SOLAR VARIATION AND WEATHER—ABBOT 145 
TaBLE 4.—Copenhagen temperature departures, smoothed. Test of 8%-month period 
[Values of all months employed. Means only given. Unit: 1/100 degree C. throughout.] 
Years November-December beginnings 
£798 to 1833- - ._-.-- Lo 24 21 7 84 ut 18 36 
1833 to 1868_------- —153 | —55 | —48 65) TO fa eek — 66 
1868 to 1903__------ 20 85 88 | 103 33 Wah es sh) 
1908 to 1937-._----- — 102 20 36 44 29 —4 =—9 —20 
1798 to 1937______-- eit ks 22 29 65 AD bee heal Loe 2e 
1798 to 1937 
Nov. and Dec--._-_-- —79 22 29 65 41 Si = 1S ee 
Jan. and Feb---_-_-- 2 i 3\| eae 34 LS ae 0 aie at 0 
Mar, atid Apr. .___-- —29 | —29 | —64 |} —19 | —66 | —39 | —29 15 
May and June.:.__--| —1l 45 44 4h) 12 24 33 46 
July and Aug___--_-- 29 Pte Ni 8 I OI eel: 16 Su) iat 
Sept. and Oct-_-_---_-- 24 32 52 6 5 18 40 42 
That there is here no progressive secular displacement of the phases of means of groups beginning at a 
constant season of the year, is shown by the extended table for November-December. But groups begin- 
ning at different seasons of the year do show displacement of phases with respect to one another. 
I soon found that while there seemed to be some tendency to perio- 
dicities in weather corresponding to the solar changes, these weather 
periodicities, unlike their solar counterparts, fluctuated in phase. It 
occurred to me, however, that this instability of phase is but a natural 
seasonal effect for the periods of shorter duration like 8, 934, or 1144 
months. For the phase of terrestrial response to a solar cause must 
evidently depend on local terrestrial circumstances. For instance, 
there will be a longer lag with stations under oceanic control than for 
those in cloudless deserts. Pursuing this thought it occurred to me 
that the phase, for example, of an 8-month period of response to solar 
change in weather must be different if the solar cause occurs in 
summer than if it occurs in winter. I investigated this idea for several 
periods and many stations. Figure 15 shows that my surmise was 
a correct one. 
Hence terrestrial responses to solar periods of moderate lengths 
should be expected to be in the same phases only when the solar causes 
occur at the same time of the year. If a solar period of exactly 8 
months existed, we must compare its terrestrial effects 2 years apart, 
for then their solar causes would occur exactly at the same seasons of 
the year. 
I will not delay to show exactly how we make use of the calendar to 
eliminate seasonal phase changes, but will content myself with show- 
ing for three stations, Copenhagen, Vienna, and New Haven (see tables 
3 and 4), that when this complication is properly allowed for, and when 
the exact length of the solar period is determined, the terrestrial re- 
sponse is proved to be exactly in phase from the year 1700 to the present 
