July 15, 1910] 



SCIENCE 



93 



The same method may be available as an 

 adjunct to either Jamin's or Michelson's in- 

 terferometers, except that here the transmitted- 

 diffracted and reflected-diffracted rays are 

 brought to interfere. To take the example of 

 the Michelson type stripped of unnecessary 

 details, let gGg' in the figure, be the grating 

 or ruled surface, n its normal, L the source of 

 white light, M and M the mirrors, and E the 

 eye. In the usual way the rays from L inter- 

 fere at E. 



Now replace i by a slit and collimator, E 

 by a telescope focussed for parallel rays. The 

 eye at E now sees a sharp line of light. At 

 D and D', however, there must be two diffrac- 

 tion spectra coinciding in all their parts and 

 hence interfering rhythmically if all adjust- 

 ments are sufficiently perfected. The other 

 two diffractions within MGg' and EGg are 

 often lost at an incidence of 45°. 



The attempt to produce these interferences 

 D, B', with replica gratings is liable to result 

 in failure: for while the transmitted system 

 NGD shows brilliant spectra, the reflected 

 system MGE is dull and hazy. Both spectra 

 are clearly in evidence and may be brought to 

 overlap. The film, however, does not reflect 

 in a degree adequate to the transmission. 

 Attempts are in progress to realize the condi- 

 tion of equality with a grating actually ruled 

 on glass or possibly with a modified film. 



What is strikingly feasible, however, with 

 ordinary plate glass and a non-silvered grating, 

 is the production of interferences between 

 pairs of diffracted spectra, D/ and D^, for 

 instance, if returned by equidistant mirrors 

 M^ and N^ to a telescope in the line D. Both 

 of these spectra are very brilliant and not very 

 unequally so, and the coincidence of spectrum 

 lines brings out the phenomenon. This is of 

 the ring type, and not of the line type referred 

 to above; but it also occupies the whole field 

 of the spectrum from red to violet. In my 

 first adjustment using sunlighl, I obtained 

 splendid large confoeal ellipses, with the dark 

 centers in the yellow, the sodium line simul- 

 taneously in focus serving as a major axis. 

 It is more usual, however, to obtain oblique 

 lines across the spectrum which are strongest 



in certain color fields. In a city laboratory 

 these are perpetually in motion, the rings 

 particularly alternating between dark and 

 bright centers. Naturally a fine slit is of 

 advantage. The theory of these ellipses will 

 be given elsewhere. 



Gael Barus 



notes on an experiment concerning the 

 nature op unit characters 



Some time ago the writer planned' a series 

 of experiments designed to throw some light 

 on the nature of unit characters. Only one 

 part, of which the following is a brief extract, 

 has been completed. If an apology is neces- 

 sary for daring to present negative results, I 

 might say that even if proof of a negative is 

 logically impossible, such evidence does give 

 an idea of the relative frequency of the occur- 

 rence of the event in question. It is some- 

 times forgotten that a small probable error is 

 as desirable in this case as when the results 

 are positive. In addition to this fact, how- ,;. 

 ever, it is a pleasure to call attention to a line 

 of experimentation which, though familiar to 

 all biologists, has not had the serious consid- 

 eration that it deserves. I mean the work of 

 MacDougal in trying to produce mutations or 

 transmissible variations by artificial means. 

 Even if one does not accept as fact that the 

 definite and transmissible changes which have 

 occurred in Dr. MacDougal's injection experi- 

 ments were caused directly by the introduction 

 of semi-toxic solutions into the mother plant's 

 ovary, he should admit that the method pro- 

 posed is well worth his earnest attention. It 

 is capable of several modifications and exten- 

 sions — two of which I shall describe — which if 

 given sufficient trial might yield results with 

 important bearings somewhat apart from the 

 original scope of MacDougal's investigations. 

 Even if many experiments on limited popula- 

 tions should give no positive results, it should 

 be remembered that progressive variation oc- 

 curs but rarely in nature, possibly but one 

 variant in millions of individuals. One ought 



^ At the Connecticut Agricultural Experiment 

 Station, under the federal appropriation known as 

 the Adams fund. 



