126 The Mechanism of Evolution in Leptinotaesa 



that of the group collectively and not as so many individual agents, and as a 

 result there are three groups that need be considered, namely, the mass of the 

 gametic system, symbolized by the elytral pattern group, which for brevity may 

 be represented by A for undecimlineata, a for signaticolUs, B for the pattern 

 condition in the larval stage of undecimlineata, b for the same in signaticollis, 

 and C for the white lipoid-color group of undecimlineata, c for the yellow of 

 signaticollis. In this cross other agents are neutral and do not act in the opera- 

 tion, so that the composition of signaticollis is abc, undecimlineata ABC. In 

 all respects this series can be and is, in reality, as far as this cross is concerned, 

 a typical trihybrid reaction, as typical and constant in its reaction as the classical 

 case of Mendel's trihybrid peas, with this important exception. Mendel's peas 

 had the three characters present in the same stage of the individual; in this 

 series the three contrasting characters are not present at the same time in the 

 ontogeny of the individuals, but at successive periods in the life of the indi- 

 vidual, and none is in any way dependent upon those antecedent for its mani- 

 festation, so that this series is of interest as showing that the condition of 

 trihybrid or any other is not limited to the manifestation at one stage, but that 

 the condition governs the entire life-cycle. 



In plate 13 the results shown are of the cross of these two species when the Ac 

 determiners are the same in value and th-e conditions of the medium are uniform 

 and neutral. The result in Pg is that, in the larvae in the second stage when the 

 first contrasting characters come in, there is a sharp separation of the population 

 into two classes, white and yellow, in the proportions of 3 white, 1 yellow, the 

 array appearing in all instances that of a monohybrid type. In the third larval 

 stage each of the two classes again separate, upon the basis of the pattern system 

 present in this state, into two groups that have the general relations of 3 without 

 pattern, 1 with pattern ; but in the entire array in the third larval stage there 

 are present four classes of larvae: 9 white without spots, 3 white with spots, 

 1 yellow with spots, 3 yellow without spots, showing the " dominance " of the 

 white over yellow and of no spots over spots, or undecimlineata pattern over 

 signaticollis. In the adult each of the four larval groups gives three kinds of 

 adults : extracted signaticollis type, undecimlineata type, and mid-types. These 

 adults are always pure for the adult characters in the extracted types, but may 

 be hybrid for one or two of the larval characters, or pure in all. The mid-types 

 are never pure for the adult characters, but may be pure for the larval, or hybrid 

 in 1 or 2. 



That there might be no mistake, I have tested out the different products in F^ 

 as a routine laboratory operation, and have found that all of the expected com- 

 binations were present, giving the actual kinds of individuals that Mendel's 

 hypothesis demands. I have found 27 different gametic conditions in the Fg 

 array and no more, and these were in entire agreement with Mendel's hypothesis 

 as to their composition and their products in F3. The combinations have been 

 found and tested to the extent as shown in table 14. 



This series has been a fortunate one, in that the condition of the larval char- 

 acters is shown in the ontogeny, so that one is able with fewer matings to make 

 an analysis of the F2 array than if the three characters were all present in the 

 same stage, and the fact that when A and a are present in the gamete the product 

 is always intermediate, separates at once, in adult characters, homozygous from 

 those that are not. With experience I learned to separate in Fg nearly all of 



