderm, the egg gradually assumes a brownish tint which at last turns into brownish slate shaded with some light pink or purple (Figs. 1, 3, 11). There may be found, however, many variants, some rather deeper, some lighter, and some with different shades. Now and again it happens that some eggs characterized by extraordinary variations of colour are found among normal ones, such as reddish brown (Fig. 2), whitish grey (Fig. 4), blue (Fig. 5), greenish slate (Figs. 6, 10, 12), crimson (Fig. 7), orange, greenish white and many others. Most of the eggs deposited by Japanese green breeds are more or less shaded with green. When newly laid, they are yellowish green and are much deeper in colour than those deposited by ordinary white breeds. Most of the Chinese or European breeds come in a similar category. Shape. The normal shape of the silk-worm eggs is oval, slightly pointed at one end, where a micropyle is situated (see Figs. 1 — 7). It is a little flattened and its surface is convex when newly laid but after a few days it becomes depressed in the middle, thus producing the characteristic form which is familiar to us. In this characteristic, as in the case of the colour, we observed many variants, some of them being quite extraordinary, for instance such as spindle-shaped eggs (Fig. 13) or other irregularly shaped ones, etc. which will be discussed minutely afterwards. K. ToYAMA 353 II. Origin of the Characteristics above enumerated. The egg consists of the shell, vitelline membrane, serosa and yolk, and each of them is coloured or shaded with certain tints or pigments, except the vitelline membrane which mostly remains colourless in nearly all breeds. The shell is usually translucent and is slightly tinted with certain colours. In Japanese breeds it is usually white or slightly shaded with brown, flesh-colour, green, or dirty-white or some other tint. That of some Japanese green, Chinese or European breeds is coloured yellowish green or pale green. The colour of the eggs is consequently more or less influenced by the colour of the shell. As to the shape, it is chiefly determined by the characteristics of the shell, which is derived from the epithelium of the oviduct. The cause of the egg-colour is, however, mostly due to the pigments deposited in the serosa which are seen through the shell. The colour of the yolk plays a certain part in the production of the egg-colour only in the case where the formation of the dark pigments in the serosa does not take place, i.e. in newly laid eggs or those oviposited by the spring brood of di-, tetra-, or multi vol tine breeds. The object of the present series of experiments is to know what influence, if any, these variants have upon the trend of heredity in their offspring. As to the origin of these variants we are quite ignorant whether they are produced by mutation or by hybrid mutation or some other causes which are yet unknown to us. We only know that they are seldom found among eggs laid by the normal-egged breeds generally reared in Japan. III. Results of Line Breeding of certain Variants. Series 1. The Reddish-Brmvn Eggs (Figs. 1, 2). In the winter of 1907, Mr K. Ishivata, one of the famous silk- worm breeders in the district of Fukushima in Japan, kindly offered me some normal (Fig. 1) and brown (Fig. 2) egg batches* laid by a divoltine white breed called " Shinkawachi," and said that both of them, even when inbred, gave the antagonistic characteristics in the offspring and thus it was very difficult to establish them as constant forms. * All through this paper, the word "batch " represents the total eggs laid by a moth. 354 Maternal hiheritarvce and Mendelism As the colour of the shell and yolk of both variants was the same, we must attribute the chief cause of those characteristics to the pigment of the serosa, a product of the combination of both parental gametes. We started our breeding experiments in the spring of 1908. The First Generation. 1908. Spring. We reared two batches of eggs from each variant. The normal series gave 72 matings or batches and the brown series 87, all of which were divoltine white in colour and we were unable to distinguish which were normal and which were brown. This characteristic, producing un- coloured eggs, is one of the normal characteristics of di-, tetra- or multi- voltine breeds in Japan. In these breeds, the eg^ laid by the spring brood generally produces no dark pigments in the serosa and conse- quently it remains whitish until the embryo is completely developed. Tropical multivoltine breeds, such as Siamese white or yellow which produce 8 or 9 broods in a year, never produce any dark pigment in the serosa, the colour of the egg therefore being determined by that of the yolk and the shell. Sometimes it happens that certain eggs of the spring brood of di- or multivoltine breeds turn into the ordinary dark slate-colour, in which case most or all of them become univoltine in character and do not hatch until next spring comes. On the contrary, all the eggs laid by the summer or autumn broods of divoltine or other multivoltine breeds deposit normal dark pigments in the serosa, thus giving various colours characteristic to the respective breeds. The Second Generation. 1908. Summer. Summer broods derived from the whitish eggs of the spring broods from normal and brown series yielded the antagonistic characteristics as shewn below. , 1. £!ggs laid by the Summer Broods of the Brown Series. Number of Matings Normal Batches Brown Batches Mixed Batches Totals No. 15. 11 0 11 7 18 No. 19. 12 10 19 29 58 Totals ... 10 30 36 76 K. ToYAMA 355 2. Eggs laid by the Summer Broods of the Normal Series Number Normal Brown Mixed of Matinga Batches Batches Batcbe* ToUls No. 18. 12 4 4 4 12 „ 14 2 5 12 19 „ 9 1 1 2 4 No. 4. 1 3 0 0 3 TotalTT^ 10 ~lo 18 38 In the former or brown series, 10 were normal, 30 brown and 7 a mixture of both normal and brown eggs in the same batch, and therefore in this series the normal colour-characteristics remained as recessive. The reverse is the case in the normal series, which produced 10 normals, 10 browns and 18 mixtures, that is to say, the brown is recessive in this series. The Third Generation of Normal Series. In the autumn of 1908, three normal-coloured batches derived from summer broods of the normal series were reared. They gave 58 batches, among which there were 28 normals, 14 browns and 16 mixed batches, that is to say, they again produced the antagonistic character. The third generation of the brown series or brown batches laid by the summer brood in 1908 were reared in the spring of 1909. They gave, as is usual, all divoltiue whitish eggs. The Fourth and Further Generations. The fourth generation of the normal series derived from normal eggs laid by the normal series in the autumn of 1908 gave divoltine white eggs, a few being univoltine normals and browns. The same is the case in the fifth generation which was reared in the spring of 1910. In the spring of 1911, i.e. in the sixth generation, we noticed for the first time that this series of normal characteristic inbreeding gave all normal batches with a few divoltine white batches, which may be considered to be normal coloured in character, that is to say, they breed true to parents. On the contrary, the fourth generation of the brown series or summer brood of 1909 yielded, without exception, brown eggs. Since then we have reared them through two generations without producing any antagonistic characteristic. Hence it may be said that this brown form is established as a constant form. Respective figures obtained by this series of experiments are given in Table I. 356 Maternal Inheritance and Mendelism TABLE I. BrotOTiregged Series. Number of Univoltine batches produced Number of Matings Total Normal Brown Mixture White batches 1908 (Spring brood). First generation. 16* 0 0 0 2G 26 19* 0 0 0 61 61 Totals ... 0 1908 (Summer brood). 