686 Memorandum on the Adjustment of Dynamometers. 
that is, the indications of the counter multiplied by the co- 
efficient C which we are in search of, whence 
I lbs. X 300 ft . . 
2 
and any of the other trials will give the same value to C. 
Having determined C, if we know the distance d, travelled 
over "by the instrument, and in any experiment the reading of 
the counter, say R, we get the mean draft at once by the equation 
J _C X It 
But it frequently happens, especially in inexperienced hands, or 
if the joints of the instrument are loose, or from the elasticity of 
the frame, or if there is an initial strain on the springs, or from 
some or all these causes, that the line joining the ordinates 
does not, when produced, cut the base at o, but within or 
beyond it, and the tops of the ordinates themselves are not 
exactly in a line : in that case draw a line, as in Fig. 2, through 
the tops of the first and the third ordinates, it will cut the base 
at o" ; take next the second and third, it will cut the base at o"\ 
and so on, combine all the ordinates, and draw the best mean 
straight line through them, and finally raise a new ordinate at 
0, and measure its mean height, which will represent the error 
of the instrument ; because if aline s^" be drawn through its 
extremity parallel to the base, the ordinates measured from sf 
will be nearly in proportion to their distance from o ; that is, 
the indications of the counter with o s, deducted in each case, 
will be proportional to the loads on the springs. 
If the mean of the diagonal lines through the ordinates cuts 
the base-line inside the triangle, as in Fig. 3, then the ordinate 
0 s must be measured downwards, and added to the reading of the 
counter. A little consideration will show that the co-efficients 
of error, as we may call o s, must vary with the distance tra- 
versed by the instrument only, and not with the strain on the 
springs. If the error arises from looseness of joints^ it is plain 
that the integrating wheel may move a little before the springs 
come into play, and revolve slowly as the instrument moves, 
recording work, though none is done ; and, therefore, the 
counter will indicate too much, so that the correction must be 
deducted : but the joints, having once taken their bearing, will 
yield no more, however much the spring is strained, and there- 
fore the amount of error will not depend on the strain on the 
springs. Again, if there is initial strain on the springs, the 
integrating wheel will remain immovable, recording no work, 
though work is done, until the draught applied exceed that 
strain, when it will commence registering ; but it will count too 
