286 
ZOOLOGY: R. W. HEGNER 
by vegetative reproduction. A specimen, numbered 58 (fig. 1, A) was chosen 
for this work because it was near the mean of the species in diameter and spine 
number, and multiplied rapidly. Figure 2 shows the principal results of the 
experiments. During the thirty-nine days before selection was begun 198 
specimens were obtained from number 58, representing 7 generations. These 
varied in spine number from 8 to 13, with a mean of 10.87. Selection was then 
inaugurated and carried on for six periods totalling sixty-four days. During 
this time 1192 specimens were reared, belonging to 22 generations. The 
work was divided into periods so that any changes due to environmental con- 
ditions would be revealed, and also because one investigator can take care of 
only a few hundred specimens at one time, and when the limit has been reached 
a new selection has to be made. At first all parents and progeny were kept 
until the end of each selection period, but later selection was also practiced 
during the periods. An effort was made to obtain a line (A) with a high mean 
spine number and another line (E) with a low mean spine number. Specimens 
within the high line that possessed 12 or more spines were selected and those 
within the low line with 10 or less. Past performance was used as the basis of 
selection, i.e., specimens in the high line that had a high spine number and had 
near relatives with a high spine number were chosen, and similarly specimens in 
the low line that had a low spine number and had near relatives with a low 
spine number were chosen for continuing the lines. During the six selections 
periods the differences between the mean spine numbers of the two lines were 
as follows: —0.07, 0.50, 0.40, 0.48, 0.84, and 1.16, and the mean difference was 
0.55. The coefficients of correlation between parents and progeny with 
respect to spine number during these periods were 0.060 ± 0.076, 0.220 ± 
0.039, 0.186 ±0.042, 0.185 ±0.040, 0.403 ± 0.044, and 0.512 ±0.039. 
Selection was then stopped and during four periods totalling thirty-five 
days, no selection was practiced. As many specimens as could be taken care 
of were kept during this time and as soon as one divided the 'parent' was 
eliminated and the offspring kept. In this way 1325 specimens were obtained 
belonging to 18 generations. At first regression occurred in both lines but 
later the mean difference between them remained almost constant. These 
differences for the four periods were 0.94, 0.07, 0.41, and 0.43, and the mean 
difference was 0.46. The decrease in the difference after selection was stopped 
was probably due to the production and inclusion within the high line of low 
spined specimens that would have been eliminated during the selection periods 
and within the low line of high spined specimens that likewise would have been 
eliminated. Each line, however, should give rise to as many high as low 
spined specimens, and hence the means of each would not vary after an equi- 
librium had been reached and the difference between them would be permanent. 
The coefficients of correlation between parents and progeny within the high 
line during these four periods was 0.170 ± 0.027 and within the low line 
0.197 ±0.024. 
The low line (E) was then subjected to selection in a similar way and a high 
