Research on some Standards of Light. 



495 



The definition of the standard quantity of light is that emitted in a 

 normal direction by 1 square centimetre of surface of platinum at its 

 temperature of solidification. 



It has already been pointed out : 



1. That it is practically impossible to obtain an instantaneous 

 photometric observation. 



2. That each part of the surface is, strictly speaking, only at the 

 standard temperature for the small time-interval during which the 

 solid film is actually forming. 



3. That owing to the difficulties inherent to all photometric observa- 

 tions, each determination should rest on a number of separate 

 readings. 



The most obvious method would be to take the mean of the readings 

 obtained during the " constant period," but in doing so we should 

 certainly be wrong, as these readings form a decreasing series, the first 

 one only being theoretically correct. The rate of decrease of the readings 

 varies with the mass of metal used, and the circumstances under which 

 it is allowed to cool ; and were we to take the mean of the readings, 

 the final determination would necessarily be dependent on these experi- 

 mental conditions. The first reading, though theoretically the most 

 correct, is practically the least reliable of the series. This reading 

 immediately follows a sudden and considerable change in the illumina- 

 tion of the photometer, and the eye has not yet had time to become 

 accustomed to the change. It is therefore out of the question to base 

 our determinations on this first reading. 



All things considered, it is best to establish the value of the inten- 

 sity of the light emitted by the platinum at the instant of solidification 

 by a graphical exterpolation. 



The shape of the curve during the period of solidification is, prac- 

 tically, a straight line. The various observations during this period 

 differ but little in absolute value, and can easily be plotted to a very 

 large scale. If we draw a straight line through these points the 

 height of this line above the axis at the time at which the " flash " 

 occurred will give us the quantity of light emitted at the temperature 

 of solidification. The values of the observations themselves determine 

 within ten seconds the time at which the " flash " occurred, and for a 

 first approximation this is sufficient. 



The above method of reducing the results is illustrated in fig. 10. 

 After a set of observations has been taken, and while the platinum is 

 still white-hot, the surface visible through the diaphragm should be 

 inspected. Should any specks of lime dust or other impurity be appa- 

 rent, the observations should be rejected. 



It is advisable to make it a rule to discard any set of readings 

 where : 



1. The " constant period " extends over less than thirty seconds. 

 VOL. lxv, 2 



