454 G. K. Gilbert—Special Processes of Research. 
work, it was known that thunderstorms, although too large to 
be fully observed from a single station, are too small to be in- 
vestigated by means of so coarse a mesh of stations as that of 
the national weather bureau. A thunderstorm observed at one 
of these stations frequently escapes notice at the other stations 
of the vicinity, while many of them occur between stations and 
are thus entirely omitted from their records. It was known 
that the local phenomena of a thunderstorm are of compara- 
tively brief duration, so that the tri-daily observations do, not 
afford a complete record of their occurrence. It was known 
that some thunderstorms move across the country, appearing 
successively at different points. It was known that they are 
characterized by electric discharges as well as by rain, wind 
and temperature change, and it was known that their occurrence 
is more frequent in summer than in other seasons of the year. 
It was believed or hoped that some conspicuous, readily ob- 
served feature of their transit might be found so persistent and 
constant in its relation to the aggregation of phenomena con- 
stituting the storm that it could be used for the purpose of 
tracing out the path and general form of the meteor. The 
society was unable to pay its observers for their services, and 
it recognized the fact that those who might voluntarily codper- 
ate with it must not be overtaxed, or they would withdraw 
their aid. 
With these considerations, and doubtless many others, in view,. 
the work was planned and organized. The initial steps were the 
arrangement of a system of stations for observation, and the 
preparation of schedules of phenomena to be observed. A 
large number of resident observers were engaged, so distributed 
that the interspaces between stations were much smaller that in 
the system of the Signal Service. Provision was made for ob- 
servations at very short intervals of time during the approach 
and passage of thunderstorms, but no record whatever was re- 
quired, except a general negation, on days when no thunder- 
storms occurred. The work was restricted to the summer 
months. The observers were classified in such way that the 
greater number were called upon to record but few features of 
the storm, while a few who were able ard willing to bestow 
much time and attention were instructed to make very full and 
minute records. The schedule of observations included, be- 
sides numerous other features, a number of definite events, 
namely, the time of the first rainfall, of the last rainfall.and of 
the heaviest rainfall, the time of the first thunder heard, and of 
the loudest thunder. 
Looking at this organization for observation from our special 
point of view, we are able to group the considerations deter- 
mining its detailsin four categories. In the first place, there 
