INHERITANCE OF CREST. 11 



of Experiments 506, 601, and 617 (with one parent homozygous) we 

 have totals of 18 non-crested to 48 crested offspring, while theory 

 calls in this case for 16.50 and 49.50 respectively. The actual result 

 accords closely with expectation. 



When a heterozygous crested (DR) parent is mated with a non- 

 crested (R) one, 50 per cent of the offspring should be non-crested. 

 Subtracting from the totals of the right half of table 2 the offspring 

 of homozygous 79 (male, Experiment 704), we get 52 non-crested to 

 42 crested, instead of the expected 47 and 47. The considerable 

 deviation from expectation is doubtless due to insufficient numbers 

 and the extraordinary run of excess of non-crested offspring obtained 

 in Experiments 604, 606, and 613. 



3. BALDNESS. 



It is frequently stated (Kidd, teste Darwin, 1876, chap, vin ; 

 Russ, 1906, 127-128) that two crested birds should not be mated 

 together because they rarely produce fine crests, but, on the con- 

 trary, bald heads. One is advised to mate the crested bird with a 

 non-crested one whose parents were crested. Blakston (1880, 129), 

 while approving this rule of breeding, scouts the idea that the produce 

 of two crests tend to be bald. Such produce tend merely to have high 

 tufts instead of the desired flat crests. The experiments listed in 

 table 2 are of a sort to throw light on this question. Are the 61 crests 

 derived from the experiments recorded on the left hand of the table 

 different from those derived from Experiments Nos. 505, 515, 604, 

 704 to 708, and 712, recorded on the right hand of the table ? An 

 examination of the distribution of bald offspring does not show that 

 they are confined to or exceptionally prevalent in the cases of matings 

 of two crested individuals. It is necessary to seek another cause for 

 baldness. 



An inspection of the records shows : first, that of my original 

 crested birds some had a perfect (or fully feathered) crest and in 

 others the crest was imperfect (/. e., more or less bald on the occiput). 

 Birds with perfect crest were Nos. 9 (male), 11 (female), and 34 (male), 

 while birds with imperfect crest were Nos. 3 (female), 6 (female), 

 12 (female), 13 (male), and 21 (female). Now, with one exception (see 

 Experiment 714), all of the bald offspring obtained in these experi- 

 ments had one of the imperfectly crested birds in its ancestry. This 

 fact suggests the hypothesis that imperfect crest is a unit-character in 

 heredity. To test this hypothesis I have made various matings between 

 individuals with and those without an imperfect crest. The results of 

 such matings are given in table 3. 



