The extracted dominants show great variability in their progeny, but the extracted recessives show practically none. This is because a positive character may fail to develop; but an absent character can not develop even a little way. The difference in variability of the offspring of two extracted recessives and two extracted dominants is the best criterion by which they may be distinguished, or by which the presence (as opposed to the absence) of a factor may be determined. The crest of fowl receives especial attention as an example of a character previously regarded as simple but now known to comprise two and probably more factors--a factor for erectness, one for growth, and probably one or more that determine the restriction or extension of the crested area. The direction of lop of the single comb is an interesting example of a character that seems to be undetermined by heredity. In this it agrees with numerous right and left handed characters. It is not improbable that the character is determined by a _complex_ of causes, so that many independent factors are involved. A series of studies is presented on the inheritance of plumage color. It is shown that each type of bird has a gametic formula that is constant for the type and which can be used with success to predict the outcome of particular combinations. New combinations of color and "reversions" receive an easy explanation by the use of these factors. The cases of blue, spangled, and barred fowl are shown also to contain mottling or spangling factors. CHAPTER I. THE SPLIT OR Y COMB. A. INTERPRETATION OF THE Y COMB. When a bird with a single comb, which may be conveniently symbolized as I, is crossed with a bird with a "V" comb such as is seen in the Polish race, and may be symbolized as oo, the product is a split or Y comb. This Y comb is a _new form_. As we do not expect new forms to appear in hybridization, the question arises, How is this Y comb to be interpreted? Three interpretations seem possible. According to one, the antagonistic characters (allelomorphs) are I comb and oo comb, and in the product neither is recessive, but both dominant. The result is a case of particulate inheritance--the single comb being inherited anteriorly and the oo comb posteriorly. On this interpretation the result is not at all Mendelian. According to the second interpretation the hereditary units are not what appear on the surface, but each type of comb contains two factors, of which (in each case) one is positive and the other negative. In the case of the I comb the factors are presence of median element and absence of lateral or paired element; and in the case of the oo comb the factors are absence of median element and presence of lateral element. On this hypothesis the two positive factors are dominant and the two negative factors are recessive.