232 PROCEEDINGS OP THE AMERICAN ACADEMY. 



matter is now undergoing experimental tests which, when complete, may 

 enable us to make predictions fjr these color categories not less jjrecise 

 than those given for white. 



In the last column of the table are given for comparison the predic- 

 tions based on Galton's law for the corresponding generations and pairs. 

 If we were to consider the grand totals only, we might conclude that 

 the Galtonian predictions are quite as good as the Meudelian, but if we 

 examine item by item the two series from which these totals are made 

 up, we see that there is no comparison in point of accuracy between the 

 two sets of i^redictions. The Mendelian predictions are very close to the 

 observed numbers throughout the table, generation by generation and 

 pair by pair. In all cases except four the predictions are either perfect 

 or within one of jDerfertion, and in one only of these cases is the error 

 greater than two. Th : s one case is the total for generation III where 

 the observed number is fourteen, the expectation eleven. Rarely do the 

 Galtonian predictions come within one, or anywhere near one, of pre- 

 fection. They demand the occurrence of white individuals in every gen- 

 eration and among the offspring of nearly every pair in the series, 

 whereas white individuals are entirely wanting, and according to Men- 

 del's law are not to be expected, among the offspring of all pairs in 

 the second generation, and of eight other pairs in later generations of the 

 series. The test is conclusive in favor of Mendel's law and against the 

 '' law of ancestral heredity," in the special case of albinism in mice. 

 Elsewhere Castle and Allen ( :03) have shown that among organisms 

 in general albinism probably follows the same (Mendelian) law of 

 inheritance. 



Numerous other cases of Mendelian inheritance covering a wide 

 range of characters are recorded in recent papers by de Vries (:01-03), 

 Correns (:01, :03), Tschermak (:01, :01", :02), Bateson and Saunders 

 (:02), Webber (:00), Spillman (:02, :02 a ), Hurst (:02, :03), and others. 

 These cases show that the Mendelian laws are widely applicable. They 

 are not laws of hybridization merely, as Vernon (:03) and some others 

 assume, but are general laws of alternative inheritance. 



III. Yule on Galton's Law and Mendel's Law. 



Bateson (:02) l ias taken the very reasonable position that Mendel's 

 law and the law of " ancestral heredity " cannot both be applicable to the 

 same classes of cases. But Yule (:02) sees no incompatibility between 

 the two, and this view Pearson (:03) endorses. Yule says (p. 22C), 



