470 G. K. Gilbert—Special Processes of Research. 
For example, in describing Prof: Davis’s preparation of the 
composite portrait, | have represented him as computing cer- 
tain distances from knowledge of the corresponding times; 
but in point of fact he saved himself that labor by a very sim- 
ple graphic expedient. The velocity of the storm being ap- 
proximately uniform, distances in the direction of motion were 
proportional to times; and as distances on the composite were 
proportional to distances on the ground, they were likewise 
proportional to times. He therefore graduated a time scale on 
a strip of paper, laid it on the composite sheet parallel to the 
storm-track axis and at the distance corresponding to a particu- 
lar station, slid it along until the graduation for the time of 
first rain at that station was over the rain-front axis, and then 
marked on the composite opposite the proper time graduations 
all the other observations made at that station. 
Another of these special devices it has fallen in my own way 
to employ, and I will take the liberty to describe it, for while 
its field of application is narrow, it wus found exceedingly use- 
ful within that field. Whether it is novel, 1 am not informed. 
It is a method of graphic integration allied in principle to the 
algebraic process known as mechanical quadrature. 
Of the many variations to which the pressure of the atmos- 
phere upon the barometer is subject, there is one having a daily 
period. arly in the morning the barometer rises, then during 
the middle of the day it falls, reaching a minimum in the after: 
noon, and there follow during the night one or more minor 
undulations. This oscillation was recognized long ago, and it 
is the subject of an extensive literature, and many theories 
have been framed to explain it. One of these, proposed by 
EHspy,* appeared to me so plausible as well as simple that I 
undertook some years ago to test it. The theory is this: Dur- 
ing the day the atmosphere, being warmed by the sun, is ex- 
panded and made to occupy greater space. Every stratum ex- 
cept the lowest is thus moved upward, and as the air is endowed 
with mass, time is consumed in putting itin motion. While 
the expansive force is acting, and so long as fully equivalent 
motion has not been produced, there is a reaction downward, 
causing arise of the barometer. Conversely, when the rate of 
heating or the rate of expansion diminishes, the upward 
momentum of the air produces the reverse effect, a defect of 
pressure on the barometer. The addition to the normal pres- 
sure of the barometer is greatest, not in the afternoon when the 
temperature of the air is highest, nor yet at noon when the 
heating of the air is most rapid, but in the morning when the 
increment to the rate of heating is greatest. Now the momen- 
* Fourth Meteorological Report. Foot note to p. 11. 
