2 
Canary Breeding 
and comparing them with those referred to, it is necessary to indicate certain points 
in the planning of Davenport's experiments, which may possibly account for 
considerable difference in our conclusions. 
It would have been advisable for Davenport to have selected the original stock 
with much greater care. Most of the original birds used were of the Hartz 
mountain variety (page 8), which is bred purely for song quite regardless of colour 
and crest, the two points concerning which the author wished to test Mendel's 
theories, and also the practical rules of fanciers. 
"A few of the Norwich type" (page 8) were used and one is represented 
(Plate I, fig. 2) which no crest breeder would have in his room, and which shows 
that the "Norwich type" used, so far as established crest properties are concerned, 
was quite as poor as the nondescript Hartz crests. 
The author's sole criterion of a perfect crest (even though the requisite shape is 
given on page 8) seems to be absence of a bald occiput. Now this bald spot 
accompanies most of the perfect crests of the present day. See my figs. 3, 5. 
Plate IV, fig. 1. 
The fancier's idea of an imperfect crest is well depicted in 118 cT {loc. cit. Plate I, 
fig. 1) which is called perfect (page 12). This bird has practically no front to its 
crest — one of the worst faults — and is deficient in radiation over the eye. It might 
well be called an intergrade form, neither crest nor plainhead, and such birds are 
well known to appear occasionally in the best bird rooms. This fact is so well 
known that no bird with a crest showing this imperfection in the slightest degree, is 
used for crest-breeding purposes. I have frequently seen, and occasionally bred, 
these intergrade forms, varying from short fronts to complete absence of frontal 
crest, the posterior half only being present, and spreading from a line on the top of 
the head, instead of radiating evenly from a central spot. Thus it would appear 
that "crest" is not strictly "alternate" in inheritance. 
While the application of Mendelian principles to the inheritance of a tuft of 
feathers, or rudimentary crest, is interesting, it is unfortunate that the terms 
" perfect " and " imperfect " were used, and that the bald occiput was regarded as 
the sign of an imperfect crest (page 11). 
Similarly the disagreement with Russ with regard to " green x yellow " mating 
always producing " mottled " progeny (page 15), may be explained ; for Russ states 
that it is " important that two pure-bred birds be mated." Now, neither in the 
Hartz nor Norwich canaries had this been the case, for pure greens and pure 
"yellows" are not bred in these varieties (the only exception being green Norwich 
in very limited numbers). 
Meaning of Terms. Before I show that results vary according to the pedigree 
of the birds used in spite of markings, it is necessary to point out certain misunder- 
standings of terms. Among fanciers, the term " yellow " is synonymous with 
"jonque," and "buff" with "mealy." Yellow and buff are the terms usually 
employed, and "yellow" does not mean merely absence of dark feathering, as in 
