456 G. K. Gilbert—Special Processes of Research. 
made it possible to develop a plan whereby a small number of | 
the necessary instruments can be so distributed as to perform 
their work effectively and economically. This assumption of 
our own is based on the analogy of other investigations, for it 
is a common feature of researches that the preliminary results 
determine the scope and field of subsequent observation. A 
research is the exploration of an unknown land, and neither 
the route nor the goal can be foreseen. The explorer climbs 
the hill before him, and from the vantage of its summit selects 
the most promising course fur the next stage of his journey. 
One of the most important obstacles encountered in the work 
of the society arises from the difficulty of making observations 
at a distance from the surface of the earth. A thunderstorm 
involves a mass of air of very considerable vertical dimension, 
and the ideally complete system of stations would include many 
more points in mid-air than upon the ground. Observations 
from balloons, though within the range of possibility, are 
limited by expense and danger, and would moreover be con- 
ditioned by uncertainty as to location at all points within the 
storm cloud.. The society has thus far attempted only ter- 
restrial stations, but it endeavors by indirect methods to learn 
something of what is taking place at greater altitudes. The 
wind cannot be directly observed there, but floating objects 
can, and observers are therefore instructed to watch the motions 
and observe the positions of clouds. Temperature and mois- 
ture cannot be directly measured, but advantage is taken of the 
fact that a certain combination of temperature and moisture 
produces condensation, so that the boundary of a cloud is 
the locus of a certain atmospheric condition which is a joint 
function of temperature and moisture. The configuration of 
the storm cloud therefore has a certain value, and is the sub- 
ject of study. It was also known antecedently that as the 
domain of saturation advances in the air, the resultant cloud 
grows, while its recession is characterized by the dissipation of 
the cloud. Observers are therefore instructed to record cloud | 
formation and clond dissipation. 
The preliminaries haying been arranged, observations were 
made during the summer of 1885, and the records of observation 
were placed in the hands of Professor Davis for the purpose of 
what is variously called “reduction,” or * working up,” or “ di- 
gestion,” or ‘‘ discussion.” What he did was to classify them in 
various ways, and observe the relations brought to light by the 
classification. But before he made a classification, it was neces- 
sary to select its basis. The particulars of observation were 
numerous. They included the intensity of the storm, the first 
thunder heard and its direction, the loudest thunder and its 
direction, lightning strokes, the first rain, the heaviest rain, the 