0 0 87 Second generation. 19, 12* 15.11 10 0 19 11 30 29 7 36 0 Totals ... 10 1909 (Spring brood). Third generation. 19. 12. 9 0 0 0 32 „ 12* 0 0 0 35 „ 25 0 0 0 all white Totals 0 0 0 67 1909 (Summer brood). Fourth generation. 19. 12. 12. 1* 0 3 0 0 „ 2 0 2 0 0 „ 3 0 18 0 0 „ 4 0 8 0 0 ,. 5* 0 33 0 0 Totals 0 64 0 0 1910 (Spring brood). Fifth generation. 19. 12. 12. 1. 2* ., 5.8 26 18 Totals 44 1910 (Summer brood). Sixth generation. 87 58 18 76 32 35 all white 67 3 2 18 8 33 64 26 18 44 19. 12. 12. 1. 2. 1 0 107 0 0 107 2 0 116 0 0 116 3 0 86 0 0 86 4 0 108 0 0 108 5 0 105 0 0 105 6 0 50 0 0 50 Normal-egged Series. Number of Univoltine batches produced Divoltine batches Number of Matings 4* 18* Normal Brown Mixture White bai 0 0 0 30 ; 0 0 0 42 ^ Totals 0 0 0 72 1908 (Summer brood). 4. 1 3 18. 9 1 18. 12* 6 18. 14 5 Second generation. 0 0 0 12 0 4 4 0 5 12 0 Totals 15 10 18 0 1908 (Autumn brood). 18. 12. 5* 0 „ 9* 0 ,. 20* 28 Third generation. 9 4 0 5 8 0 0 4 0 Totals 28 14 16 1909 (Spring brood). Fourth generation. 18. 12. 5. 17 4 0 0 18 „ 9. 25 0 10 10 „ 20. 18* 2 0 0 20 Totals 6 0 48 1910 (Spring brood). Fifth generation. 18. 12. 20. 18. 19* 6 0 0 20 18. 12. 20. 18. 26 0 0 0 13 Totals ... 6 0 0 33 1911 (Spring brood). Sixth generation. 0 0 4 18. 12. 20. 18. 19 (5 matings) 25 Totals ... 0 572 0 0 572 | * Eggs laid by the mating marked with an asterisk are used as the parents of the next generation. K. TOYAMA 357 To avoid complication, we give below a graphical summary : Bs: brown egg batch; N=nonnal-coIoared; 1^= divoltine white ; 3f= mixture of brown and nonnal-colonred eggs in a batch. Common parent ... N Parent egg First generation inbred IT (87) >0 I IT (72) I II I Second generation inbred B (30) + iV(lO) + If (36) B(10) + N{15) + M{1S) Third Fourth Fifth Sixth t> i» »» >» jr(67) B(64) W{U) B(U) + N(28) + M{U) N{6)+B(l) + W{48) N (6) + W (33) B (572) N{25) + W (4) From the results above obtained, we observe that complete segrega- tion between the two characteristics, the brown and the normal-coloured, took place and that each may be established as a constant form from their common stock. It is much easier, however, to establish the brown as a constant form than the normal. Moreover, we learn that during two or three generations both characteristics even when inbred produce antagonistic characteristics in their offspring, a fact which apparently seems to run counter to Mendelian principles, but which in reality is in perfect accordance with the principles, as will be seen in " General considerations." Series 2. The Blue-egged Variant (Fig. 5). The phenomena of inheritance, similar to those above described, were observed in the inbreeding of the blue variant. This form is a sport from a divoltine normal-egged breed called " Kuni-nishiki," and is characterized by the special blue colour of the egg. In the spring of 1910, only one batch (No, 20) was obtained, through the Kindness of Mr S. Saito in Ghifu-Ken. The worms, cocoons, and 358 Maternal Inheritance and Me7idelis7n moths derived from them were all normal in character. They, inbred, gave 30 batches of eggs, of which, 12 batches were divoltine white, and the remaining 18 all normal-coloured, which suggest to us that all the batches in this generation would be all normal-coloured ones. Three divoltine white batches (Nos. 1*, 3 and 4) which were reared in the summer gave three kinds of eggs, some batches being blue, as in the parental blue, others blue shaded with a brown, which we called " intermediate colour " and the rest normal-coloured batches. There were no batches in which both blue and normal or intermediate forms were found mi.xed. Respective figures obtained are given below ; Number of Matings Number of blue batches produced Number of intermediate batches Normal- coloured Totals 20. 1* 21 54 68 143 20. 3 12 65 76 153 20. 4 14 46 60 120 Totals 47 165 204 416 Of 416 batches derived from three parent batches, 47 were blue, 204 normal-coloured and the remainder intermediate. In the blue- coloured eggs, we distinguished both light and dark-shaded ones. The former we called " light blue " and the latter " dark blue." Five blue batches (Nos. 6, 14*, 24*, 25 and 28) laid by the mating No. 20. 1 were reared in the spring of 1911. They oviposited, without any exception, 251 batches of eggs, all of them being divoltine white. Eight white divoltine batches were reared in the same summer. Both the light and dark blue series yielded again the antagonistic characteristics as shewn below : Number of Matings Light Blue Dark Blue Normal Totals 20. 1 light blue 14. 3 13 0 8 21 14. 15 21 12 15 48 14. 20* 4 4 0 8 Totals 38 16 23 77 20. 1 dark blue 24. 4 0 0 3 3 24. 9 1 0 0 1 24.14 2 0 2 4 24.25 0 1 2 3 24. 27 0 0 3 3 Totals 3 1 10 14 Grand totals . . . 41 17 33 91 Of the eight parent batches, four gave both blue and normal- coloured batches, two all blue batches and the rest only normal-coloured ones, the total number of batches produced by them being 91. K. TOYAMA 359 In the spring of 1912, six blue batches (three light and three dark blue) derived from the lineage which produced only blue batches were reared separately as in the former generations. They gave the following egg-batches : Number of Mfttinga Number of Univoltine blue batches Number of Divoltine white batches Mixed batches Total number of Dark blue series No. 20. 1. 14. 20. 2 0 63 0 63 »> )» 3 1 18 0 19 >i >> 4* 22 64 1 87 Light bine series No. 20. 1. 14. 20. 1» 5 91 0 96 >> >> 8 0 52 0 52 n >» 10 0 18 0 18 Totals 28 306 335 Of 335 batches derived from the six blue parent batches, 306 were divoltine white as is usual in the divoltine breed, and 28 were univoltine blue coloured and only one was a mixed batch consisting of both divoltine white and univoltine blue-coloured eggs. The summer brood derived from four batches of light blue series and four batches of dark blue series gave the following batches: — Light Blue Series. Number of parent batch Number of dark blue batches laid Number of light blue batches laid Normal Totals No. 20. 1. 14. 20. 4. 4 29 42 — 71 5 22 36 — 58 19 14 16 — 30 22 6 2 1(?) 9 Totals 71 96 Dark Blue Series. Number of parent batch Number of dark blue batches laid Number of light blue batches laid No. 20. 1. 14. 20. 1. 1 6 33 >> »» 12 20 45 >» >i 15 25 46 »» >> 16 19 49 >» »» 25 13 31 1(?) Normal 2(?) 168 ToUls 39 65 71 68 46 Totals 83 204 2(?) Now we are able to extract the blue-egged characteristic as a constant form. As to the light and dark forms, they seemed to be fluctuations of the same character, blue. 360 Maternal Inheritance and Mendelism Resume: B = light and dark blue batches ; / = intermediate ; N = normal ; W= divoltine white ; M=mixei batch consisting of divoltine white and univoltine blue-coloured eggs. Parent eggs B Eggs of the first generation N{20) I j second generation { N (204) + 1 (165) + B (47) } third „ W{251) fourth „ B(9){B (49) + N (27)} N (6) fifth „ {B(28) + Pr(306)} „ sixth „ all B (445) Series 3. The Whitish-Grey egged Variant (Fig. 4). This variant from the normal egged breed is characterized by the peculiar structure of the shell. As is well known, the shell of normal breeds is elastic and translucent, its surface being smooth. That of the variant, on the contrary, is rather brittle and opaque, and is milky white in colour, in consequence of which the colour of the serosa can barely be seen through the shell and thus a peculiar whitish grey colour is produced. The surface of the shell is not smooth as in the normal shell, but begins to become irregularly corrugated as soon as the ventral plate is formed. There is no depression in the middle, which is a common characteristic of the egg laid by normal breeds. In the spring of 1909, we obtained two batches of grey eggs, one being derived from the univoltine white reared in the district of Hyogoken, and the other which came from Fuktishimanken is derived from the normal divoltine white, " Aobiki." They were reared separately and each of them gave both normal and grey batches, that is to say, the former deposited one grey and one normal batch (Table III) and the latter 15 normals and 20 greys (Table II); no mixed batches were produced in these cases. (See Tables II and III.) In the second generation (summer brood of 1909), moths derived from both normal and grey eggs paired inter se, yielded again, with no K. TOYAMA 361 Parent egg TABLE 11. Pedigree of the Whitish-Grey Variant, No. 24. All whitish-grey r --1 1 First generation (Spring of 1909) 1 15 normals + 20 1 grey batches Normal-egged Series. Grey egged Series. Second generation Number Xumber of of normal Mstings batches No. 24. 