358 S. H. BARRACLOUGH AND T. P. STRICKLAND. 
Of these three, the first was the point under immediate investi- 
gation, the second and third being more especially required for 
the purpose of correcting the velocities in the first, for small and 
unavoidable changes of hydraulic radius and temperature. 
3. General plan of apparatus.—When the object of an investi- 
gation is the determination of the law of variation of some par- 
ticular quantity, rather than the magnitude of coefficients or 
constants required for practical use, there are decided advantages 
to be gained by planning the apparatus on a small scale, so that 
the various conditions of the experiment can be easily controlled. 
For this reason it was decided to use a comparatively short 
channel, made of planks the full length of the channel, and to 
limit the water-way to a small section, so as to ensure the water 
attaining its uniform régime within a short distance from the 
entrance. The general disposition of the apparatus is indicated 
in Fig. 1. The water was drawn from the ordinary service 
supply into the supply tank, whence it passed by one or more 
orifices to the tin-lined inlet box at the channel entrance, then 
along the channel to the gauging tank, where its amount was 
accurately determined, and finally into the drain. For the 
temperature tests the water in the tank was heated by the conden- 
sation of steam from a neighbouring boiler, the steam being lead 
into the tank by steam piping, not shewn in the figure. The slope 
of the channel could be readily altered by raising or lowering its 
supporting trestles. For convenience of observation and compu- 
tation the metric system was used, as far as possible, throughout 
the investigation. 
4. Supply and control of the water—The details of the supply 
tank are shewn in Fig. 1. It consists of an ordinary 400 
gallon tank with part of the top cut away, and having strong 
wooden stays bolted to the interior sides to prevent bulging. The 
water discharges itself from the main horizontally through a rose, 
and this, together with the series of baffle plates fixed across the 
tank, effectually checks any disturbance due to influx. Throughout 
the whole of the experiments nothing in the nature of an oscillation 
