G. K. Gilbert—Special Processes of Research. 463 
adjusted purely to considerations of convenience. Then for 
each pair of simultaneous values of the phenomena he finds 
the intersection of the corresponding lines and marks the point 
with a dot. If the value of either phenomenon falls between 
the numbers represented by two adjacent lines, the position of 
the dot is correspondingly interpolated. The relations of the 
dots to each other are then considered by inspection. If they 
are irregularly scattered over the sheet, the absence of a law is 
inferred. If they fall in line, either accurately or approxi- 
mately, the existence of a law is demonstrated. In that case a 
line is drawn through or among the platted points, and this 
line is an expression of the law or relation sought. The in- 
vestigator may or may not proceed to the computation of the 
algebraic equation. Asarulesuch procedure is profitable only 
when the platted points fall well in line, for it is only then that 
the proper form of equation can be selected with confidence, 
and without a knowledge of that form extrapolation is hazard- 
ous. For purposes of interpolation the free hand curve drawn 
among the platted points is practically as serviceable as the 
algebraic equation. 
Ordinarily, as in the employment of section paper, the lines 
used to represent numerical values are straight, parallel and 
equidistant, and the two sets intersect at right angles, but none 
of these characters are essential to the graphic method. The 
lines may be curved, they may converge, their interspaces may 
follow any law of increase or decrease, and the two sets may 
intersect at any angle. Nor is it essential that more than 
one line of each set be drawn—indeed, in a large group of 
practical examples even these are omitted, the edges of the 
paper standing for them. The graphic notation is thus highly 
elastic, adjusting itself freely to the convenience of the investi- 
gator. 
It is equally versatile in the character of the results it attains. 
Even when the phenomena do not practically admit of algebraic 
expression, and no representative line can be drawn upon the 
graphic chart, the form of the area occupied by the dots, or 
some other feature of their distribution may convey a meaning. 
If one in ignorance of the principle of the psychrometer should 
plat upon a sheet the simultaneous observations with a dry bulb 
thermometer and a wet bulb thermometer, representing the dry- 
bulb readings by abscissas and the wet-bulb by ordinates, he 
would find that the group of determined points could not be 
connected or generalized with any degree of approximation by 
a line, but if the number of his observations was large, he 
could not fail to perceive that the area occupied by the dots 
was definitely limited above by a diagonal straight line passing 
through the points of intersection of the lines representing 