3 4 Numbei of grey batches 4 Total batdMM 8 1 Namber Number of of normal Blatings batches 24. G 1 2 Number of grey batches 1 Number ^bi^ 0 Mixture of C and BG: T bat (Summer of 1909) 6 4 1 5 „ 3 2 5 0 „ 21 2 0 2 „ i* 10 15 0 ' — S „ 22 3 1 4 16 1 2 0 — ,, 24* 7 0 7 20 „ 16x1 10 0 10 1 0 0 2 Totals 20 6 26 Totals 25 34 0 — ~1 Third generation (Spring of 1910) 24. N 24. 1 2* 34 41 0 0 34 41 24. G 4. 1* „ 4 7 18 13 22 1 4 — 2 A 2x1 2 0 2 „ 25 15 9 0 — 2 3 21 0 21 „ 26 9 5 3 1 3x1 5 0 5 „ 28 18 8 1 2 4 4x1 22 10 0 0 22 10 Totals 67 57 9 — "l3 ,, 5 23 0 23 Totals 158 0 158 Fourth generation (Summer of 1910) 24.^^24.2.xVl* 142 0 142 24 G4. Gl. Gl* 0 7 36 — 4 2 24 2(?) 26 7 11 38 7 ~5 5 7 8 22 23 5 71 87 66 3 0 0 0 0 0 5 71 87 66 3 8 11 21 28 32 22 9 26 41 47 31 16 24 27 11 0 — 9 9 5 4 Totals 100 173 69 — Totals 398 2 400 34 Fifth generation 24. N 24. 2. No. 1. (N 10, 14, 18, 22, 25) Mixture of 5 batches do. No. 12 38 47 0 0 38 47 24. 64.G1.61.) G6 1 7 16 22 24 Mixture of 5 ) batches )" 0 0 0 0 2t 2+ 18 19 4 18 5 56 14 17 9 18 8 40 1 0 0 0 6 8 3 3 1 3 O Totals 85 0 85 i Totals 120 106 • Those matings marked with an asterisk are the parent of next generation, t Normal eggs found in these matings are not true normal form, the shell being rather thin when compared with that of normals. tBG=B-greys. Joom. of Gen, u as 15 362 Maternal Inheritance and Mendelism exception, normal and grey batches. Thus five matings of the normal series from the divoltine breed yielded 20 normals and 6 greys J six matings of the grey series from the same breed similarly gave 25 normals and 35 greys. (See Table II.) The third generation of the normal series which were reared in the spring of 1910 yielded all normal egged batches which when inbred remained true to parents in subsequent generations : i.e. they became homozygous. This was not the case in the grey series. Five matings of the grey series in the spring of 1910 (third generation) gave 67 normals and 67 greys, in addition nine batches of a new variant which we have as yet never observed in our breeds. This new variant is characterized by the thin translucent shell which has fine wrinkles over it and by a shape a little longer than normal eggs. There is no depression in the middle. We shall call this kind of variant " B-grey," since it more resembles the grey form than the normal ones. In the case of moths laying B-grey eggs the actual number of eggs laid is always much smaller than the number laid by moths laying eggs of normal colour, even though the parents belong to the same batch. The worms which came out from the BG are so weak that we can hardly get any moth and consequently we are unable to trace the order of its inheritance. Of six grey matings of the grey series which were reared in the summer of the same year (fourth generation), five again yielded 100 normals, 173 greys and 69 B-grey batches. One mating, on the contrary, gave no normal eggs except the grey and B-grey, the respective figures obtained being 7 and 36. The fifth generation derived from the grey mating, which in the last generation yielded no normal batches, gave 245 batches in which 120 were grey, 106 B-greys, 5 mixture of grey and B-grey, and 4 which look like an intermediate form between normals and B-greys. Details of figures obtained in each mating of each generation will be seen in Tables II and III. From the results above quoted, we are able to say that, as in the case of first and second series, both normal and grey characteristics segregate from one another, and it is easier to get rid of the antagonistic characters in the normal than in the grey. The appearance of the new form which may probably be due to the new combination of allelomorphs renders the phenomena of inheritance rather complicated. Hence if we consider the G and BG forms as a single form, the results come in the same category, which was mentioned in the former series of experiments. K. ToYAMA 363 TABLE III. Pedigree of Whitish-grey Eggs derived from the Normal Univoltine White. Parent egg batch, No. 12 All Grey ^gs (laid in the Spring of 1908) First generation (Spring of 1909) G (1) + N (1) H- 1 Second generation (Spring of 1910) G (2) + N (4) N (8) I 1 Third generation N (\S) + G (16) These figures will be again summarized graphically as below A. G = Grey egg batch, iyr= Normal, BG=B-Grey. 1. Parent eggs in 1908 (No. 24) Q -~i 2. „ 1909 (Spring) (N + G) 3. „ 1909 (Summer) N {N + G) (N + G) 4. „ 1910 (Spring) all N 5. „ 1910 (Summer) all N (G + BG) (N + G + BG) 1 {N + G 1 + BG) I j 1911 (Spring) all 'N (G + BG) B. 1. Parent eggs 1908 G I .J 1909 (Spring) I — (G + N) 1910 (Spring) {G + N) aU N 4. „ 1911 (Spring) (G + N) 25—2 364 Maternal Inheritance and Mendelisni Series 4. The Spindle-shaped Eggs (Fig. 13). In the early spring of 1909, we obtained half a batch of the eggs laid by a Japanese normal univoltiue white. The egg is long and spindle-shaped, and is slightly pointed at both ends. There is no depression in the middle which is a characteristic common to normal silk-worm eggs. The first generation which was reared in the spring of 1909 gave eggs which were quite normal in shape and other characteristics. The egg-batches obtained were only six in number. The second generation derived from the normal eggs yielded moths which paired inter se deposited 46 batches of eggs in which we found both normal and spindle-shaped ones, the number found in each mating being as follows : Number of Matings Number of normal batches Number of spindle-shaped batches Totals 1 18 5 23 6 15 8 23 Totals 33 13 46 Of 46 batches derived from two normal matings, 13 were spindle- shaped and 33 normal-shaped batches, no mixed ones. In the third generation which was reared in the spring of 1911, both normal and spindle-shaped eggs gave moths which, when inbred, laid two kinds of eggs, normal and spindle-shaped ; the respective figures obtained in each mating are shewn below : Eggs laid Number of Matings Normal batches Spindle-shaped batches Mixture Totals Spindle-shaped egg, No. 1. 10 j „ 12 V 25 2 — • 27 „ 24 ) Normal eggs, No. 6 (8 batches) 24 3 1 28 Owing to the great havoc made by " flacherie," the mortality of worms was so great that we only obtained a small number of moths, yet we are able to prove that both characteristics even when inbred again produce the antagonistic characteristic. Thus the order of inheritance of these characteristics may be represented as below : K. ToYAMA 365 S = spindle-shaped eggs; ^r= normal-shaped. Parent eggs ... ... S I Eggs of first generation N I i Eggs of second generation {N + S) Eggs of third generation (N + S) (N + S) Although we are not yet able to establish this variant as a constant form, we may infer from the above facts that it comes in the same category as the variants just referred to. Of the various egg-characteristics discussed in the series of experi- ments above referred to, we know that those in the first and second series are derived from the colour of the serosa, those in the third and fourth series from the shell, whose special structure gave the egg some characteristics different from normal-shelled eggs. Notwithstanding their origin being different, their order of inherit- ance is nearly the