288 PHYSIOLOGIC GENETICS 



of variance. The F 2 generation in our experiment was reared by F x mothers whose 

 hybrid vigor created a quite different maternal environment. It would have been 

 desirable in this case to have fostered F 2 puppies upon mothers of the parental strain. 

 With the dog this is not always easy to accomplish, since cross fostering can be success- 

 fully carried out only between females in approximately the same stage of lactation. 



Another means of obtaining estimates of the effects of random and environmental 

 factors is comparison of successive litters from the same mating. Such litters will, 

 on the average, have similar heredity, although random variation will occur if the 

 parents are heterozygous. 



Another solution of the problem of equalizing or eliminating maternal effects is 

 hand rearing. Under such circumstances, of course, it is possible that the behavioral 

 patterns usual in the species will not appear, since the eliciting stimuli may be omitted 

 in the hand-rearing situation. Thus, the greater control of environmental variation 

 may detrimentally affect the appearance of the behavior of interest. Isolation of 

 subjects to insure equality of environmental conditions suffers from the same drawback. 

 It is important in studying the development of behavior to place the subject in an 

 expressive environment. 410 Such an environment is one designed to favor the elicita- 

 tion of species-specific behavioral patterns. It is conceivable that the technique of 

 semi-isolation 408 may prove to have some value in behavioral genetics. In this 

 method, animals are isolated at weaning and held in cages which permit minimum 

 stimulation. They are removed, however, at regular intervals and placed in a highly 

 stimulating environment. This method insures control of development while providing 

 essential conditions for psychologic development. 



The measurement of behavior. — As stated earlier, behavior may be defined as the acti- 

 vity of the intact organism. Behavior is complex and obviously can vary in more 

 than one dimension. Behavioral analysis of a character such as fighting in the mouse 

 begins with a set of descriptions of the various behavioral patterns involved. When 

 two mice are placed together, the measurements consist of the latency, frequency, 

 and intensity of the fighting pattern. Latency and frequency can be measured with a 

 stopwatch and counter. An enormous amount of behavioral work can be done with no 

 more elaborate instruments than an arena, timer, and counter. The intensity of 

 fighting behavior is more difficult to measure, as one cannot usually apply direct 

 measures of force. This is a technical and not a theoretical difficulty. Presumably 

 each blow and each bite involves the expenditure of a definite amount of energy, but the 

 application of a measuring instrument would interfere with the behavior. Rating scales 

 for intensity have been used, but these create problems of reliability and of equality 

 of intervals on the rating scale. 



Latency, frequency, and intensity of behavioral patterns are often not independent 

 measures. A pair of mice that fight often are apt to fight quickly and vigorously; but 

 since the correlations between these measures are less than unity, one might expect to 

 arrive at somewhat different genetic conclusions depending on the particular measure 

 employed. 



