structure of a species must be essentially altered by external influences。 Two methods of experimental research may be adopted； the effect of crossing distinct species and； secondly； the effect of definite factors of the environment。
The subject of hybridisation is dealt with in another part of this essay。 It is enough to refer here to the most important fact； that as the result of combinations of characters of different species new and constant forms are produced。 Further； Tschermack； Bateson and others have demonstrated the possibility that hitherto unknown inheritable characters may be produced by hybridisation。
The other method of producing constant races by the influence of special external conditions has often been employed。 The sporeless races of Bacteria and Yeasts （Cf。 Detto； 〃Die Theorie der direkten Anpassung。。。〃； pages 98 et seq。； Jena； 1904； see also Lotsy； 〃Vorlesungen〃； II。 pages 636 et seq。； where other similar cases are described。） are well known； in which an internal alteration of the cells is induced by the influence of poison or higher temperature； so that the power of producing spores even under normal conditions appears to be lost。 A similar state of things is found in some races which under certain definite conditions lose their colour or their virulence。 Among the phanerogams the investigations of Schubler on cereals afford parallel cases； in which the influence of a northern climate produces individuals which ripen their seeds early； these seeds produce plants which seed early in southern countries。 Analogous results were obtained by Cieslar in his experiments； seeds of conifers from the Alps when planted in the plains produced plants of slow growth and small diameter。
All these observations are of considerable interest theoretically； they show that the action of environment certainly induces such internal changes； and that these are transmitted to the next generation。 But as regards the main question； whether constant races may be obtained by this means； the experiments cannot as yet supply a definite answer。 In phanerogams； the influence very soon dies out in succeeding generations； in the case of bacteria； in which it is only a question of the loss of a character it is relatively easy for this to reappear。 It is not impossible； that in all such cases there is a material hanging…on of certain internal conditions； in consequence of which the modification of the character persists for a time in the descendants； although the original external conditions are no longer present。
Thus a slow dying…out of the effect of a stimulus was seen in my experiments on Veronica chamaedrys。 （Klebs； 〃Kunstliche Metamorphosen〃； Stuttgart； 1906； page 132。） During the cultivation of an artificially modified inflorescence I obtained a race showing modifications in different directions； among which twisting was especially conspicuous。 This plant； however； does not behave as the twisted race of Dipsacus isolated by de Vries （de Vries； 〃Mutationstheorie〃； Vol。 II。 Leipzig； 1903； page 573。）； which produced each year a definite percentage of twisted individuals。 In the vegetative reproduction of this Veronica the torsion appeared in the first； also in the second and third year； but with diminishing intensity。 In spite of good cultivation this character has apparently now disappeared； it disappeared still more quickly in seedlings。 In another character of the same Veronica chamaedrys the influence of the environment was stronger。 The transformation of the inflorescences to foliage…shoots formed the starting…point； it occurred only under narrowly defined conditions； namely on cultivation as a cutting in moist air and on removal of all other leaf… buds。 In the majority （7/10） of the plants obtained from the transformed shoots； the modification appeared in the following year without any interference。 Of the three plants which were under observation several years the first lost the character in a short time； while the two others still retain it； after vegetative propagation； in varying degrees。 The same character occurs also in some of the seedlings； but anything approaching a constant race has not been produced。
Another means of producing new races has been attempted by Blaringhem。 （Blaringhem； 〃Mutation et Traumatisme〃； Paris； 1907。） On removing at an early stage the main shoots of different plants he observed various abnormalities in the newly formed basal shoots。 From the seeds of such plants he obtained races； a large percentage of which exhibited these abnormalities。 Starting from a male Maize plant with a fasciated inflorescence； on which a proportion of the flowers had become male； a new race was bred in which hermaphrodite flowers were frequently produced。 In the same way Blaringhem obtained； among other similar results； a race of barley with branched ears。 These races； however； behaved in essentials like those which have been demonstrated by de Vries to be inconstant； e。g。 Trifolium pratense quinquefolium and others。 The abnormality appears in a proportion of the individuals and only under very special conditions。 It must be remembered too that Blaringhem worked with old cultivated plants； which from the first had been disposed to split into a great variety of races。 It is possible； but difficult to prove； that injury contributed to this result。