same. Let us now compare the results obtained in the third and fourth series, which are represented as below : 1, The Resnlts of the Third Series. (G and BG are considered to be a single character G.) Parent eggs G Eggs of first generation ... (.V + G) ._! I. I III second „ ... X (N + G) (^ + G) i third „ ... N (.Y + G) I 1 fourth „ ... N ^o + If) + G fifth „ ... .V 366 Maternal Inheritance and Mendelism The Results of the Fourth Series. Parent eggs s Eggs 0 f first generation second ,, {F I, 'I >> + • 1 1 r 1 ■ 1 1 >> third [N + 1 1 {N + 1 S) In this case, if we consider the parent egg, G in the third series, as the N of the first generation of the fourth series, both results come in a single form which may be represented as below : A = G in the third and N in the fourth series ; B = N in the third and S in the fourth series. Parent eggs ... Eggs of first generation „ second ,, ,, third „ ,, fourth ,, „ fifth „ sixth ,, I B I I ^ (A + B) I {A + B) I r^ n I I (A + B) 1 I 1 I I A {A + B) I A If we compare the results obtained by the first and the second series of experiments, we can easily see that they behaved in inheritance in a similar manner to those above described, certain inconsistent results being due perhaps to the appearance of divoltine white eggs which prevented the elimination of the antagonistic characteristics during one generation. K. ToYAMA 367 IV. Crossinq of various Breeds or Variants possessing DIFFERENT EGG-CHARACTERISTICS. Series 1. Crosses between the vrild {Theophila mandarina, M.) and the domesticated {Bombyx mori, L.) mulberry silk-worms. The egg of the wild mulberry silk-worm (Fig. 10, a, b) is deposited in a small group on stems or twigs of the mulberry tree. When newly laid, it is a straw yellow (Fig. 10, b) which with the formation of the blastoderm gradually assumes a brownish tint and at last turns greenish grey (Fig. 10, a). The shape is nearly the same as those of cultivated ones, while the size is little smaller than the latter. The shell is straw yellow and translucent. The egg of the domesticated silk-worm used in this series of experiments is light greenish white when newly oviposited. It gradually assumes, as in the wild form, a brownish tint and turns brownish slate with some shade of purple or pink, i.e. it assumes the normal colour of Japanese silk-worm eggs (Figs. 1, 3, 11). The shell is nearly white, rarely faintly shaded with a greenish, brownish, or other tint. In the spring of 1905, five wild female moths were mated with domesticated males (tetravoltine Tobuhime). They deposited, with no exception, eggs whose characteristics are the same as those of pure wild ones in every respect, such as colour, shape, size and brood character (voltinism). On comparing them with those laid by pure wild parents, we were not able to find any difference at all. Eleven reversed matings (uni-, di-, and tetravoltine females with wild males) gave, on the contrary, eggs which are similar in shape, colour and voltinism to those of pure domesticated ones (Fig. 11). Even the eye of experienced breeders is not able to distinguish the cross-bred eggs from those laid by maternal pure breeds. Other five batches derived from divoltine females in the spring brood mated with wild males are all divoltine white in colour, and thus the order of inheritance is entirely maternal, no paternal influence being observed in those reciprocal matings. The worms which emerged from the reciprocal F^ eggs were reared in the summer of the same year. Moths derived from the eggs laid by the wild female matings (five matings) gave 56 batches of eggs, all 368 Maternal Inheritance and Mendelism of them being uniform in their characteristics. When oviposited, they were light greenish yellow and much lighter in colour than the ^i and gradually assumed a brownish colour which finally turned a greenish slate. Although they resemble the F^ eggs very much in colour, they are darker and the colours are more decided than the latter, and have no dirty or dull shade which is common in the eggs of Theophila. The shell is of clear greenish yellow and deeper than that of the .^i or pure wild forms. Therefore, we may easily distinguish F^ eggs from F^ eggs. There are, however, certain variations of colour in the same batch or between different batches, but no trace of the colour-characteristics of the domesticated parents, and consequently we may say that the colour-characteristic derived from the wild parent dominates over that from the domesticated parent. In the spring of 1906, we reared worms derived from the F^ eggs. Owing to the prevalence of grasserie and diarrhoea, all of them died without attaining their mature stage. It will be noted here that, as far as our experiments went, the hybrid form is much more easily injured by those diseases than the pure domesticated form, especially in the case where the male parents were of the wild form. It is, therefore, very difficult to rear a good supply of the hybrid form for experiments. We were therefore compelled to continue our experiments with back- crossed form paired with domesticated one which is healthier than the first cross. Back-crosses. In the summer of 1905, cross-bred moths from the ^i eggs were mated with pure domesticated ones. Reciprocal matings gave, as in the case of Fi, diametrically opposite results. Four Fi females mated with pure domesticated males gave all greenish slate eggs whose colour is quite the same as that of the F^ eggs before mentioned, while those laid by 12 domesticated females (tetravoltine white) mated with the cross-bred Fi males produced, without exception, eggs with character- istics quite maternal. The worms derived from the former matings all died in consequence of the two diseases above mentioned, while those from the latter, being much more able to resist those diseases, gave some moths in the end of the autumn of 1905, that is to say, the third brood of 1905. 36 batches of eggs resulted from the inter se breeding, in which we found many different coloured batches as is shewn below : Number of Matings Or«eniflli-«I*t« batches I 0 Ih 1 11 a 9 lie 0 lie S K. ToYAMA 369 Mixture of Japaneae nonnal Tariona colour- batches shaded eggs Totals 2 8 5 0 0 1 8 5 22 2 18 0 2 5 Totals 13 12 11 36 Of 36 batches or matings obtained, 13 were greenish slate as in their parents, 12 Japanese normal colour, the rest being a mixture of both kinds of eggs and some intermediate ones in various pro- portions. Owing to certain variations found in a batch, or between various batches, and the scanty number of matings obtained from each parent, we are unable to give the exact numerical proportions of these various coloured batches, but we certainly see that the uniform coloured F-i characteristic disintegrated into various coloured forms. In the spring of 1906, we reared worms derived from normal- coloured eggs. Moths paired inter se gave all divoltine white eggs. The fourth generation from these divoltine white eggs were reared in the summer of 1906. Three matings gave three kinds of egg batches, as in the former generations, namely : Number of Matings Greenish-slate Japanese normals Mixture Totals Ila, 3 5 7 3 15 .. 4 4 10 5 19 „ 5 2 4 10 16 In 1907, we made similar experiments. The worms which came out from normal-coloured batches gave all divoltine white eggs in the spring. The summer broods gave 30 batches of eggs, all of them being of the normal Japanese colour. Since then they liave bred true to parents, never giving any greenish coloured ones. Resume : T = greenish-slate coloured batches like those laid by pure Theophila. N= brownish-slate coloured egg batches, common to normal Japanese breeds. if = mixture of those two kinds of eggs above-mentioned and some intermediate coloured eggs. JF= divoltine white. (1) (2) ( ? Theophila x d Bombyx) ( ? Bombyx x « 4 0 2 2 — 4 11 39 0 0 39 ,, 6 16 32 18 — 66 17 37 0 0 37 ,, 11 0 14 18 — 32 20 41 0 0 0 41 29 " 44* 0 42 35 — 77 .. 