32 BULLETIN 1339, U. S. DEPARTMENT OF AGRICULTURE 



that there are striking differences in the conditions of the colony; in 

 the spring the brood rearing is increasing and the population of the 

 colony is not only greater but is made up predominantly of young 

 bees; in the fall the population is smaller, brood rearing is decreas- 

 ing, and many of the field bees are older than in spring. In the spring 

 in the vicinity of the Bee Culture Laboratory the honey flow is heavy 

 and nectar is abundant, whereas in the fall the daily gain is usually 

 small and the plants yielding nectar do so in relatively small quan- 

 tities. There may be an undetermined difference in the time of day 

 at which nectar is secreted (as is found between clover and buck- 

 wheat, for example) . In the spring, when nectar is unusually abun- 

 dant, the gains may be largely attributable to the influence of 

 increased temperature on bee activity, whereas in the fall the greater 

 gains at lower temperatures may be associated with the inverse re- 

 lationship which has been assumed between nectar secretion and 

 temperature. Until more observations are made this must remain 

 a matter of speculation. 



It will be noticed in Table 2 that the correlation between diurnal 

 variation of temperature and net gain in spring is .5967 ±.0704. A 

 wide daily variation in temperature on days of good gathering has 

 long been noted by beekeepers. There is some evidence, as previously 

 pointed out, that a preceding low temperature is necessary to induce 

 nectar secretion, and that once started this secretion will continue at 

 higher temperatures. The correlation of this same variable with the 

 fall changes of weight (Table 5) gives approximately the same figure, 

 namely, .5570 ± .0863. The fact that diurnal variation of temperature 

 affects the weight changes alike, both in the spring and fall, and the 

 observation of this phenomenon by beekeepers, would indicate that 

 this factor has a greater influence on nectar secretion than it does 

 upon bee behavior. 



For the fall period there is a correlation of — .7586 ±.0531 between 

 the average diurnal temperature and morning loss. The morning 

 loss is largely brought about by flight activity, and is therefore really 

 a quantitative measurement of it. Since there exists for this period 

 a positive correlation between morning loss and change of colony 

 weight, this fact would seem to indicate that the colder the weather 

 the greater the gain, as in reality the correlation between the changes 

 of colony weight and diurnal average temperature ( — .2310) actu- 

 ally shows, though to a small degree. The correlation between diur- 

 nal variation of temperature and change of colony weight is positive, 

 and the correlation between morning loss and diurnal variation of 

 temperature is also positive, despite the fact that there is a large 

 negative correlation between morning loss and the diurnal average 

 temperature. In calculating this latter coefficient of correlation, the 

 average temperature for the entire diurnal period was used, while 

 the period of morning loss covers but half of this time. Since the 

 maximum average temperature comes later than the end of the per- 

 iod of average morning loss, as shown in Figure 10, a correlation 

 with the temperature for the hours actually coinciding with morning 

 loss would undoubtedly result in a lower correlation between these 

 two variables. The evidence in favor of the theory that during the 

 fall temperature has a greater effect on nectar secretion than it has 

 on bee behavior is thus not materially weakened. 



