300 CALVIN B. BRIDGES 



value is maintained to about 27°, or throughout the whole range 

 of 'room' temperature at which the breeding work is ordinarily- 

 conducted. At 29° the crossing over is slightly freer than 

 normal, but between 29° and 31° the amount of crossing over 

 nearly trebles (18.2). This extraordinarily sharp and extensive 

 rise is followed by a slight fall at 32° (15.4 per cent). Above this 

 temperature, and below 9°C. it was found that the flies either 

 died or produced too few offspring to be workable. It seems 

 probable that here also these two sharply marked maxima 

 separated by a long interval of no or slight change may represent 

 two distinct phenomena. 



When the coincidences are calculated for these various tem- 

 peratures, it is seen that the curve of temperature variation of 

 coincidence is a slightly rising but practically straight line 

 cutting alike through both of the maxima and the normal in- 

 terval. When the variations in crossing over due to exposure 

 to abnormal temperature are followed through the life of the 

 fly by means of tube counts, characteristic rises and falls appear 

 on the curve which expresses this relation. When a coincidence 

 curve is calculated from these same data, it is seen that it cuts 

 through the rises and falls due to temperature exactly in the 

 same way as it did in the temperature and coincidence curves 

 constructed from experiments at diverse temperatures. This is 

 a significant difference from the relation previously observed in 

 the age variations, and would seem to indicate that the age and 

 temperature variations w^ere accomplished by different mecha- 

 nisms — by effects upon different physiological factors. Those 

 double crossovers, that do occur have the same distribution along 

 the chromosome at all temperatures, which shows that the 

 method of handling the chromosomes is unchanged. In accord- 

 ance with the analysis already given (p. 298) the cause of the 

 temperature variations in crossing over is to be sought rather in 

 variation in the coefficient of crossing over — in the crossing over 

 capacity of the chromosome itself because of some variation in 

 its structure or framework. 



The conclusion just drawn from the failure of the temperature 

 variation to affect the coincidence, namely, that the change in 