77 29 0 Totals 2 matintrs 28 71 39 1.38 Totals 248 0 0 248 3 5» 0 58 55 — 113 2. 8. e , BG 3 0 11 11 22 >» BG 9 0 1 3 — 4 >J BG 12 0 0 1 — 1 Totals .. 0 12 15 — 27 2. 4. 1. 8 6 15 8 _ 29 9 0 8 12 — 20 12 11 14 7 — 32 17 2 13 6 — 21 20 3 6 3 — 12 • 27 0 4 6 — 10 19 3 4 3 — 10 Totals 5 matings 25 52 27 — 104 2 " 0 12 18 — 30 i-7 1911 2. 2. 7. 5. 13. 3 (Spring) „ 6 Fg 2. 8. 6. 44. 3 0 5 8 — 13 17 - 243 0 0 243 6 0 4 1 — 5 18 21 0 2 5 — 7 19 6fe 0 3 5 — 8 96 0 15 9 3 27 126 0 1 2 1 4 12c 0 7 12 0 19 16c 0 6 6 1 13 19c 0 6 1 0 7 Totals 0 49 49 103 * Those marked with an asterisk are the parents of next generation. 374 Maternal Inheritance and Mendelism The respective figures obtained in each mating in every generation are represented in Table IV which will be summarized as below : (Normal x grey No. 24) Ft eggs r I Fi eggs N F3 eggs iV I Fi eggs N Ffi eggs N Fg eggs N N 1 {N + G) N (O + BG) "1 {N + G + BG) I I I I I I {N + G + BG) (G + BG) {G + BG) (N + G + BG) I ill (G + BG) (N + G + BG) (G + BG) (G + BG) The results are nearly the same as those obtained in the original grey variant which is inbred, except the appearance of the constant N form in F^. (b) Exactly the same results were obtained when we crossed a normal-egged breed extracted from a cross between divoltine " Chiyodzuru " and the univoltine albino before referred to with the gi-ey-egged variant No. 24. The results of experiments are tabulated below (Table V). (c) In this series we made again similar matings as in the previous two series of experiments. Three females from the divoltine normal- egged white called " Shinkawachi " and two from the normal-egged extracted form called E III were mated with males derived from the very same grey variant used in the preceding experiments. They gave as in former cases, all Fi normal egg-batches, no grey ones. Two batches from the former and one from the latter were reared in the next season. The former gave 93 F2 batches and the latter 13 F2 batches, all of them being normal batches. Five batches of the former and one batch of the latter were again reared in the next season. The former gave 968 F3 batches of normal eggs and the latter 10 ^3 normal batches. I\. TOYAMA 37o Fi eggs (Spring, 1909) Fj eggs (Summer, 1909) ^3 eggs (Spring, 1910) TABLE V. ( 9 formal x 14 18 12 0 30 „ 13 6 4 0 10 „ 8 ., 9 8 17 1 4 i 1 0 28 22 86 6.16. 7 (B-grey 0 series) 1 30 31 „ 8 0 11 10 21 .. 18 0 0 4 4 6. 18. 1 0 0 6 6 6. 19. 1 0 2 0 2 „ 8 0 14 3 17 Grey series Grey 1 series Fa eggs (Spring, 1911) Number of Parent Normal Gr ey B-grey Total Number of Parent Normal Grey B-grey Total 6. 7. 3 I „ „ „ .- 6. 16. 12 ) (4 and 5) j ^ ^ o ii (5andl6))" 16 24 * Parents of next generation. In the next generation paired inter se, both series again yielded only normal- egged F^ batches, the number of batches produced in the former series being 66 and in the latter 123. Now it is quite certain that in these matings there is no grey factor which has lain dormant as in the former matings. The facts obtained in these three series of experiments and those from the second series of the line breeding suggest to us, firstly that among the eggs of grey-batch No. 24, there are two kinds of grey eggs, one having the grey factor in its zygotic composition, while another has no grey factor, in spite of its being grey in colour; that is to say, some grey eggs are heterozygous for the normal factor, some homozygous 76 Maternal Inheritance aiid Mendelimn normals ; secondly that the normal and the grey segregate from each other in their succeeding generation as other Mendelian characters do ; thirdly that it is much easier to free the normal form from the antagonistic character than the grey ; and, lastly, that the normal form does not produce any other form when it becomes free from the grey form, while the grey form segregates into another form even after being freed from the normal form. From this fact we may safely infer that the grey is more complicated in its constitution than the normal. Series 3. Grosses between Yellow and White forms of Japanese Tetravoltine Breed, " Onodahime." (Fig. 9, a and b.) Normal Japanese tetravoltine breeds are generally white cocoon- spinners, as far as we are aware. In the year 1905, we obtained a mixed breed consisting of white and yellow cocoon-spinners, the latter being a yellow-blooded form. Each form was reared separately and was established as a constant form. In the spring of 1907 reciprocal crossings between these forms were made. Yellow females mated with white males gave Fi eggs which are all yellow when newly laid^ (Fig. 9, b). This is the characteristic egg- colour of the yellow form. The reversed mating gave, on the contrary, all pale white Fi eggs (Fig. 9, a) which is also the characteristic colour of the white form. It will be necessary to note here that the colour of newly laid eggs is determined by that of the shell and the yolk, both of them being maternal in their origin. The worms which came out from the reciprocal Fi eggs were reared in the late autumn of the same year. All the worms were yellow- blooded and spun yellow cocoons without any exception. The moths paired inter se gave all yellow F^ eggs which are quite the same as those laid by the pure yellow forms. In the spring of 1908, the Fz yellow eggs gave two kinds of worms, the one being yellow-blooded, the other white-blooded ; the total figures found in those matings are shown below : — Number Number of yellow- of Group blooded worms I 485 II 567 Totals ~ Z. 1,052 Mendelian expectation 1,035 1 In this series of experiments, we only refer to the colour of the egg when newly laid, i.e. the colour of eggs before the formation of the blastoderm takes place. Number of white- blooded worms Totals 169 654 160 727 329 1,381 345 1,380 K. To YAM A 377 The moths derived from these yellow- and whitu-bluoded worms were paired in the following ways : I ? White- blooded moths x g White-blooded II ? White „ „ X „ ,, M 10 ,, ,, >> „ f» „ »i „ 13 ,, „ „ ,, ••>. ,, •' ,, 21 " " " " " " II II No. 6 white yellow pale white 59 51 110 ,, „ 13 „ ,, „ 17 0 17 ,, ,- 14 ,, „ ,, 242 0 242 »» „ 23 " " '• 27 19 . 46 III II No. 9 yellow yellow brownish yellow 0 0 0 „ „ 3 ,, ,, » M 109 30 139 ,, ,. 4 „ „ »i >i 39 14 53 i> » 5 »i ,, greenish yellow 0 0 0 „ >, 11 „ ,, brownish yellow 71 17 88 „ „ 16 >) ,, )> >> 0 0 0 " „ 24 " >• greenish yellow 54 13 67 Totals 273 74 347 IV I No. 6 yellow white yellow 65 55 120 As the Table shews, if yellow -blooded females (both homozygous and heterozygous) are used, whether they are mated with their own males or other white males, the results are always the production of brownish or greenish yellow eggs which is a characteristic of the yellow-blooded form. In like manner, white-blooded females mated with yellows (homo- and heterozygous) or whites gave all whitish eggs, characteristic of the white form. Thus we may say in thi^ case as in the other cases before cited, the colour characteristics of the egg are not influenced by the zygotic composition of the egg after fertilization, but by their maternal zygotic constitution before fertilization, and therefore there is no sign of male characteristics to be seen in the egg, but in the larval stage their relation is quite Mendelian. Journ. of Gen. ii 26 378 Maternal Inheritance and Mendelism Resume : r= yellow eggs ; W^=white eggs ; (y) = yellow-blooded worms and moths ; (w>) = white- blooded worms and moths. (1) (2) ? (i/) X f >i 2 >i If >• 3 >> >f »» 3 all Theophila coloar 2 >> >i 380 Maternal Inheritance and Mendelism (6) Reciprocal crosses between European breeds (Italian white and Sina blanc) and Japanese normal-coloured breeds (divoltine Shinkawachi, tetravoltine Tatsutahirne, and divoltine Asakanishiki). The egg-colour of these European breeds are practically the same as that of Papillons noirs, i.e. greenish-shaded slate. Matings $ Italian white x s Shinkawachi ? Shinkawachi x ^ Italian white ? Tatsutahirne x >i «» The extracted normal breed Divoltine white (Eenzoku) Male parent Crimson Number of Mating 6 Eggs laid (Fi) divoltine white 7 11 12 normal dark colour 13 14 15 i» »» »> >i 16 K. TOYAMA 383 Of eight matings, five gave all normal dark eggs, while three gave all divoltine white eggs, which is the maternal characteristic of tetra- or divoltine females in their spring brood. 