A third method has been adopted by MacDougal （MacDougal； 〃Heredity and Origin of species〃； 〃Monist〃； 1906； 〃Report of department of botanical research〃； 〃Fifth Year…book of the Carnegie Institution of Washington〃； page 119； 1907。） who injected strong （10 percent） sugar solution or weak solutions of calcium nitrate and zinc sulphate into young carpels of different plants。 From the seeds of a plant of Raimannia odorata the carpels of which had been thus treated he obtained several plants distinguished from the parent…forms by the absence of hairs and by distinct forms of leaves。 Further examination showed that he had here to do with a new elementary species。 MacDougal also obtained a more or less distinct mutant of Oenothera biennis。 We cannot as yet form an opinion as to how far the effect is due to the wound or to the injection of fluid as such； or to its chemical properties。 This； however； is not so essential as to decide whether the mutant stands in any relation to the influence of external factors。 It is at any rate very important that this kind of investigation should be carried further。
If it could be shown that new and inherited races were obtained by MacDougal's method； it would be safe to conclude that the same end might be gained by altering the conditions of the food…stuff conducted to the sexual cells。 New races or elementary species； however； arise without wounding or injection。 This at once raises the much discussed question； how far garden…cultivation has led to the creation of new races？ Contrary to the opinion expressed by Darwin and others； de Vries （〃Mutationstheorie〃； Vol。 I。 pages 412 et seq。） tried to show that garden…races have been produced only from spontaneous types which occur in a wild state or from sub…races； which the breeder has accidentally discovered but not originated。 In a small number of cases only has de Vries adduced definite proof。 On the other side we have the work of Korschinsky （Korschinsky； 〃Heterogenesis und Evolution〃； 〃Flora〃； 1901。） which shows that whole series of garden…races have made their appearance only after years of cultivation。 In the majority of races we are entirely ignorant of their origin。
It is； however； a fact that if a plant is removed from natural conditions into cultivation； a well…marked variation occurs。 The well…known plant… breeder L。 de Vilmorin （L。 de Vilmorin； 〃Notices sur l'amelioration des plantes〃； Paris； 1886； page 36。）； speaking from his own experience； states that a plant is induced to 〃affoler；〃 that is to exhibit all possible variations from which the breeder may make a further selection only after cultivation for several generations。 The effect of cultivation was particularly striking in Veronica chamaedrys （Klebs； 〃Kunstliche Metamorphosen〃； Stuttgart； 1906； page 152。） which； in spite of its wide distribution in nature； varies very little。 After a few years of cultivation this 〃good〃 and constant species becomes highly variable。 The specimens on which the experiments were made were three modified inflorescence cuttings； the parent…plants of which certainly exhibited no striking abnormalities。 In a short time many hitherto latent potentialities became apparent； so that characters； never previously observed； or at least very rarely； were exhibited； such as scattered leaf… arrangement； torsion； terminal or branched inflorescences； the conversion of the inflorescence into foliage…shoots； every conceivable alteration in the colour of flowers； the assumption of a green colour by parts of the flowers； the proliferation of flowers。
All this points to some disturbance in the species resulting from methods of cultivation。 It has； however； not yet been possible to produce constant races with any one of these modified characters。 But variations appeared among the seedlings； some of which； e。g。 yellow variegation； were not inheritable； while others have proved constant。 This holds good； so far as we know at present； for a small rose…coloured form which is to be reckoned as a mutation。 Thus the prospect of producing new races by cultivation appears to be full of promise。
So long as the view is held that good nourishment； i。e。 a plentiful supply of water and salts； constitutes the essential characteristic of garden… cultivation； we can hardly conceive that new mutations can be thus produced。 But perhaps the view here put forward in regard to the production of form throws new light on this puzzling problem。
Good manuring is in the highest degree favourable to vegetative growth； but is in no way equally favourable to the formation of flowers。 The constantly repeated expression； good or favourable nourishment； is not only vague but misleading； because circumstances favourable to growth differ from those which promote reproduction； for the production of every form there are certain favourable conditions of nourishment； which may be defined for each speci