﻿Measurement of 



calculations are very quickly carried out by the use of a 

 table of squares, Crelle's Multiplication Tables and a calcu- 

 lating machine ; a 150-entry table can easily be computed 

 in an hour. Calculations should be entered with one signi- 

 ficant figure more than is to be employed finally, and first 

 differences should be tabulated for convenience in interpola- 

 tion and as a check on the accuracy of the work. 



If the anemometer is to be employed in the measurement 

 of very low velocities (of the order of 10 cm. /sec. or less),. 

 the disturbing effect of the free convection current set up by 

 the heated wire may become sensible. From data derived in 

 the course of the experiments by the writer already referred 

 to, the " effective velocity " of the free convection current 

 set up by a 3-mil wire at 1000° C. is estimated at about 

 15 cm. /sec, is only reduced to about 8 cm. /sec. at 200° C, and 

 does not diminish materially with the diameter of the wire*. 

 The effect of the free convection current on the determination 

 of velocity from formula (1) at low velocities would demand a 

 separate investigation. Some information on this point can 

 be derived from the experiment described under fig. 3, in 

 which the distribution of the flow of air into a slit in a plane 

 is measured and compared with the distribution calculated 

 from theory. The evidence there discussed points to the 

 fact that the anemometer registers velocities as low as 

 15 cm./sec. with an accuracy of about ten per cent. For 

 use in low-velocity measurement it is, however, more satis- 

 factory to calibrate the anemometer under such conditions 

 that the air-velocity makes the same direction with the 

 vertical as in the experiment in which the instrument is to- 

 be employed. 



The linear relation expressed by formula (1) has been 

 tested experimentally for velocities as low as 17 cm./sec. t, and 

 was found to hold good within limits of experimental erroiv 

 Theoreticallv the linear relation mentioned is the asymptote 

 to a transcendental curve expressing the true relation 

 between heat-loss and velocity : it is shown, however, that to 

 an accuracv of 2\ per cent, a linear formula of the type (1) 

 may theoretically be employed when the velocity is as low 

 as "that given by the relation Yd=l'Sl x 10~ 2 . Y being 

 expressed in cm./sec, and the diameter d in cm. J For a 

 2^-mil wire this limiting velocity is as low as Y=2'9 cm./sec, 

 probably much lower than the lower limit imposed by the 

 disturbing effect of the free convection current. 



* Reference (1), Table V1IL, p. 424. 



t Reference (1), Table III., p. 416. 



% Reference (1). Description of Diagram I., p. 426. 



