144 Capt. W. de W. Abney. [June 



tain 



however, I have discovered that this law breaks down under cert 

 conditions, and I think the fact worthy the attention of those interested 

 in the subject, since it is possible that these conditions may arise 

 with other experimenters. Quite lately 1 have described the method 

 of comparing the photographic value of sunlight with that of candle 

 light ('Photographic Journal,' June, 1893), which was as follows: 

 A beam of sunlight, after three reflections from plain glass mirrors, 

 was admitted through a narrow slit to sensitive bromide paper 

 stretched round a drum of about 4 inches in diameter. The drum 

 could be caused to rotate round its axis at any speed up to about 

 sixty revolutions per second, by means of an electro-motor. A 

 small exposure with this light was given to the paper during the 

 rotation of the cylinder. Subsequently an amyl acetate lamp was placed 

 in position at any convenient distance from the same slit, and a fresh 

 portion of the same sensitive paper exposed to its action during a 

 much longer period, the rotation being continued as before. The slit 

 was next replaced by a small square aperture, of some -J inch side, and 

 further portions of the same paper exposed to the amyl acetate light 

 at the same distance, for varying but known exposures, with the drum 

 at rest. On development the paper showed three images, a narrow 

 band of deposit of the width of the slit caused by the sunlight, a 

 second band of the same width due to the light from the amyl 

 acetate lamp, and a third row of squares of varying blackness of 

 deposit due to the different exposures given with the drum at rest. 



If the width of the slit be accurately measured, the band formed 

 by the amyl acetate lamp is evidently superfluous, supposing the 

 usually accepted law to hold good under all circumstances, as by 

 measuring the blackness, or rather want of whiteness, of the different 

 squares, and using them as ordinates to the abscissae which were the 

 times of exposure, and drawing a curve through them, the blackness 

 produced by the sunlight could be referred to that produced by 

 the light of the amyl acetate lamp, and its equivalent value in 

 terms of the latter light be calculated. The band of deposit produced 

 by the amyl acetate lamp was introduced as a check, for its black- 

 ness could also be referred to the curve, and the width of the slit 

 be calculated from it. On making such calculations I was surprised 

 to find that in every case the calculated width of the slit was always 

 considerably less than what it was in reality, the difference being far 

 beyond that which would be caused by any error in the measure- 

 ment. This led me to commence an investigation into the cause of 

 this difference, and what has already been carried out is sufficient to 

 show that there is a failure in the usually accepted law. It may be 

 pointed out that if it held good the sum of any number of very short 

 exposures should be equivalent to a single exposure for the same 

 length of time. 



