THE NITRATES AND PHOSPHATES. 123 



to bands of great intensity the visual observations were found to be 

 best. The results given for the fluorescence spectra of other hydrates 

 and for the anhydrous salt are based upon visual observations exclu- 

 sively. 



THE HEXAHYDRATE: U0 2 (N0 3 ) 2 +GH 2 0. 



The hexahydrate crystallizes in the rhombic system with the axial 

 ratio a: b: c = 0.6837 : 1 : 0.6088. The crystals were grown in the form of 

 plates by using a water solution whose depth was equal to the thick- 

 ness of the plate desired. Single crystals as large as 15 mm. in diameter 

 were obtained with relatively little difficulty. All of the results here 

 discussed are based upon observations made with single crystals. 



In selecting the data to be used in taking a final average, each nega- 

 tive was carefully studied and measurements that seemed for any 

 reason doubtful were discarded. The elimination of doubtful observa- 

 tions was made without reference to the agreement or lack of agree- 

 ment between the different measurements, and was, in fact, completed 

 before the measurements of the different negatives were compared. 

 About 40 negatives were used, although the number for any one line 

 was rarely more than 10. 



The errors of calibration of the spectrograph and spectrometer can 

 hardly exceed 1 A. u., except perhaps in the extreme red end of the 

 spectrum. The uncertainties due to the faintness of certain bands, to 

 their finite width, and to photographic broadening are more difficult to 

 estimate and undoubtedly differ greatly with the character of the band 

 and its position in the spectrum. In the case of the sharper bands of 

 moderate intensity we feel that the averages that are here tabulated 

 are reliable within 1 A. u. In other words, the reciprocal wave-lengths 

 are accurate to within about 0.02 per cent. For the faint or hazy bands 

 the possible error is undoubtedly much greater. 



Of the 55 fluorescence bands observed, 46 can be arranged in 9 series, 

 as tabulated below, the frequency interval being nearly constant in 

 each series. Two of the remaining bands have the same interval, and 

 apparently form part of a series whose other members were too weak 

 to detect. The 7 bands that do not fall in any series arrangement are 

 all extremely weak, and since in most cases they are recorded only once, 

 their existence is subject to considerable doubt. Estimates are given 

 in table 40 of the intensities of the different bands and of the reliability 

 of the measurements. In some cases the series seem to extend into the 

 region of absorption, and in such cases the absorption bands that seem 

 to form part of the series are also given. 



The data for series B, D, E, and F, which are made up of the stronger 

 bands and those of medium intensity, are undoubtedly the most reli- 

 able. The values of the average interval between bands in these series 

 are 86.0, 85.8, 85.9, and 86.1 respectively. In taking these averages, 



