136 



Mr. H. Hartridge. Factors Affecting the [Nov. 1, 



the accuracy of measurement, either of the mean wave-length or of the 

 apparent boundaries of bands by the cross-wire and micrometer. Where 

 such a method is alone available, the use of a broken line is indicated. 



Fig. 3. — To show the Effects of Contrast on the Appearance of Absorption Bands. 



A, normal density curve ; B, curve altered by contrast ; C, effect of breaking continuity 

 of density gradient by the line of the micrometer; D, suggested use of "broken" 

 line in the micrometer. 



A brief theoretical consideration may now be given of the manner in which 

 threshold and intensity of incident light control the appearance of absorption 

 bands. Any desired value may be given to the density at any point of the 

 spectrum, by causing the incident light, on transmission, to encounter the 

 correct number of pigment molecules. Any part of a spectrum may 

 therefore be made to fall below the value of the threshold of vision, and, 

 when such is the case, these portions must give to the eye the appearance of 

 uniform black areas having no visible luminosity. This explains why, by 

 starting with a dilute solution of such a pigment as haemoglobin, and 

 gradually increasing either its concentration or thickness, a point is reached 

 at which a narrow black area appears in the centre of the a-band, and, 

 further, why this area increases in width on either side as the concentration 

 or thickness is further increased, until fusion of neighbouring areas ultimately 

 takes place. Similar changes may clearly be brought about by altering 

 either the initial intensity of the incident light or the threshold. In fig. 4 

 these changes are shown diagrammatically. 



Such considerations explain the existence of certain conditions upon which 

 the accuracy of the estimation of CO in blood is found to depend. As much 

 of the bands as possible should be below the threshold, without the conditions 

 being such as to impair the luminosity of the area separating the « and /3 

 components. These optimum conditions are always aimed at when making 

 CO estimations ; their existence is of great practical importance, and is well 

 shown in nV 5. 



