3G2 Messrs. J. D. Fry and A. M. Tyndall on 



where P] stands for pitot pressures measured in thousandths 

 ot: a millimetre of water, P 2 squares of corresponding 

 parabolic pressures, and x the difference between them for five 

 different mean velocities. These are arranged in vertical 

 columns, each giving the values for a given position (?■) of 

 the pitot measured from the centre of the pipe along a radius 

 in millimetres. 



For the two lowest velocities, in which it will be seen from 

 Curve III that the flow is stream-line, the value of x is roughly 

 constant for all positions of the pitot and increases as the 

 mean rate of How through the pipe increases. This is true 

 also for r=15*6 cm. per sec. (if the reading at r = is 

 omitted). F v = 22'0 and # = 28*3 this does not appear to be 

 the case, although it may be so for positions close to side. 



Now Curve III shows that the static pressure velocity line 

 leaves the straight gradually, that is to say the transition from 

 stream-line to eddy motion is also gradual. It would appear, 

 therefore, that with increasing velocities of flow, eddy motion, 

 which must first appear at the centre of the pipe's section, 

 gradually spreads towards the walls. Consequently it is 

 reasonable to suppose that, at the velocity v = 15'6 cms. per 

 sec, stream-line motion existed in the pipe except near its 

 centre, but that at v — 22 and ?' = 28*3 eddy motion had 

 extended to a radius of at least 5 mms. This effect would 

 account for the low and sometimes negative readings of x in 

 that region. In these cases, therefore, we may assume that 

 the value of: x near the side is that which would have 

 held across the pipe had stream-line persisted. The average 

 values of x (x) for different velocities are given in the last 

 column, those values of x in brackets being omitted on the 

 eddy motion assumption in taking means. The relation 

 between x and the mean rate of flow of air through the pipe 

 is shown in Curve V, where the values of x are plotted as 

 ordinates and rates of flow as abscissa?. 



Considering the extreme smallness of x — the largest value 

 obtained being only just over a thousandth of a millimetre of 

 water — the results are fairly definite and show that ~x is 

 proportional to the mean rate of flow — at any rate at these 

 low velocities. 



The authors have not succeeded in obtaining a satisfactory 

 explanation of this phenomenon, but the following con- 

 siderations present themselves : — - 



1. The x effect could not be accounted for by an un- 

 commutated temperature effect such as is mentioned on 

 page 361 because the difference of pressure which such would 

 produce would, unlike x, be constant for all velocities of flow. 



