8 PROCEEDINGS OF THE NATIONAL MUSEUM ^ol. 122 



1598, and 2304 square inches. Employing a tape recorder, the be- 

 havior was described in a semicoded fashion by an observer concealed 

 behind a screen. The verbal descriptions on tape were transcribed 

 later, employing a constant 10-second time interval to allow further 

 for a consideration of the temporal changes in the behavior patterns. 

 An operations recorder coupled to a keyboard was also employed for 

 recording behavioral units in time. Both data-recording methods 

 are outlined in previous publications (Eisenberg and Kuehn, 1966; 

 Eisenberg, 1963b). Such data were analyzed by various methods 

 that permitted an accurate determination of the changes in the fre- 

 quency of responses, the duration of specific activity patterns, and 

 the frequency and durational changes over extended periods of time. 



A problem in quantification arose in comparing acts that had a 

 brief duration, coupled with a clear onset and cessation, with those 

 acts having a variable duration of unperfectly definable onset and 

 cessation. The more or less brief actions generally were counted and 

 described in terms of straight numbers per unit time. On the other 

 hand, such activities as driving, chasing, fighting, etc., were recorded 

 either in terms of actual duration or were counted for a frequency 

 analysis by employing a 10-second time criterion. Such acts having a 

 duration of 10 seconds or less were treated as one unit; if the duration 

 were greater than 10 seconds but less than 20, the action was treated 

 as two units, and so on. Indication is made in the text whenever 

 the 10-second criterion was appUed for a frequency analysis of acts 

 having a long duration. 



Patterns of social interaction were also recorded by the general 

 methods just described — by verbal means on tape or with an opera- 

 tions recorder. In addition, a thu'd method involved a weekly check 

 of cages containing freely growing populations. Pairing tendencies, 

 incidence of wounding, litters born, and breeding condition of indi- 

 viduals were censused in this manner. The three approaches to the 

 analysis of social interaction patterns are as follows: 



(1) The encounter in a neutral arena: In this situation, two 

 animals were placed simultaneously in a standard environment and 

 the resultant interaction recorded. Such variables as the sex, the 

 age, and the physiological state of the encountering animals could be 

 controlled (see Eisenberg, 1962a, 1963b). The glass-fronted en- 

 counter arenas included 748, 1598, and 2304 square inches of floor 

 area. Data from such standard encounter situations permitted 

 species comparisons with respect to differences in the frequency and 

 duration of various social behavior patterns. 



(2) The territorial encounter: This was a variant of the simul- 

 taneous encounter technique and involved the use of a cage divided 

 into two or three equal compartments as described previously (Eisen- 



