375 



because the winter of 1915-16 was, if anything, warmer than that of 

 1916-17. 



Data from Tables XII-XIIIG were used in making Fig. 9, which 

 shows smoothed curves of percentages of larvae failing to pupate in 

 experiments conducted at approximately constant temperatures. The 

 actual failure per cent is shown by circles, and mean data for experiments 

 within two degrees of each other, by crosses. Curves were first drawn 

 through the average points. These were then plotted on cross section 

 paper as in Fig. 10. and the same per cent of mortality connected between 

 the different humidities, and smoothed. (See Huntington '19, p. 252.) 

 The original curves were then corrected to fit the isofailiirc lines of Fig. 

 10. To make relations of the two figures clear, compare the failure per 

 cent at dift'erent temperatures on humidity 95% of Fig. 10, with tempera- 

 ture and failure per cent on that humidity in Fig. 9. (For fuller explana- 

 tion of these methods of graphic representation of results, see below, pp. 

 3S;3-393.) 



Townsend's ('26) results indicate that prolonged exposure to a tem- 

 perature of 50° F. decreases the percentage of pupation. Baumberger 

 ('17) secured similar results. Townsend showed, further, that soaking 

 in water increases the percentage of pupation and that the number of 

 soakings and the temperatures at which soaking is done are important. 

 Soaking frequently at 50° F., is most effective. The data graphed (Figs. 

 9 and 10) are representative, however, as they show a great many weather 

 possibilities in combination. 



Pupal mortality in constant-temperature experiments is shown for 

 the several mean humidities in Figs. 11 and 12. The method of drawing 

 the curves and smoothing them was the same as in Figs. 9 and 10. In 

 both cases (compare Figs. 10 and 12), the conditions are most favorable, 

 i.e., show low mortality (20% or less) and failure (50% or less), in the 

 neighborhood of 74° F., and 70-75% humidity. There are differences in 

 detail, but a drop in the mortality and failure lines at high temperature 

 for humidities of 75-85% occurs in both, leaving an upward extension of 

 favorable conditions at high temperature, both wet and dry. The diagrams 

 represent the relations in question only roughly, as the data were few and 

 quite irregular. This irregularity was evident in laboratory-hibernated 

 larvae (probably because of dift'erences in contact with water), some lots 

 showing higher mortality and failure to pupate and others showing little 

 or none. 



It is evident that variability is very important at the lowest tempera- 

 tures. One lot of larvae kept at 18°-50° F. showed no signs of pupation 

 until an accident to the thermostat sent the temperature up to 78° for an 

 hotir. In about two days several larvae pupated when the temperature 

 was abottt 48°., but all the pupae died without emerging. In one large 

 series of hibernating individuals, none pupated at 52° F. except within a 

 day or two after being transferred from 70° F. The influence of the 

 higher temperattires apparently persisted a few days. This may result 

 from one or more of the following catises : (a) lag in change of metabolic 