2. ( ? crimson breed x ^ normal breed.) The reversed matings gave results identical with those represented. Female parent ( Crimson ( i» Male parent Tetravoltine Onodahime If »' The extracted normal-egged \ form from the cross be- 1 tween Bombyx mori and [ Theophila mandarina ) Chinese "Dragon horn " Number of Matings Eggs laid [Fi) normal dark 2 and 3 divoltine white greenish slate The results of the reciprocal matings shew us that in the ^i eggs normal coloured or greenish coloured characteristics dominate over the crimson, while the voltine characteristics are, as in other egg-character- istics before enumerated, maternal in inheritance. B. F, Eggs. The worms and moths derived from the reciprocal normal-coloured Fi eggs in the summer of 1910 were all dark-eyed, and the moths paired inte7' se gave the following F^ eggs : Matings Number of Matings Eggs laid by each moth Normal Crimson Totals 6. 1 342 136 478 6.2 294 103 397 6.8 369 103 472 6.9 374 103 477 ? Tetravoltine " Onodahime " < 6. 10 314 124 438 »> 3 306 141 447 >» )> 14 367 104 471 Totals 1077 Grand totals 2152 Expectation 2156J 357 1434 723 2875 718| 2875 These figures suggest to us at once that the crimson-coloured characteristic is a Mendelian one, recessive to the normal-coloured as is the former crimson breed. We reared these eggs separately in the spring of 1910. Crimson-coloured eggs from two batches of the second breed gave, without any exception, crimson-eyed worms and moths, while normal-coloured gave all normal-eyed worms and moths. Both K. TOYAMA 387 crimson and normal-eyed moths derived from the mixed batches of the above breeds when paired inter se oviposited the following batches: ^= normal-eyed; 12 = crimson-eyed moths. Parent Moths Eggt Uid Names of breeds ? ranging from 40*8 7o ^o 98 7o io each batch. In 42 batches of eggs derived from the same breed, we have counted 15,194 eggs, of which 12,796 eggs were found dead in the spring; i.e. a death-rate of 84 7o" In the eggs laid by a cross-bred crimson form (a cross between an extracted crimson from the cross, " Divoltine crimson x Divoltine Chiyodzuru," and the crimson derived from the divoltine " Tamanashi ") a considerably smaller percentage of dead eggs was noted than in the cross between divoltine crimson and tetravoltine white. K. TOVAMA TABLE VII. Number of batches Total number of eggs laid Number of Death-rate 1 481 52 10-8 °/o 8 487 45 9-2 «/„ 8 481 52 10-8 »/o 4 466 3 0-6 °/„ 5 471 260 66'2 % 6 476 241 50-6 °/„ 7 514 90 17-6% 8 543 169 311 °/o 9 515 342 66-9 °/o 10 490 16 3-2 °/o 11 529 377 71-2 7o 12 471 45 9-5 °/o 13 459 223 48-5 °/„ 14 554 3 0-5% 15 477 224 -47-5 0/, 389 Totals 7414 2142 28-8 % In this crimson form the average death-rate is only 288 % ; in certain matings nearly all the eggs were hatched, while in some others the death-rate mounted as high as 712 °j^. In other crimson batches laid by an extracted crimson form derived from a cross between the divoltine crimson and "jE'III" breed (an extracted breed from a cross between the wild mulberry silk-worm and tetravoltine " Tobuhime "), the figures quoted below were given : TABLE VIII. Number of batches Total number of eggs laid Number of dead eggs Death-rate 1 234 181 77-3 % 2 383 348 90-8 °/o 3 335 102 30-4 % 4 467 329 70-4 °/o 5 360 66 18-3 °/o 6 248 192 79-0 % 7 326 303 92-9 % 8 261 233 89-2 % 9 362 199 54-9 »/o 10 277 153 66-2 "/o 11 394 283 71-8 °/o 12 313 279 89-1 °/o 13 340 322 94-7 °/o 14 391 341 87-2 o/o 15 374 355 94-»»/o Totals 5065 3686 72°/o 390 Maternal Inheritance and Meiidelism In this form the death-rate is 72%, varying, in individual cases, from 18-3% to 94-9%. We have observed moreover that the death of the embryo inside the egg took place in its earlier stage. So we often found some dead eggs in August or September when a white patch appeared in one side of the egg (Fig. 8) which gradually became enlarged in size and at last the clear crimson colour became paler, while others were found dead after the embryo had completely developed. The majority of the deaths seem to occur, however, in the earlier developmental stage of the embryo. Even in the same breed, the death-rate differed greatly according to the colour of the esrg, and the voltine characters. Hence the divoltine white eggs laid by the very same breed which laid the eggs recorded in Table VI yielded a much smaller number of dead eggs than in the univoltine coloured eggs. Ten batches of eggs selected at random from fifty divoltine white egg-batches gave the following figures : TABLE IX. Number of batches Total number of eggs laid Number of dead eggs Death-rate 7 191 60 31-4% 8 361 18 4-9 % 9 111 51 45-9 °/o 10 329 168 51°/o 11 272 196 72% 12 309 12 3-8 % 13 186 18 9-6 °/o 13a 272 164 61-4% 14 413 64 15-5 % 15 381 78 20-8 % Totals 2825 829 29-8 7o that is to say, the average death-rate is 29'3°/q, which varies from 3'8 7o to 72% ii^ individual batches. The average death-rate found in fifty batches of eggs laid by the same breed is 19 7c' while in the case of univoltine eggs it is 84 7o> ^s already recorded in Table YI. In certain cases it occurred that some crimson-coloured eggs hatched in the summer. The death-rate of such divoltine crimson-coloured eggs was nearly the same as that of the white divoltine eggs. Even more striking facts were observed when we examined the number of dead eggs found in the batches in which three different coloured eggs, normal, crimson, and albino, were found. These are the F^ eggs laid by the cross, " H^' albino and the di voltine crimson breed. K. TOVAMA :v.)i TABLE X. Number of tmUhtm 1 2 3 4 5 Xonnal eggs ToUl number Deed (tf eggs eggs DeaUi-rmte Crimeonegga Albino eggc 194 209 195 213 213 33 64 23 23 79 17°/o 30-5 % 13-8 % 10-7 °/o 37% Total 95 66 99 75 94 Dead 77 49 76 61 62 Deftth-nte 81°/o 74-2 °/^ 76-7 % 81-3 % 64-8 °/o ToUl 91 86 88 99 87 Deed 12 7 80 22 5 Death-rmU 13-1 °/o 8-1 % 34°/o 22-2 »/o 5-7 °/o Totals 1024 222 21-6 % 429 325 757 % 451 76 16-8% In spite of their being laid by the same parents, the mortality in those eggs whose colour was crimson was much greater than the others. While in normal and white-coloured eggs, the percentage of deaths was only 21'6 7o ^^^ 168 % respectively, that of the crimson eggs was 75*7 7o' Other ten similar batches gave the following figures: Normal eggs Crimson efsgs Albino eggs Total namber 2315 956 1076 Nnmber of dead eggs 406 650 288 Death-rate 17-5 % 67-9 °/o 26-7 % The facts above enumerated taught us that in every case in every breed which we have studied, those eggs coloured crimson have a greater death-rate than normal-coloured eggs, while divoltine eggs which are crimson in colour did not shew so high a death-rate as univoltine crimson-coloured eggs. These facts led us to conclude that the embryo of crimson-coloured eggs is not so long-lived as that in the normal- coloured eggs. As to the cause of the early death of the embryo of the crimson- coloured eggs, we are quite ignorant at present, but we are inclined to believe that it may be due to the lack of certain pigments in the serosa which in some way help the respiration of the embryo during its development. V. General Considerations. From the results of these series of experiments in both line and cross breedings above quoted, it now becomes clear that (1) those egg- characteristics above enumerated, except in the crimson breed, are determined by the characteristics of the female parent, on account of which the paternal characteristics even when