Despite the many disturbing factors with which the observations had to contend, it is nevertheless seen by this experiment that the development of the hybrids, with regard to those characters which concern the form of the plants, follows the same laws as does _Pisum_. With regard to the colour characters, it certainly appears difficult to perceive a substantial agreement. Apart from the fact that from the union of a white and a purple-red colouring a whole series of colours results, from purple to pale violet and white, the circumstance is a striking one that among thirty-one flowering plants only one received the recessive character of the white colour, while in _Pisum_ this occurs on the average in every fourth plant. Even these enigmatical results, however, might probably be explained by the law governing _Pisum_ if we might assume that the colour of the flowers and seeds of _Ph. multiflorus_ is a combination of two or more entirely independent colours, which individually act like any other constant character in the plant. If the flower colour A were a combination of the individual characters _A_{1} + _A_{2} + ... which produce the total impression of a purple colouration, then by fertilisation with the differentiating character, white colour, _a_, there would be produced the hybrid unions _A_{1}_a_ + _A_{2}_a_ + ... and so would it be with the corresponding colouring of the seed-coats[44]. According to the above assumption, each of these hybrid colour unions would be independent, and would consequently develop quite independently from the others. It is then easily seen that from the combination of the separate developmental series a perfect colour-series must result. If, for instance, _A_ = _A_{1} + _A_{2}, then the hybrids _A_{1}_a_ and _A_{2}_a_ form the developmental series-- _A_{1} + 2_A_{1}_a_ + _a_ _A_{2} + 2_A_{2}_a_ + _a_. [44] [It appears to me clear that this expression is incorrectly given, and the argument regarding compound characters is consequently not legitimately developed. The original compound character should be represented as _A_{1}_A_{2}_A_{3} ... which when fertilised by _a_{1} gives _A_{1}_A_{2}_A_{3} ... a as the hybrid of the first generation. Mendel practically tells us these were all alike, and there is nothing to suggest that they were diverse. When on self-fertilisation, they break up, they will produce the gametes he specifies; but they may also produce _A_{1}_A_{1} and _A_{2}_A_{2}, _A_{1}_A_{2}_a_, &c., thereby introducing terms of a nature different from any indicated by him. That this point is one of the highest significance, both practical and theoretical, is evident at once.] The members of this series can enter into nine different combinations, and each of these denotes another colour[45]-- 1 _A_{1}A_{2}_ 2 _A_{1}aA_{2}_ 1 _A_{2}a_ 2 _A_{1}A_{2}a_ 4 _A_{1}aA_{2}a_ 2 _A_{2}aa_ 1 _A_{1}a_ 2 _A_{1}aa_ 1 _aa_.