222 DONALD WALTON DAVIS 



progress of regeneration of the stripes there is a tendency further 

 to increase temporarily the proportions of instances of odd 

 numbers. 



Comparison of the data for the separate lots is rather sug- 

 gestive. Lots 1 and 2, collected July 12 and 18, respectively, 

 differ in that the earlier lot includes a shghtly larger proportion 

 of specimens with a very low number of stripes indicating re- 

 peated, rapidly succeeding divisions, while the later set shows a 

 few more individuals with high numbers of stripes indicating 

 more nearly completed regenerations. Lot 3, representing a 

 group collected September 22 from the same place as lot 2, 

 shows a relatively much greater number of individuals with 

 twelve or more stripes. In this group the mode is at twelve 

 with prominent secondary modes at fourteen and sixteen and 

 only lower modes at seven and five. This indicates progress in 

 regeneration with less frequent divisions since the collection in 

 July. Lot X represents a group of specimens selected for their 

 large size. The distribution as shown in the table suggests that 

 the group is composed of specimens that have, for the most part, 

 completely regenerated. Examinations of my records, which 

 show for each individual the number of stripes in areas of all 

 different ages, confirms this suggestion. Davenport's curve ap- 

 parently represents a group of individuals of which a very large 

 proportion have recently divided. It includes a few specimens 

 with more than twelve orange stripes and a large number with 

 twelve, but the great bulk of individuals show fewer than ten. 

 Davenport gives no indication of the time at which these speci- 

 mens were collected. The large number of individuals with few 

 orange stripes, in connection with my curves, suggests the prob- 

 abiUty that they were collected early in the summer. Among 

 the groups below ten Davenport's curve in contrast with mine 

 shows an excess of specimens with even numbers of stripes. A 

 possible partial explanation of this lies in the readiness (p. 167) 

 with which the line of fusion of cut edges soon after division may 

 be mistaken for an orange stripe. One stripe thus added in a 

 certain proportion of cases would alter the relative numbers of 

 specimens with odd and even numbers of stripes from the state 



