52 
PACIFIC SCIENCE, Vol. IV, January, 1950 
A POSSIBLE RHYTHM 
From about January 10 to March 20 there 
are several fluctuations with ranges of about 
0.6°. For most regions 0.6° would not be 
significant, but it might be significant for 
Honolulu where the conventional "mean an- 
nual range of temperature” is only 6.8° 
( 783 ° _ 71.5°) and the extreme range is 
only 38° (90° — 52°). Therefore a study 
was undertaken of the fluctuations in the first 
3 months of each of the 18 years from 1931 
to 1948. The value of 0.6° is about 9 per 
cent of the 6.8° mean annual range and is 
about 1.6 per cent of the extreme range. In- 
spection of Figure 2 shows troughs with cen- 
ter dates of Jaunary 16, January 28, Febru- 
ary 13, and March 8 or 9- Eighteen sets of 
7 -day progressive means were computed for 
the first 3 months, one set for each of the 
18 years. These means were plotted as 18 
graphs, and were inspected to see to what 
extent there were troughs that coincided with 
one another and with the troughs of Figure 2. 
In Figure 2 there are four more or less 
definite troughs and four or five more or less 
definite crests between January 1 and March 
31. The 18 graphs for the 18 years show 
from three to six troughs and from two to 
six crests, their distribution being shown in 
the following tabulation. 
Frequency of Numbers of Troughs and Crests 
per Year 
1 II III 
2 0 1 
3 - 5 2 
4 5 5 
5 7 6 
6 1 4 
L The number of troughs (or crests) per year. 
IL The number of years with the number of 
troughs indicated in I. 
III. The number of years with the number of 
crests indicated in I. 
The frequencies shown in the tabulation 
differ so much from the numbers of troughs 
and crests in Figure 2 that they make unten- 
able the idea that some meteorological rhythm 
causes the troughs and crests of Figure 2. 
Moreover, if there were some rhythmic 
cause that repeated in a significant number 
of the 18 years, we ought to find recurrences 
of troughs and crests on or about the same 
dates. A table was prepared with (a) the 
84 days for which 7-day progressive means 
were calculated and (b) the 18 years as argu- 
ments. Each trough or crest was entered in 
the body of the table at the proper date and 
year. Only a few dates gave more than 3 
years with either troughs or crests. January 5 
with four crests and February 21 with five 
crests were the highest, but neither of these 
dates is noteworthy in Figure 2. January 21 
had four troughs but also had one contradic- 
tory crest, and is moreover close to a crest 
in Figure 2. February 16, which follows by 
3 days the trough of Figure 2, had three crests 
in the 18 years. The data for January 21 and 
February 16 definitely contradict the hypoth- 
esis of a rhythmic control. 
- Smoothed averages for the 4 1 -year and the 
17-year periods showed no exact coincidences 
of troughs and crests with one another, nor 
with the 5 8 -year graph of Figure 2 . So we 
conclude that there is no evidence of an an- 
nually recurring, rhythmic cause to explain 
the troughs and crests of Figure 2. No doubt 
the troughs and crests of graphs for single 
years are due to the influence of fronts related 
to highs and lows that pass near enough to 
affect Honolulu’s weather. There certainly is 
no mystic relationship to the winter solstice. 
THE WARMEST DATE 
Selection of a warmest date- is less simple 
because the smoothed values form a sort of 
plateau of 36 days, including 30 dates of 
78.4°, with 2 dates a tenth of a degree lower 
and 4 dates a tenth of a degree higher. 
Among the 7-day progressive means for the 
58-year data, we find 78.5° on August 16, 
19, 20, 21, and 22, with 78.4° on August 17 
and 18. The 7 -day progressive means were 
subjected to a second, similar smoothing, 
which gave a series of five dates, August 18 
to 22, with 78.5°. The middle date is 
August 20, which may therefore be taken as 
the warmest date. 
