GAUGING OF FLOW IN STKEAMS 349 



drift away to every portion of the stream ; snags and hollows in the 

 bed, bends, and falls, all produce some (irregular) disturbance of the 

 flow, with the result that the velocity at a point in the surface may vary 

 by 20 per cent., and at a point near the bed by as much as 50 per cent. 

 (Harlacher) in a short interval of time. 



In experiments on the St. Glair River (1899) the velocity-time curves 

 showed two sets of waves, small ones of 15 to 60 seconds amplitude and 

 larger ones of 3 to 6 minutes amplitude. The range of velocities as 

 found from the larger waves was in some cases 35 per cent, of the mean 

 velocity shown by the meter reading taken over ten minutes. These 

 experiments indicate that the pulsations are very limited in extent in a 

 direction at right angles to the current. The whole depth of the river 

 is affected by them, although their effect decreases from the bottom 

 towards the surface. 



It follows that a float, measuring as it does the velocity due to a single 

 pulsation, may give results which are greatly in error, and the only 

 chance of obtaining a fair estimate of the mean velocity over a single 

 section of the stream is to take the mean of a large number 40 or 50 

 of the values given by floats. The complexity of the motion is very 

 evident when floats are used. Of a series dropped into a stream at the 

 same point, no two will trace out the same path, and as may be well 

 understood when the multitude of observations necessary to give any 

 pretensions to accuracy is remembered, this method though at first sight 

 so simple, may easily prove the most expensive method of determining 

 the discharge. With current meters, on the other hand, the mean 

 velocity at any point may be obtained with great accuracy, provided the 

 period of observation is sufficient to cover a series of the pulsations of 

 velocity. Professor Unwin found that the mean time of successive 100 

 revolutions of such a meter in the Thames, when plotted, gave a very 

 irregular curve, while the mean times of successive 500 revolutions gave 

 an almost straight line. In general the time of a single observation should 

 not be less than five minutes, a period of six to ten minutes being 

 advisable. 



This renders it essential that in order to avoid spending an excessive 

 length of time in the field and thus running the risk of serious fluctua- 

 tions in the water level, the discharge be found from single observations 

 in comparatively few verticals, and that the ratio of the velocity at the 

 depth chosen, to the mean velocity, be known from vertical velocity 

 curves. This emphasises the importance of a thorough investigation 

 of the relation between velocity and depth in a vertical longitudinal 