dominant are almost negligible in their influence upon the character of the egg, that is to say, phenomena of inheritance are maternal ; (2) gametic segrega- tion of parental characteristics takes place as in normal Meudelian 392 Maternal Inheritance and Mendelism allelomorphism ; and (3) in certain generations, both parental character- istics even when inbred give rise to the antagonistic characteristics which at first suggests that there is a departure from the normal rules. Suppose, now, there are certain Mendelian characteristics which behave as maternal in ioheritance. If they were reciprocally mated, what would be the result as regards their offspring ? Let D represent a dominant and R a recessive factor, the results of their reciprocal matings would be diametrically opposite. In the case of a i) female mated with an R male, the resulting F^ eggs would be all D, while an R female mated with a D male would give all R F^ eggs, in spite of their zygotic constitutions being the same in both matings, namely, DR. And therefore, all the worms and moths derived from the F^D ov R eggs will have the constitution DR^ in which the D behaves as an active factor in determining their characteristics. In the same way the egg- cell which has the composition of DR during its development in the parent body is influenced by the D factor only, and consequently after segregation, when it lost the antagonistic factor and became pure D or R, it retains the D characteristic before acquired. Thus the results of fertilization will be the production of all normal-coloured ^2 batches. Zygotioally considered, however, the F^. D eggs are not the same in their constitution. As the result of the fertilization, some of them will be DD, some DR and the rest RR. Consequently, the constitution of the F^ moths derived from the F^ D eggs will be a mixture of DD, DR and RR» Thus all the F^ eggs, whether fertilized with D ox R sperma- tozoon, will be all D characterized. As there is no means in this case of distinguishing a DD worm or moth from a DR or an RR, random matings between them are expected to occur. The result will be as below : Colour of the Zygotic composition Fi eggs laid of the egg a. ? DD X : 3-K or 2D : IR. As in the case of the F^, the zygotic constitution of F3 D and R batches is not simple D or R. As the formulae quoted just above shew, the constitution of certain F^ D batches is DD (series 1 a), some batches DR (series 1 c), some a mixture of DD and DR (series 1 6 and series 2 a), or DR and RR (series 2 c), and the rest DD, DR and RR (series 2 b). We get similar results in the case of the F3 R eggs, some batches being DR (series 3 a), some (series 3 c) RR, and the rest (series 3 6) a mixture of DR and RR. If moths derived from F3 D or R batches were inbred, what will be the result in the dominant series ? In F3 D batches, as we have already observed, there are five different kinds of batches whose zygotic compositions are respectively : (1) DD, (2) (DR + DD), (3) DR^ (4) {DD + DR + RR), (5) (DR + RR). The moths derived from each kind paired inter se will produce the following Ft batches : Mating Zygotic composition Outward appear- ance of F^ eggs 1. DD inter se = DD x DD = DD D 1. 9DRx d DR = (DD + DR + RR) D 2. {DR + DD) inter se = 2. 3. ?DBx s DD = (DD + DR) 9DD X s DD = DD D D 4. 'iDDx i DR = {DD + DR) D 3. DR inter se = 2DRx d DR=:{DD + DR + RR) D rl. ^DD X g DD=DD D 2. ^DDx 6 DR = (DD-rDR) D 3. 9DD X e RR = DR D 4. ?Z)B X seeds ... Indent Indent Fs seeds ... (Indent + Round) (Indent + Round) In certain cases, however, strict segregation took place or an inter- mediate form was produced in F2. It was, moreover, mentioned that in certain varieties those characteristics behaved quite normally in in- heritance, namely, F^ seeds are all indents which in F^ segregate into three indents to one round. Hence we are led to say that certain characteristics of the eggs of animals and of the seeds of plants behave in inheritance in a similar way. We shall discuss in our next paper the appearance of intermediate or mixed batches in certain line- or cross- breedings, a fact which seems to be inconsistent with maternal inheritance. We are at present collating the facts gathered from certain experiments which we have just concluded, and from the trend of our results up to the present, we 400 Maternal Inheritance and Mendelism will, we think, be able to put forward a satisfactory explanation of this phenomenon. We merely, at present, say that this appearance is not really contradictory to the maternal inheritance. Causes of Maternal Inheritance. Concerning certain characteristics such as the whitish grey, spindle- shaped, or yellow and white colour of newly laid eggs whose origin is due to the shell or yolk which are entirely derived from the maternal body, the maternal inheritance is the natural consequence, and may be compared with the inheritance of certain characteristics of the seed-coat of plants which are of purely maternal origin. Concerning the colour of the egg, whose origin is due to the special pigments deposited in the serosa, the case is quite different. As the serosa is formed of cells derived from the conjugation of paternal and maternal nuclei, the egg-colour ought to be influenced by the paternal characteristics if they are dominant, but, as we see, it is entirely maternal in certain colours, such as the reddish brown, blue, normal colour, etc. Nothing is as yet known as to why these serosa characteristics behaved as maternal in inheritance. We are now waiting the results of the further series of experiments which we have been engaged upon concerning this question. There are other characteristics of the silk-worm which behave maternally in inheritance. They are the brood characters such as uni-, di-, or multivoltine, or "voltinism" of the silk-worm. The fact of maternal inheritance of these characteristics was first observed by me (1906) and was proved by McCracken (1909), who was led to the conclusion that the order of inheritance was non-Mendelian, while Castle (1910), upholding the fact that they are maternal in inheritance, says that univoltinism is a Mendelian dominant to divoltinism. Voltinism. McCracken's results may be compared with those obtained by us in the series of experiments above referred to, but in the case of voltinism, as there are many causes disturbing the proper elimination of parental characteristics which are entirely neglected by her, it is rather premature to consider the phenomena of inheritance displayed by the character " voltinism " as non-Mendelian. We enumerate here those disturbing causes : (1) divoltine character may easily be changed by the influence K. TOYAMA 401 of temperature during the incubation of the egg, more strictly, during the embryonal stage after sexual cells are liberated from the mesodermal tissue. If we expose eggs at this stage to a temperature of about 60 — 65° F. or lower until hatching, all the moths derived from them will lay divoltine whitish eggs; on the contrary, if we subject them to a temperature of 80° F. or more, all the eggs will become univoltine coloured ones. This is a well-known fact among Japanese breeders and has been made use of for industrial purposes for the last twenty years. (2) The eggs laid by the second brood of the divoltine breed are identical in appearance with the univoltine eggs and hibernate without hatching. In the case of crossing, we are, therefore, unable to eliminate divoltine characterized eggs from the univoltine in every alternate generation. (3) The maternal inheritance referred to above, which also prevents the proper elimination of antagonistic characters. These are the causes why, I think, the character " voltinism " behaved so irregularly that McCracken considered it to be non-Mendelian. Generally speaking, I believe, the order of inheritance of the " voltinism " of the silk- worm will follow the course before mentioned in our scheme. In a later paper I propose to give a fuller account of the phenomena connected with voltinism. Before concluding this paper, I wish to express my sincere thanks to Prof. W. Bateson who has kindly assisted me in many ways when pre- paring this paper for press. Thanks are also due to Mr S. Hashimoto, assistant in our laboratory, who has helped me in rearing the worms used in our experiments since 1906. VI. Summary. 