534 SECTIONAL TRANSACTIONS.— A. 



Experiments have been made using wires of different diameters, and these have 

 given identical values for the temperature of the flame investigated within the limits 

 of experimental error. 



In the second method a beam of light from an incandescent tungsten sphere was 

 focussed through the flame on to the slit of a spectroscope. Sodium was introduced 

 into the flame, and when the temperature of the flame was greater than that of the 

 tungsten sphere, bright sodium lines showed up on a continuous spectrum background. 

 If the temperature were lower, reversal of the sodium lines took place. By careful 

 adjustment of the temperature of the tungsten sphere a point was reached when all 

 trace of either bright or dark lines disappeared. This balance could be effected 

 within a range of a few degrees. The corresponding temperature of the sphere was 

 then determined by means of an optical pyrometer. 



The first question studied was the influence of flame thickness, and it was found 

 that the method gave results independent of the thickness of the flame. The second 

 series of experiments was directed towards answering the question whether the 

 temperature obtained in the spectrum-line reversal method was independent of the 

 particular spectrum line employed. For this purpose the red lithium line and the 

 yellow sodium lines were utilised. To obtain satisfactory reversals with the lithium 

 line a thick flame from a water-cooled burner 10 inches in length was used. Separate 

 experiments were made in which sodium and lithium salts were sprayed in the gas 

 supply to the burner, and the red and yellow lines were also obtained simultaneously 

 by using a solution of both salts in the spray. It was found that the temperatures 

 obtained were the same whether the red or the yellow lines were employed as 

 indicators. It may here be remarked that the lines used were principal lines of a 

 spectral series, and the conclusion may be applicable to these lines alone. 



The third series of observations was made on heterogeneous flames. Two flames 

 at different temperatures were measured individually, and then superimposed and 

 again measured. A mathematical investigation was also made of the problem and a 

 formula deduced for the ' apparent ' temperature of two superimposed flames from 

 the individual values. A comparison of the observed and calculated data showed 

 satisfactory agreement. It may be remarked that the value obtained for the 

 ' apparent ' temperature of two superimposed flames differs markedly from the 

 mean of the two individual temperatures and is dependent on the relative positions 

 of the flames. It would therefore appear that the method cannot be applied to 

 estimate the average temperature of a heterogeneous flame. 



A noteworthy feature of this method of measuring temperature is that it is devoid 

 of ' time lag,' and so, for example, may prove of value in measuring temperatures in 

 the explosion cycle of an internal combustion engine. 



Some preliminary experiments have been carried out in conjunction with Mx. 

 Fenning of the Engineering Department N.P.L., to test this appUcation of the method, 

 and gave quite promising results. The ' match ' could be adjusted by successive 

 trials in a series of five or six explosions. 



Dr. J. Jackson. — Free Pendulum Clocks. 



The clocks Shortt 3 and Shortt 11 were installed at the Royal Observatory, 

 Greenwich, in November 1924 and May 1926 respectively. Each clock consists of a 

 ' free pendulum ' of invar swinging in a vacuum (1 inch of mercury) and a slave clock. 

 The slave clock does all the work, including the release of the gravity lever which 

 maintains the free pendulum. The only interference with the latter is for part of a 

 second every 30 seconds. The slave is controlled by a mechanism brought into 

 action when the impulse is ended, which takes place at a definite phase of the free 

 pendulum. 



The accuracy of such clocks greatly exceeds that of earlier types. The principal 

 irregularities, shown by both clocks, is a temperature coefficient of 0-003 seconds per 

 day per 1° F., and a gradual slowing down of the pendulum attributed to a growth of 

 the invar rod. The growth for the 2-seconds pendulum is about 1 micron (ycmttj mm.) 

 in 120 days, producing a decrease in the daily rate of -037 seconds in 100 days. The 

 growth of the invar of Shortt 3 appears to have been nearly constant since October 1925. 



These clocks greatly improve the accuracy with which time signals can be sent, 

 especially when cloudy weather prevents astronomical observation, but it appears 

 that they are not sufficiently accurate to check small irregularities in the earth's 

 rotation. 



