SCIENCE. 



LN. S. Vol. XVII. No. 439. 



ally widened the colors rapidly lose their pure 

 hues, the change at the central portion being 

 most marked at iirst. Finally, if the slit be 

 made wide enough, the color entirely disap- 

 pears from the middle part, while at one end 

 (accepting Dr. Kirschmann's description) 

 there is a band of red, orange and yellow; at 

 the other, one of blue and blue-green. The 

 explanation of this is well known. A wide 

 slit may be considered as the sum of a great 

 number of narrow ones, each of which gives 

 rise to a pure spectrum, but these spectra are 

 superposed, producing perfect white in the 

 middle and the colors mentioned at the ends. 



Dr. Kirschmann explains the absence of 

 green proper by saying that its two neighbors, 

 blue and yellow, are here separated, and so 

 are deprived of the power of cooperation, 

 just as in the ordinary spectrum red and blue 

 are separated, and thus can not produce 

 purple. But the nature of the mixture is 

 very different in the two -cases, and even 

 though we should grant that the sensation of 

 green is due to the superposition of blue and 

 yellow, we should hardly be justified in con- 

 cluding that purple should be considered as 

 simple a color as green, since blue and yellow 

 have wave-lengths nearly equal, while the 

 wave-length of red is approximately twice that 

 of blue. We hnoiv, from physical considera- 

 tions, that purple is not simple like spectral 

 green. 



From the explanation given "above it would 

 appear that the ' inverted spectrum ' is far 

 from being a pure one, though Dr. Kirsch- 

 mann thinks that this statement can be 

 proved unfounded. When we examine with 

 a spectroscope the light reflected from a very 

 thin sheet of mica, we see the spectrum crossed 

 transversely by a number of dark bands. This 

 phenomenon is one of the large interference- 

 family of ' colors of thin plates,' and is ordi- 

 narily known as the 'channeled spectrum.' 

 Many investigations have been made on it. 

 Dr. Kirschmann states that he was able to 

 obtain these 'channels' in the 'inverted 

 spectrum.' For the production of such, how- 

 ever, the spectrum need not by any means be 

 pure. In a single experiment with a direct- 

 vision spectroscope and a sheet of mica about 



1/100 of a millimeter thick; I was able to see 

 the * channels ' when the slit was 0.6 milli- 

 meter wide, while to show the most prominent 

 Fraunhofer dark lines, and thus have a slight 

 approach to purity, the slit had to be less than 

 0.25 millimeter wide. 



To use the channeled spectrum for the 

 purpose of measuring wave-lengths, as sug- 

 gested, is not very convenient, since the 

 thickness of the thin plate, its index of re- 

 fraction, the angle of incidence (or of refrac- 

 tion), as well as the 'order' of the interfer- 

 ence, would all have to be determined. But 

 if som,e wave-lengths are known, others may 

 be conveniently located by this means. A 

 very elegant application of this method was 

 made by Maxwell* in his classical experi- 

 ments on the mixing of spectral colors. His 

 thin plate was a layer of air between two 

 plane plates of glass, and by the channels in 

 the spectral image shown at the end of hia 

 color-box he was able to calibrate in wave- 

 lengths an arbitrary scale put across it. 



C. A. Chant. 



Department of Physics, 

 Universitt of Toronto. 



SURFACE tension; MOLECULAR FORCES. 



In deducing the surface tension equations 

 by the method of Laplace we start with the 

 assumption that the force with which one 

 element, dv, of the liquid attracts another 

 S element, v, is 



--1 



- p" ■ dv ■ fc '/(r) 



(usually the h is wrongly omitted), where p 

 is the efiective density of the liquid, and f (r) 

 is the law of the variation of the force with 

 the distance. Finally we find that the sur- 

 face tension is 



where 7 is a definite integral (and hence a 

 constant) derived from fir). In measuring 

 the surface tension of liquids we are usually 

 content to stop when we have found T, or we 

 endeavor to find relations between the values 

 of T for different liquids. We can do much 

 *Phil. Tram, of B. 8., 1860. 'Scientific 

 Papers,'' Vol. I., p. 410, § 6. 



