356 
NATURE 
[FEBRUARY II, 1904 
in three different ways. First, by displacing air the density 
of which is three-eighths greater ; second, by evolving latent 
heat in its condensation; and thirdly, by causing ascending 
currents, and thus reducing dynamically the pressure of ‘the 
atmosphere as a whole. ‘The first and second of these 
processes do not indeed directly reduce the pressure but only 
the density of the air stratum while they increase its volume. 
In order, therefore, that the observed effect may follow, 
a portion of the higher atmosphere must be removed, and 
this will necessarily flow away to regions where the pro- 
duction of vapour is at a minimum, viz., the polar and 
cooler portions of the temperature zones, and more especially 
those where a cold dry Jand surface radiates rapidly under 
a winter sky. Such an expanse is the great northern plain 
of European Russia and Western Siberia north of the 
Altai.”’ 
In 1886 we got the first fruits of the observations of the 
widened lines in sun-spots, which had been obtained on a 
definite plan, since 1879. “The changes which occurred from 
a spot-minimum to a spot-maximum, and some distance 
beyond, had therefore been recorded. The changes were 
most marked, showing a great change in the chemistry of 
the spots at these times. At minimum the lines chiefly 
widened were those of iron and some other metals, but at 
the maximum the lines widened were classed as “* un- 
known,’’ because they had not been recorded in the spectra 
of the terrestrial elements. It was reasonable to suppose, 
therefore, that the sun was not only hotter at maximum, 
but hot enough to dissociate iron vapours (Proc. Roy. Soc., 
1886, p. 353)+ 
In 1891 Janssen’s suggestion of 1869 was brought into a 
practical shape for observatory work by Hale and 
Deslandres (Comptes rendus, August 17, 1891), and the 
prominences on the sun’s- disc and surrounding it were 
photographed in full daylight by using only the light 
radiated by the calcium vapour, which they always contain. 
By the year 1900 we had accumulated at South Kensing- 
ton observations of the widened lines for a period of more 
than twenty years. There was a curious break in the 
regularity of the results obtained after 1894, and the Indian 
meteorologists reported contémporaneous irregularities in 
the Indian rainfall. 
I determined, therefore, to make a connected inquiry into 
both these classes of phenomena. ‘Thanks to the establish- 
ment of the Indian Meteorological Department in 1875, we 
had rainfall tables extending over a quarter of a century, 
and in the tropics, where the problems might be taken as 
of the simplest, to compare with the new solar data. 
I have already stated.that in the preliminary discussion of 
the most widened lines observed in the sun-spots up to the 
year 1885 a most remarkable difference was observed in the 
lines observed at sun-spot maximum and minimum. This 
continued until about 1895, another ten years. As the curve 
of iron lines went up, the curve of “‘ unknown ”’ lines came 
down; there were therefore crossings of the curves which 
might, on the hypothesis before referred to, be taken as 
the times at which the temperature of the sun had a mean 
value. These crossings turned out to be about half-way 
between the maxima and minima of the spotted area which 
had to be considered as.the times at which the sun was 
hotter and colder than the mean. 
We were then brought into the presence of three well- 
marked stages of solar temperature—it was no longer a 
question merely of spots and no spots, but of heat pulses. 
The next point was to study these heat pulses in relation 
to the Indian rainfall, and it was found that in many parts 
of India the plus and minus heat pulses on the sun, which, 
of course, occurred immediately after the time of mean 
temperature, when the sun was getting either hotter or 
colder, were accompanied by pulses of rain in the Indian 
Ocean and the surrounding land. It was next found, from 
a study of the Indian Famine Committee's reports, that 
the famines which have devastated India during the last 
half century have occurred in the intervals between the 
pulses. 
