ON ARTIFICIAL ATAVISM. 23 



taken as such at the outset are really the true units, or at least the true 

 principal units of the combination. 



The result of this last crossing experiment may be regarded as a case of 

 atavism and as the type of a long series of instances of atavism caused by 

 crossing. To show this more clearly we must suppose that the original dark- 

 red type had died out and was therefore unknown, or perhaps it may suffice 

 to assume that the relation of the garden varieties to the wild form was 

 doubtful. Not having tlie component characters of the red color, we would 

 only be justified in saying that the crossing of the two varieties in question 

 produced a new character not seen in any of them, but belonging to the wild 

 or supposed ancestor. Evidently many crossings of cultivated varieties must 

 in this way lead to cases of reversion. 



Briefly stated, the results described here for AntirrJiiiiiiui and controlled 

 by experiments with the other species, we may say : 



1. It is possible to split up the colors of some Aoivers by crossing the 

 colored type with the white variety. 



2. The constituents arrived at by this splitting often follow Mendel's lazvs. 



3. By crossing the appropriate constituents the original compound color 

 may be rebuilt. 



4. Instances of atavism may in this way be artificially produced. 



H. F. Roberts: Before raising a point suggested indirectly by this paper I would 

 like to ask whether in making one's first cross and obtaining one's first group of 

 hybrids, it may not be necessary (in view of the fact that the plant is a composite and 

 all its organs and the sporophyls included are variable and differ from each other) to 

 take all of one's pollen from a certain parent from one sporophyl? In other words, 

 supposing you were using as a male parent a flower which is didynamous, the stamens 

 of two lengths, is it not possible that the pollen from stamens of different lengths may 

 differ from each other in potency; and if so, would it not be necessary to preserve one's 

 pollen that is used in producing a certain group of hybrids from a single sporophyl in 

 order to be within the limits of the greatest possible accuracy? 



W. J. Spillman: I am not going to answer the question, but I do say that there 

 is only one place to get the answer, and that is in his laboratory. 



W. Bateson : The interest of the subject of Professor De Vries is perfectly apparent 

 to everybody. Two points present themselves to me. I am myself engaged in trying to 

 determine the constituents which break up a compound character into its component 

 parts, and I am endeavoring to work out the relation of those to each other. My first 

 comment is this: Say it is imagined that you can analyze a character into its component 

 characters; it is obviously easy to form the converse conception, namely, that you can 

 synthesize characters so as to rebuild a compound. But the question arises, how can it 

 be possible to rebuild a compound character which consists of more than two compo- 

 nents? According to the rules, the two opposite characters are contradictory to each 

 other, and no individual can carry more than two of them. Consequently, if you decom- 

 pose a compound character into more than three components, I do not see how by any 

 scheme that we could formulate you can rebuild even those organs which shall contain 

 those three characters. The second point is that I don't think Professor De Vries is 

 strictly accurate in describing such rebuilding as a synthesis at all. Your purple sweet 

 pea as you will breed it up by crossing is not a synthesis of the original sweet pea, 

 because it will not break up into the original components which the original sweet pea 

 contained, but it will only break up into the two components which you put into it. 

 The original, from which the original forms were produced, bred true, but the synthesis 

 will not. It is an apparent synthesis, but not a real one, and in my opinion the problem 

 of synthesis is one which remains wholly unexplained. 



O. F. Cook: The question of the synthesis of these characters has been raised, 

 and that of the identity or the dissimilarity of hybrids and mutations. I simply want to 



