fact that the peak of the experimental spectrum ( Fig. 1) falls midway between the single-and double-transition frequencies, and from previous results on the quadrupole-induced lines, 7 we expect the single and double transitions to contribute roughly equal amounts to the spectrum of Fig. 1. Assuming exactly equal contributions would give l$l=0.42e« 0 4 and l$'l = 0.33etf 0 3 . Finally, we mention other possible induction mechanisms, in addition to the hexadecapolar mechanism. The quadrupole-induced double transitions S 0 (D + S i( 3 )> Si(l) + S 0 (3), and S 0 (2) + S 1 (2), which fall within the linewidth of Fig. 1, can be ruled out. To do this the group of double-S transitions near 5300 cm" 1 (outside Fig. 1) were observed 8 and their intensity compared with the expected intensity of the double-S transitions in question; the latter is expected to be about 5% of the intensity of the lines near 5300 cm" 1 . Since the observed peak of Fig. 1 is about 50% of the intensity of the 5300 cm" 1 lines, we see that the double-S transition mechanism fails, by an order of magnitude, to explain the observed peak. Rough estimates of the dispersion and overlap mechanisms indicate that they too are small. The "level-mixing" mechanism of Herman 9 is estimated to be of importance here only if <£> is of order 0. lea 0 4 or less.

No abstract

No abstract