Practical Remarks on Hydrometry. 227 
The float was then thrown into the main thread of the 
stream, a few yards above the upper line, and the number 
of seconds observed to elapse in its passage from the upper 
to the lower line. The number of seconds consumed in this 
passage was observed about a dozen times, and that observa- 
tion which gave the least number of seconds was obviously 
adopted as the most correct. With this number of seconds, 
the distance of fifty feet, and the table, the mean velocity 
corresponding to the portion of stream under examination 
was computed. 
Then this mean velocity, multiplied by the mean sectional 
area of the stream within the specified limits, gave the discharge 
of the stream much more accurately than if one cross section 
only had been measured. 
If the observer be provided with Brunning’s, Woltman’s, 
or any other accurate tachometer, then Mr. Stevenson’s 
mode of determining the discharge of a river, by subdividing 
the cross section into a number of trapezoids, and determining 
the mean velocities for each of them as multipliers for the 
corresponding areas thereof, is a very correct one. I have 
followed this method when I formerly had occasion to deter- 
mine the quantities of water that passed to, and through, 
opposite a town on a tidal estuary, during the flow and ebb 
of a tide that rose and fell a certain number of feet and — 
inches at atide-guage. I tried Pitot’s tube during the opera- 
tion, but found it unsatisfactory for great velocities; also the 
hydrometrical pendulum, which was not adapted for deter- 
mining the great variations that occurred in the surface velo- 
city; for when the ball of the latter instrument worked well 
for’ moderate velocities, it proved quite unmanageable for 
high velocities, whilst for low velocities it showed no appre- 
ciable deflexion from the vertical. I subsequently con- 
‘structed a hydrometrical pendulum, furnished with four balls 
of gradually increasing specific gravity, so that whilst the 
lightest ball was very sensible to the influence of very small 
surface velocities, the heaviest was not too violently acted 
upon by the highest velocities that occurred in ordinary 
rivers. 
The co-efficients for each ball were respectively determined 
by direct experiment; and the numbers affixed to the balls 
and corresponding coefficients were registered on the in- 
strument. 
Then, in the measurement of the surface velocity of a 
stream, that ball was employed which proved on trial to be 
the best adapted for the measurement of such velocity, by 
