in the early morning, and a lesser peak in the evening. Winnowing at 
other times of the day or night was so sporadic as to give no indica- 
tion of the number of pairs actually present. Figure 1 gives a compar- 
ison of winnowing activity in the evening as compared with the early 
morning. The heavy lines in figure la and lb portray the amount of 
winnowing on a fairly typical morning or evening, whereas the dashed 
lines show how the presence of a bright moon will spread out the 
winnowing period and reduce the amount of winnowing at the usual time. 
Observations are depicted in terms of number of winnowing calls heard 
per minute (Eastern Standard Time), rather than the actual number of 
birds winnowinge The average frequence of winnowing was about 5 per 
minute, so "5" on the vertical scale indicates steady winnowing by 
one male, "10" corresponds to two males, and "15" means that all 
three males were performing. A typical pattern for a single bird is 
shown between 2:55 and 3:28 a.m. in Figure la. 
Four evening and six early morning counts made during favorable 
weather conditions showed that evening winnowing started between 25 
and 62 minutes after sunset and continued until 62 to 78 minutes after 
sunset, whereas morning winnowing started frem 70 to 88 minutes before 
sunrise and lasted wmtil 30 to 6 minutes before sunrise. The usual 
winnowing periods ran fram 4O to 70 minutes after local sunset, and 
from 75 to 35 minutes before sunrise, The heavy line in Figure la, 
is typical of the moonless morning observations; by Jume 22 there 
was a decline of one-third to one-half in winnowing. 
The evening data were much more variable, and no one night Could 
be considered typical. The time of starting and the total duration 
of activity were extremely variable, and pronounced irregularities 
were noted each evening. Records for Jue and Jume 21 are shown in 
Figure lb, and a condensed record for June 7 appears in Figure lc. 
The greatest evening activity took place on Jue 21, so this date was 
chosen for inclusion in Figure lb, in spite of the fact that it was 
comparatively late in the season. On the evening of June 1 (not 
pictured) there was a brief winnowing period from 9:51 to 9:58, and 
another from 10:18 to 10:23; during 6 consecutive minutes two of the 
three birds were winnowing, in 8 other one-minute periods one bird 
was winnowing, and for more than half of the usual 30-minute winnowing 
period there was no wimowing at all. The evening period, therefore, 
is relatively uwndependable as compared with the pre-dawn period, 
An all night count made on the night of the full moon is shown 
in Figure lc, Each entry on the graph represents an average of six 
consecutive minutes, rather than a single minute as in Figures la and 
lb; the vertical scale is the same as in Figures la and lb. On this 
particular night not more than one bird was heard at any time. Un- 
fortunately, a thunderstorm disrupted observations just at the start 
of the pre-dawn winnowing periods one or more birds were winnowing 
actively during the period show by the dotted line, but high winds 
made it almost impossible to hear them. 
47 
