67 
1913 - 14 .] Inclination of Curves of Thermoelectric Force. 
(1 inch = 05 kg. for abscissae, = 10~ 6 volt for ordinates), was examined to 
determine (i.) if it was a parabola, and (ii.) if so, if its axis was vertical or 
inclined. The method of examination differed from that employed by 
Dickson, in that it was first tacitly assumed that the curve was a parabola, 
the fact that the midpoints of several parallel chords were found to lie on 
one straight line giving support to this assumption, while not completely 
justifying it. The straight line in question then gave the direction of the 
axis. To determine the vertex of the curve, the tangent to it at right 
angles to the axial direction was drawn : this is easily performed with 
considerable accuracy by the aid of a long slip of glass having one straight 
line ruled on it all its length, and a shorter one at right angles across the 
slip. The shorter one is made to cover the determined axial direction, and 
then slid along it until the longer one just touches the curve. (The glass 
is, of course, turned with the lines next the paper.) The axis was then 
drawn at right angles to the tangent at the vertex, and the focus determined 
by trial, again with the glass slip, by finding the point on the axis whose 
distance from the vertex was half its perpendicular distance from the curve. 
This determined the latus rectum. Lastly, the inclination of the curve- 
axis to the E.M.F.-axis was measured roughly by protractor and (much 
more accurately) by square-counting, to find its tangent. The results of 
these various measurements were : 
Co-ordinates of vertex . . (22*48", 17*34") 
Length of latus rectum .... 25*2" 
Axial inclination, 
(i.) by protractor (mean of 8 readings) . 3° 54' (about) 
g 
(ii.) by square-counting (tan o> = Y^g) . 3° 48' 
From these data it is now possible to calculate the equation of the parabola, 
and the final stage of the work is then the verification that the values 
given in Table I. satisfy this equation to a sufficient degree of accuracy. 
The equation, as calculated to seven significant figures for each coefficient, 
was 
9956077a? 2 - 1322564 xy + 43922 if - 407990900* + 279653900 y - 206538300 = 0. 
For any assigned value of x the equation gives two values of y, one of which, 
since the coefficient of y 2 is very small compared with the other coefficients, 
will be practically infinite (and negative). Neglecting these infinite solu- 
tions as foreign to the problem, Table II. gives a comparison of the E.M.F.’s 
(y) corresponding to various loads f~) as calculated from this equation. 
