172 
to improve my results for blue and 
white ones. 
Establishment of Distant Mountain 
Solar Stations 
Our program of day-to-day obser- 
vation of the solar constant to detect 
possible variations in the sun’s output 
of radiation, which might be of im- 
portance for weather forecasting, re- 
ceived a great impetus in 1916. 
H. H. Clayton, then chief forecaster 
of Argentina, sent us a paper® in 
which he showed that if our Mount 
Wilson solar-constant values were 
used in large groups, graded with 
respect to size, temperatures in all 
parts of the world seemed to be cor- 
related with them. I reported on this 
paper to Dr. Walcott, then Secretary 
of the Smithsonian Institution, and 
he was also impressed. 
Excellent as Mount Wilson at- 
mospheric conditions are for photo- 
graphic stellar work, the variable 
haziness, associated with nearby cities 
and the ocean, make it difficult to 
determine there the variations of the 
solar constant of radiation. After 
much inquiry it appeared that the 
region of Calama in northern Chile 
might prove more _ suitable. Dr. 
Walcott authorized an expedition to 
Calama, and in 1918 the Smithsonian 
established a solar observing station 
there, directed by Alfred F. Moore. 
In 1920, by the generous assistance of 
John A. Roebling, the Chilean sta- 
tion was removed to the summit of 
Mount Montezuma. This _ station, 
about 10 miles from Calama, lies in a 
nearly rainless desert at about 9,000 
feet altitude. We have occupied many 
solar observing stations in far-away 
desert lands, but Mount Montezuma 
has proved the best, and is still giving 
us nearly daily determinations of the 
solar constant of radiation. Another 
station on Burro Mountain, N. Mex., 
was occupied until late in 1946; a 
8 Smithsonian Misc. Coll., vol. 68, No. 3, 
May 1917. 
ANNUAL REPORT SMITHSONIAN INSTITUTION, 1948 
third, on Table Mountain, Calif., is 
still in daily operation. 
The variation of the sun’s output of 
radiation, though small in percentage, 
appears to be a factor of importance 
in weather. By extensive statistical 
studies of the quarter-century of 
nearly daily solar-constant values now 
available from Montezuma and the 
other Smithsonian stations, I have 
shown that the short-interval solar 
fluctuations of 3 to 5 days in length 
strongly affect temperatures. The 
temperature changes are large. 
In a recent paper ° I show that there 
is a regular periodic variation of solar 
radiation of 6.6456 days. Its ampli- 
tude, not yet thoroughly determined, 
seems to vary from zero to over 1 
percent of the sun’s output of radia- 
tion. The average magnitude of the 
variations seems to be about 0.13 
percent. Though perfectly regularly 
periodic as a solar variation, its effects 
on terrestrial temperature show what, 
by mechanical analogy, we might 
call “‘backlash.”? That is, they some- 
times are 1, 2, or, rarely, even 3 days 
early or late compared to the average, 
but rarely change in phase much in 
the four or five recurrences of each 
single month. The magnitude of the 
temperature changes is also variable, 
but surprisingly large. It ranges from 
2° F. to 20° F. Thus these short solar 
changes are major influences on the 
weather, hitherto unrecognized be- 
cause irregular in recurrence to the 
extent just described. In a test fore- 
cast of 36 days, based on the assump- 
tion that there would be no change 
in “‘backlash” or in amplitude during 
that interval, the ups and downs of 
temperature at Washington were cor- 
rectly predicted as to times, and the 
average daily difference, observed 
minus predicted temperature, for 27 
days, until a change of amplitude did 
occur, contrary to the preliminary 
assumption on which the forecast was 
based, was 6.6° F. If the causes of 
® Smithsonian Misc. Coll., vol. 7, No. 4, 
Apr. 4, 1947. 
