IN RATS AND GUINEA-PIGS. 25 



Our general conclusion from a study of these two series of guinea-pigs, includ- 

 ing 1,048 individuals, is that one can by selection either increase or decrease 

 the extent of the pigmented areas, but it is impossible by selection to fix this 

 pigmentation in a particular pattern, retaining pigment areas on certain parts 

 of the body and eliminating them from others. As the pigmentation changes 

 in extent, under the influence of selection, the various areas typically pigmented 

 are affected in the following order: Shoulder, side, rump, and head, the change 

 being greatest in the first-named and least in the last-named area, irrespective 

 of what particular spots were present in the selected ancestors. 



STATISTICAL ANALYSIS OF THE DATA FOR SERIES D AND H. 



The foregoing conclusions may, we think, fairly be drawn from an exami- 

 nation of the tables as they stand. For those, however, who place confidence 

 in the more precise methods of statistical analysis devised by Pearson and 

 others, it may be more satisfactory to treat the tables, which have been 

 constructed for the various groups of individuals, as correlation tables, and 

 derive from them the constants which measure the variability of parents 

 and children respectively in the several groups, and the degree of correlation 

 between the two. Such constants are given, in tables 26 and 27, for the 

 several groups of Series D and H, respectively. 



From the left-hand columns of table 26 we learn that, in Series D, the 

 number of spots borne by an individual increases from group to group, in 

 the case of both parents and offspring; but the number of spots is, in every 

 case, 20 to 26 per cent greater in the offspring than in their parents. This 

 indicates a tendency for the offspring to become pigmented in regions which 

 lacked pigment in their selected parents. 



Table 27 shows the existence of a similar, but still stronger, tendency in 

 Series H, the offspring bearing 50 to 105 per cent more spots than their 

 parents. This tendency may be considered regression, a tendency to return 

 to a condition of more widely distributed pigmentation, and to acquire 

 spots where the selected ancestors lacked them. It does not imply that the 

 young bear more pigment than their parents. The regression is stronger 

 the more reduced the pigmentation of the parents, as we might expect. 

 This is seen to be true both within Series H and in that series as a whole 

 compared with Series D. 



The standard deviation from the average number of spots is for the 

 offspring a very constant quantity, being close to 3.6 in both Series D and 

 Series H. This indicates no change in the variability in number of spots 

 as a result of selection, or as a consequence of change in the number of pig- 

 mented areas. The parents show throughout Series H and in Group D% of 

 Series D a less standard deviation than their offspring. This fact, however, 