1. In the egg of the silk-worm there are certain special character- istics of shape, colour, etc. which differ in different breeds or even in the same breed. Japanese green breeds generally lay green-shaded eggs varjdng in depth of colour, often mixed with normal coloured ones. Most of the Chinese and European breeds lay similar green-shaded eggs, both of them, however, being distinguishable from each other by special lustres and shades. Eggs of the Japanese normal breeds are, however, brownish slate shaded with some light pink or purple. Among the eggs laid by Japanese normal-egged breeds we often find many variants in shape and colour, a smaller number of the variant being sometimes found in a batch, frequently in Mendelian proportions. 402 Maternal hiheritance and Mendelism while in other cases the whole of a batch will be found to consist of a variant. 2. Breeding experiments were made on the following egg-character- istics : 1. Greenish-shaded TheopMla colour (Fig. 10). 2. Various green colours of Japanese green and some Chinese or European breeds (Fig. 6). 3. The reddish brown variant derived from the normal-egged breed (Fig. 2). 4. The blue variant from the normal breed (Fig. 5). 5. The whitish grey variant from the normal breed (Fig. 4). 6. The spindle-shaped variant from the normal breed (Fig. 13). 7. The crimson variant from the normal breed (Figs. 7 and 8). 8. Yellow, brownish-shaded yellow and white colours of newly laid eggs of the yellow and white cocoon breeds (Fig. 9). 9. The brownish slate colour of Japanese normal breeds (Figs. 1, 3, and 11). 3. Some of these characteristics such as the normal brownish-slate, reddish brown, blue, crimson, etc., arose from the special pigments deposited in the serosa, which is produced by the conjugation of parental nuclei, while others, such as whitish grey, spindle-shaped or the colour of newly laid eggs are due to the shell or yolk which are of purely maternal origin. The colour of greenish-shaded eggs, such as are found in Japanese green breeds, and some Chinese or European breeds, is mostly derived from the serosa, but it is more or less influenced by the colour of the shell which is slightly tinted with green, or some other colours. 4. Those characteristics, in spite of the fact that their origin is different, behave in the same way as regards inheritance, except the crimson-coloured variant, which Mendelize in the normal order. The order of inheritance is represented in the following schemes (page 403). The order of inheritance represented by the first scheme seems to be non-Mendelian, but really it is Mendelian, the cause of the dis- turbance of the proper order being due to the fact of maternal inherit- ance, in which paternal characteristics remain dormant, even dominant ones, in the egg stage. K. TOYAMA 403 (1) The case where characteriHticH behaved maternally in inh(>ritanc<>. Huch ax normal, reddisb brown, grey, etc. D = dominant ; i? = recessive cbaracteri sties. (1) (2) (?Dx,jii) (9 R X d D) i, i i, J, Fi eggs F-i eggs F3 eggs Ft eggs Fi eggs Fs eggs F7 eggs I i I {D + R) D {D + R) D J. I Constant 1. (I> + J?) D (D ^ R) R R R R I Constant (2) The case where characteristics behaved in normal Mendeliau order, snch as the crimson colour. (1) (2) iD X gR 9R X iD Fi eggs 1 D Fi eggs (3D 1 : ] — 1 1 IB) 1 «) ^3 eggs 1 D (32) ? In the former scheme, the D represents dominant coloured batches, the R recessive and (D + R) & mixture of both D and R batches, while in the latter the D and R represent a single batch of D and R eggs and the (3D : IR) a mixed batch consisting of D and R in the proportion of 3 : 1. 404 Maternal Inheritance and Mendelism 5. Thehereditaryrelationsofthosecharacteristics above enumerated: as to the shape, the whitish grey is dominant towards the Japanese normal which is in turn dominant to the spindle-shaped characteristic. As regards the colour, the greenish slate stands first in dominancy, next comes the Japanese normal brownish slate ; hypostatic to it comes perhaps the blue and then reddish brown and lastly the crimson. In the colours of newly laid eggs, the white is hypostatic to yellow or brownish yellow. The relation between the normal and the crimson is the ordinary Mendelian one, the former being epistatic to the latter. 6. In the crimson-coloured eggs extracted from various breeds or crosses, the death-rate is always much greater than that of eggs from the normal, or albino breeds. Even when those three kinds of eggs are laid by the same parent, the same is the case. In divoltine crimson breeds, the divoltine white eggs laid by the spring or first brood are much healthier than those crimson-coloured eggs laid by the summer or second brood. 7. The phenomena of inheritance observed in the eggs of the silk- worm may be well compared with those observed in certain seed- characteristics of plants, such as maize, peas, wheat, etc. EXPLANATION OF PLATE XX. Normal-coloured egg of divoltine " Shinkawachi." Reddish-brown variant from divoltine "Shinkawachi." Normal-coloured egg of the original breed of the whitish-grey variant. Whitish-grey variant. Blue variant derived from divoltine " Kuni-nishiki." Green-shaded egg from the Japanese green breed. Crimson-coloured variant. Dead egg of the crimson variant. Newly laid egg of tetravoltine white. Newly laid egg of tetravoltine yellow. Egg of Theophila mandarina. Fig. 10a. Matured egg. Fig. lOh. Newly laid egg. Fig. 11. jFj eggs between female divoltine " Shinkawachi " and male Theophila mandarina. Fig. 12. F^ eggs of the above mating. Fig. 13. Batch of spindle-shaped eggs laid by a moth. Every figure except No. 13 is very much magnified, actual size being about 1-15 mm. in length and 0*95 mm. in breadth. Fig. 1. Fig. 2. Fig. 3. Fig. 4. Fig. 5. Fig. 6. Fig. 7. Fig. 8. Fig. 9a. Fig. 9b. Fig. 10. JOURNAL OF GENETICS. VOL II N° 4 PLATE XX ^% l.< 12 K. TOYAMA 405 LITERATURE CITED. 1. Bateson, W. MendeVs Principles of Heredity, 1909. 2. Baur, E. Einfuhrung in die experimentelle Vererhungdehre, \Si\\. 3. BiFFEN, R. H. " Mendel's Law of Inheritance and the Wheat Breeding.'* Joum. Agric. Sci. Vol. L No. 1, 1905. 4. Castle, W. E. " The Effect of Selection ujwn Mendeliau Characters manifested in one Sex only." Joum. of Exp. Zool. Vol. viii. No. 2, 1910. 5. CoRRENS, C. " Ba.starde zwischen Maisrassen." Bibliotheca Botanica, 1901. 6. Lock, H. "Studies in Plant Breeding in the Tropics." II and III. Ann. Roy. Bot. Gard., Peradeniya, Vol. in. 1906. 7. MX'rackex, J. " Heredity of the Race- characters Univoltinism and Bivol tiuism in the Silk-Worm. A case of non-Mendelian Inheritance." Journ. of Exp. Zoology, Vol. ll. No. 4, 1909. 8. Totama, K. " The Hybridology of Insects. 1. On some Silk- Worm Crosses, with special reference to Mendel's Law of Heredity." Bvlletin of the College of Agriculture, Tokyo Imperial University, Vol. vii. 1906. 9. . " On the Polygamous Habit of the Silk-W^orm." Ibid. 1906. 10. . " A Sport of the Silk- Worm, and its Heredity Behaviour." iijrf. 1909. 11. . "On Varying Dominance of certain White Breeds of the Silk- Worm." Zt. f. indiikt. Abst.- und Vererbxingslehre, Bd. vii. Nr. 3, 4, 1912. 12. TscHERMAK, E. " Uber die gesetzmiissige Gestaltungsweise der Mischlinge," Zt. f. d. landic. VerswJiw. in Osterr. 1902. CAHBRIOaE : PRINTED BY JOHN CLAY, M.A. AT THE UlfrrEBSlTT PRESS QH Journal of genetics A1J6^ V.2 cop. 2 Biolo{(ical & Medical Serials PLEASE DO NOT REMOVE CARDS OR SLIPS FROM THIS POCKET UNIVERSITY OF TORONTO UBRARY r-