VISION 



117 



and P20. It is easy to condition bees so that they visit a pattern 

 composed of relatively many contours or outhnes such as crosses 

 or a chessboard and to avoid a plain circle or square of the same 

 colour. So far, however, it has not proved possible to condition 

 bees to show the opposite preference. Hertz's conclusion that a 

 predilection for patterns with many contrasting outlines is present 

 in bees, before any training to \'isit specific patterns has taken 

 place, has been confirmed by Zerrhan (1933). These san.e results 



Pl5 



ri2 



G5 



P20 



Fig. 54. In this figure all the patterns extend over equal sized 

 superficial areas. G5 shows the largest amount of outlines 

 (2,000 mm.), but, because of its uniformity, it is less attractive 

 to bees than P20, which is composed of numerous small units 

 with a greatly reduced amount of outlines (570 mm.). F12 

 (1,600), however, is much more attractive than either of the 

 foregoing patterns ; it is only eclipsed in attractiveness by one 

 composed of a very large number of small units P15, but with 

 reduced amount of outlines (1,026 mm.). An experiment of 

 this kind forms an infrequent example in which bees can be 

 trained both ways with regard to a pair of patterns. Thus, P20 

 is more attractive than G5 but less so than F12. (After Hertz.) 



have also been obtained for other insects. Use (1932) used the 

 same patterns as those employed by Hertz in connection with 

 experiments with butterflies. She found that if the colour of the 

 pattern were very attractive to the insect that pattern which had 

 the largest area of colour and less outlines was preferred to a 

 pattern of smaller area but greater amount of outlines. Where, 

 however, the colour used only exercised a weak stimulus, the 

 behaviour of the butterflies was of the reverse kind and the same 

 response was obtained as that described by Hertz. The last- 