In 1902, with the view of getting more light on the im- 
portant issues raised by the comparison of the solar heat 
pulses and the Indian rainfall, I determined to reduce the 
observations of prominences made by Tacchini at 
Observatory of the Collegio Romano since 1874, and to com- 
NO. 1789, VoL. 69] 
the | 
pare the Indian meteorological conditions with them. The 
reason for this step was that the admirable photographs of 
the prominences on the solar disc, published by Hale and 
Deslandres, showed the extensive area over which they 
were distributed. An argument which has been used 
against the possible connection between solar and terrestrial 
changes was based upon the small area covered by spots. 
In 1877 Eliot wrote as follows (Report on the Meteorology 
of India, 1877, p. 2) :— 
““ So far as can be judged from the magnitude of the sun- 
spots, the cyclical variation of the magnitude of the sun’s 
face free from spots is very small compared with the surface 
itself; and consequently, according to mathematical 
principle, the effect on the elements of meteorological observ- 
ations for the whole earth ought to be small.” 
Now the photographs to which I have referred exhibited 
broad bands of prominences extending almost across the 
whole disc, and if we assume two belts of prominences, 
north and south, 10° wide, with their centres over latitude 
16°, a sixth of the sun’s hemisphere would be in a state 
of disturbance. Hence it followed that the prominence 
effect, when fully studied, might be much more striking 
and important than that produced by spots. 
The prior work in connection with the Indian rainfall had 
shown not only that there was a close connection between 
pressure and rainfall, but that the pressure was much the 
more constant element over the different areas. The com- 
parison with the prominences obtained from the discussion 
of Tacchini’s results was in the first instance compared with 
the Indian pressure curve. 
The result was magnificent. In addition to the well- 
marked prominence maximum at the maximum of the 
spotted area, there were others corresponding approximately 
with the “‘ crossings ’’ of the widened lines, and all were 
re-echoed by the Indian barometers ! 
The sun-spot cycle of eleven years gave way to a promin- 
ence cycle of about 3-7 years, and by this interval, as a 
rule, are the Indian pressures separated. 
To see whether such a striking and important result as 
this was limited to Indian ground, the important series of 
pressure observations obtained at Cordoba in South America 
were studied. Here the same effect was also most marked, 
but with the important difference that the curves were in- 
verted, that is, high pressure years in India were represented 
by low pressure years in Cordoba. 
In order to extend the Indian and Cordoba areas and to see 
how far these conditions prevailed, the pressure variations 
of stations as widely distributed as possible were examined. 
The result of this inquiry showed that the world might be 
divided roughly into two portions. The Indian area was 
found to extend to Australia, East Indies, Asiatic Russia, 
Mauritius, Egypt, East Africa, and. Europe, while the 
Cordoba region might be said to include not only South 
and Central America, but the United States and Canada, 
extending further west than Honolulu. 
The discovery of this barometric surge, which has been 
corroborated since by Prof. Bigelow, was an important 
advance, and will enable the investigator to connect up 
regions that undergo similar pressure changes. 
In addition to the two periods, namely, 11 and 3-7 years, 
mentioned above,  Briickner (** Klimaschwankungen,’” 
Eduard Brickner, Vienna, 1890) has pointed out that there 
is a long period weather variation. His discussion of all 
the available data of pressure, rainfall, temperature, Xc., 
led him to conclude that there is a periodical variation in 
the climates over the whole earth, the mean length of this 
perioa being about thirty-five years. 
Since this work, a recent discussion of the sun-spot data 
by Dr. W. J. S. Lockyer (Proc. Roy. Soc., vol. Ixviii. pp- 
285-300) has brought to light a similar long period, and 
this has taught us that each eleven-year cycle is different 
from the one immediately preceding and that following it. 
A further inquiry into the distribution of the solar promin- 
ences, as observed by Respighi, Secchi Tacchini, Ricco, and 
Mascari (Memorie della Societa degli Spettroscopisti 
Italiani), has resulted in increasing our knowledge of the 
circulation of the solar atmosphere. The centres of promin- 
ence action, or the centres of the prominence belts, have a 
tendency to move from low to high latitudes, the opposite 
of spots; generally speaking, two belts in each hemisphere 
