358 Messrs. J. D. Fry and A. M. Tyndall on 



the curve given by crosses in Curves I, and it appeared, 

 therefore, that assuming that there was a source of error in 

 the original work with this method, it had not been eliminated. 

 It being desirable to check these readings in some way, 

 attempts were made to obtain them at still lower mean 

 velocities when the motion in the pipe would no longer be 

 eddy bnt stream line. 



With stream-line motion the distribution of velocity across 

 a diameter is parabolic. Assuming then that K is constant 

 and of value l'OO (as obtained by the centrifugal method), 

 the curve of root pitot pressure across a diameter must also 

 be a parabola. The forms of these parabolas of root pressure 

 for various mean velocities (referred to in what follows as 

 " theoretical parabolas "") may be obtained by calculation and 

 compared with the experimental curves. 



Pitot pressure readings, however, for mean velocities 

 lower than 50 cms. per second could not be obtained with 

 any accuracy, because the motion of the gasometer caused an 

 irregular bending of the rubber connexions from the pipe to 

 the pressure-gauge, which at low velocities resulted in 

 fluctuations of the observed pressures comparable with the 

 readings themselves. In the hope of obtaining stream-line 

 flow for such velocities a one-inch pipe was substituted for 

 the two-inch, but it was found that the motion of the 

 gasometer was then too slow to be maintained regularly. 



The falling gasometer method was, therefore, abandoned, 

 and the stream of air obtained by fixing the pitot and a 

 surrounding one-inch pipe to the top of a 50-gallon tank and 

 gradually filling the latter with water from the mains. 



The rate of now of the air-stream was obtained from the 

 rate of change of the water-level in the tank, and this 

 remained fairly constant provided no water was being drawn 

 elsewhere in the building. The pitot pressures were obtained 

 by the change in gauge reading on commutation. The 

 absence of moving rubber and the more constant mean 

 stream velocity increased the working sensitiveness of the 

 gauge over fifty-fold, so that fairly accurate readings could 

 be obtained to as low a mean velocity as 6 cms. per second. 



It was not possible now to allow for the presence of water 

 vapour with certainty in calculating the density of the air in 

 an experiment, because it was probably not accurate to assume 

 that the air was saturated or that its humidity was that of 

 the atmosphere ; this, however, was a small correction and 

 neo-lioible compared with the discrepancy investigated. 



As it was necessary to know whether the motion was 

 really stream-line at these low velocities, the static pressure 



