252 PROCEEDINGS OF THE AMERICAN ACADEMY. 



plicate tilings, and I have drawn attention to them merely to show that 

 we must not expect to explain matters by too simple a mechanism. 



A word or two about the bismuth arc may not be out of place. 

 Various plans were tried, such as immersion of the electrodes in water, 

 burning in the carbon arc, etc. The best arrangement was found to be 

 a shallow iron dish about 4 cms. in diameter (pounded from a piece 

 of thin sheet iron), filled nearly full of molten bismuth, and kept hot 

 over a small burner. The dish of metal formed the positive electrode, 

 the negative being a bar of iron which could be raised or lowered by 

 a rack and pinion. The arc required constant attention, fresh metal 

 being put into the dish every ten or fifteen minutes, and as exposures 

 of eight hours were necessary, it will be seen that an enormous amount 

 of very fatiguing work was necessary in all cases where open-air arcs 

 were used. 



Copper Excitation. 



I have been unable thus far to obtain photographs of the fluorescence 

 excited by the separated copper radiations. The lines are close together, 

 and the arc climbs about over the electrodes. I hope next year to im- 

 prove matters in this respect. On Plate 2, a, is seen the fluorescence 

 excited by the total copper radiation. Only the three green lines are 

 operative in stimulating the vapor. The lines in the fluorescence spec- 

 trum appear to bear no very definite relation to the lines of the magnetic 

 spectrum, as will be seen by the chart. There are many coincidences, 

 however, with lines in the spectrum excited by zinc 4811, and by zinc 

 468 and 472. By comparing the three spectra I have made a provi- 

 sional determination of the lines which belong together in the spectrum 

 excited by the copper radiation. These lines are indicated by crosses 

 and vertical dashes placed above them ; other lines, which do not appear 

 in the zinc spectra, have not been marked. I suspect that excitation 

 with copper 5152 will produce a doublet at this point, and probably at 

 other points, j ust as does cadmium 5086. An attempt will be made 

 to verify this surmise. I have on one or two occasions, when trying to 

 stimulate the vapor with this isolated line, been of the impression that 

 I saw doublets distinctly, but at the time I attributed it to incomplete 

 separation of the exciting lines. 



Lead Excitation. 



A lead arc, operated in a manner similar to the one described for 

 bismuth, was used for exciting the vapor. The only line operative was 

 the one at 5001, and it gave rise to a well-marked series of fluorescence 

 lines, which coincided exactly with the first series of the magnetic rota- 



