12 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 82 



After the temperature and relative humidity in the dark-room had 

 become nearly constant, both shutters at the bottom of the photo- 

 geotactic box v^^ere inserted and 25 adult beetles were then put in the 

 box on the cloth-covered glass shutter, after which the glass shutter 

 at the top of the box was inserted, and finally the testing apparatus 

 was assembled as shown in figures 2 and 3. The blue electric light 

 bulb was then placed 10 inches directly above the water screen, which 

 prevented the infra-red or heat rays from the light from reaching the 

 insects. After a period of 15 minutes the wooden shutter was re- 

 moved and the box was gently raised and turned toward the light so 

 that the observer could see through it from end to end. The position 

 of the beetles in the ten sections was then quickly noted and recorded 

 in a table. 



In test number 2 the beetles were again put in the bottom of the 

 box on the cloth-covered shutter, but this time the box was gently 

 turned upside-down and rested on the water screen, 10 inches below 

 which was the light. In this case the beetles clung to the cloth and 

 did not fall when turned upside-down. During the same forenoon 

 tests number i and 2 were repeated alternately three times, and 

 finally the results obtained in the two sets, each consisting of four 

 tests, were treated statistically and recorded in table i. 



To secure a frequency distribution the responses were given values 

 ranging from i to 10, corresponding to sections i to 10 in which the 

 25 insects were counted at the end of a test. If all 25 insects remained 

 in section i, the total value would be 25, or i as an average ; and if all 

 moved to section 10, the total value would be 250, or 10 as an average ; 

 but neither one of these extremes was actually observed, because the 

 insects were always counted in two or more sections. To secure the 

 arithmetic mean, which in each test lay somewhere between i and 10, 

 the total value was divided by 25, and since each set of experiments 

 consisted of four tests the frequency curve was represented by only 

 four means. Owing to the small number of statistical items, Bessel's 

 formula 



(f. £. ;;/=± 0.6745 J -f^) 

 V ^ n{n-i)/ 



was used to calculate the probable errors. 



The values were accurate for those insects counted in sections 2 to 

 10, but not so for those counted in section i, because they may have 

 responded little or none, although most of them had left the cloth- 

 covered shutter on which they were put. Consequently, the mean 

 positions given in table i under the headings " up " and " down " 



