SKdT. 41 



433 



10 II 12 13 14 15 



Fig. 13. Optical system of a-meter described by Kozlyaninov (1958). 



between the windows will, under these circumstances, cause a change in flux 

 distribution on the surface of the detector but, if the latter has uniform sensiti- 

 vity over its surface, or constant coupling, no error will result. 



The lamp must deliver a steady flux output during the measurement period 

 and should be run on regulated voltage. Voltage control of the light output is 

 not desirable because the flux output of the lamp is a function of the tempera- 

 ture of the container, which may be quite different in water than in air. Probably 

 the most satisfactory method for controlling the light output is to provide 

 another photocell near the lamp which monitors its flux output directly. 



The inherent errors associated with the measurement of a have been dis- 

 cussed by Wills and Jones (1953) and by Preisendorfer (1958c). The major 



Fig. 14. Compilation of data for the total attenuation coefficient, a. [After Dawson and 

 Hulburt (1934), Hulburt (1945) and Curcio and Petty (1951).] 



15 — s. I 



