K. Pearson 
333 
of local races in other cases, and ultimately, perhaps, the differentiation of species. 
But the safe way to reach the latter is through the problem of local races. 
If Mr Bateson wishes to attack the problem of evolution by what he terms 
discontinuous variation, he must go far further than forming a useful catalogue of 
museum and collectors' deviations from " type." He must trace first whether in any 
given case they are or are not inherited, secondly he must discover whether or not 
the individuals who possess them are more fertile than the " type," thirdly whether 
the death-rate is with regard to them selective or non-selective. Shortly starting 
with a race having among its members a few with a recognisably discontinuous 
variation, he must show how its descendants at a later period have the discon- 
tinuous variation of the earlier period as a dominant character. In other words 
he must deal with the vital statistics of a population, or proceed biometrically. 
Mr Bateson writes (p. 202) as if I were to-day inclined to allow more to 
" discontinuous variation " than I did in 1895, and this although he cites on 
p. 204 a passage from my memoir on Homotyposis of 1901 practically identical in 
spirit with a second he has cited previously from my memoir of 1895. The only 
basis for his belief lies in the fact that I should heartily welcome any attempt to 
demonstrate by a satisfactory statistical investigation — none other is valid — that a 
significant change has occurred in any wild species in its natural environment by 
a " discontinuous variation " which, is sufficiently marked to be distinguishable 
from a continuous series of variations *. Till such a biometric investigation 
as I have suggested is made I must adhere to my statement as to the dis- 
tribution of variation, for it accords well with the populations I have myself 
examined and measured in the case of both animals and plants. These popu- 
lations may be far fewer than those upon which Mr Bateson bases his statements, 
but as far as I know he has not published large series of the frequencies of 
various organs in different populations f, which would enable me to test whether 
or not my " description accords ill with the observed facts of variation." It does 
not accord ill with the many series I have myself published or with the still 
more numerous data which I have still unpublished, and which have also influenced 
my judgment on this point. To sum up then, "discontinuous variation," which is 
sufficiently marked to be separable from continuous variation, is so infrequent 
(I do not say it does not occur) as to be statistically negligible for the purpose 
* Careful selection of slight variations appears to be effective in the case of artificial selection. See, 
for example. Sir W. T. Thiselton Dyer: The Cultural Evolution of Cyclamen lati/olium, "The striking 
results obtained by cultivators have been due to the patient accumulation by selection of gradual but 
continuous variation in any desired direction." R. S. Proc. Vol. 61, p. 147. 
t In Mr Bateson's Materials in hardly any case are statistics of the general population given. In three 
cases — those of the common earwig (p. 40), the Java beetle Xylotrupes gideon (p. 39), and the cockroach 
Blatta Americana (p. 417) — are statistics given for the general population of a locality. In none of these 
cases is evidence given as to the inheritance of the " discontinuous variation," and in one it is suggested 
that the variation is possibly due to regeneration. It would not I presume be difficult to test the 
question of inheritance by separating the dimorphic forms ; and one instance of death-rate correlated 
with such dimorphism in a population would in my opinion be worth a whole catalogue of "meristic 
variations." 
