130 
is greater probability of obtaining perfect homogeneity and 
accuracy of dimension, whilst they may be more readily 
and thoroughly magnetized. I have, after much trial and 
consideration, fixed upon two inches as the most convenient 
length, and on one quarter of an inch as the section of the 
bar. The bars are first filed, then thoroughly hardened, and 
afterwards ground to an exact figure, the operations being 
finished by rubbing the bars together with a little cutting 
powder between them. The bars may easily be made so 
true that when dry and perfectly polished, they will adhere 
strongly together. When finished, they must be measured 
by a standard rule under a microscope — the means of obser- 
vations in many places being taken for the respective dimen- 
sions. Their weight must also be accurately determined. 
The moment of a homogeneous parallelepiped in vibra- 
tion is found by multiplying the square of the diagonal of the 
upper surface by the weight of the bar, and dividing by 12. 
If it be thought desirable to test the accuracy of the 
moment thus deduced, the best plan is to fasten small cubes 
of lead to the ends of the bar, by an elastic band. These 
cubes must be carefully wrought and measured, and their 
moment may be found by subtracting the moment of a bar 
of lead two inches long from that of a bar two inches long, 
plus the length of the two pieces. When the moment of 
the pieces of lead is determined, that of the steel bar 
may be found by vibrating it first by itself, and then with 
the lead cubes attached. Another plan is to hang pieces of 
lead from the ends of a filament of silk thrown over the bar, 
but in this case great care is requisite that the filaments have 
very little force of torsion, and that the vibration is very slow. 
The moments of the four bars I have finished, as deter- 
mined by the above means, are as follows :■ — 
