Page:Proceedings of the Royal Society of London Vol 69.djvu/172

off the upper and lower portions, utilizing only the light which passes in the median plane, where the change in density appears to be quite uniform, this method having been used in preparing the dispersion curve alluded to above. The effective angle lies probably somewhere between 90 and 130, but even this estimate is mere guess work.

If a single observation could be made with a vapour prism of known angle, in which we could be sure that the density was uniform, quantitative values could be assigned to determinations made with the dispersion tubes. I have accordingly tried in every way possible to obtain a prism fulfilling the required conditions. These endeavours have thus far met with only partial success, but an account of the devices tried may be of value to others working along similar lines.

Glass and mica are so quickly attacked by the sodium vapour that the use of these substances for prism faces seems to be out of the question. Added to this there is the difficulty of making gas-tight joints between plate-glass and metal which will stand a temperature but little below a red heat. Repeated failures to secure prisms in this way compelled me to seek some other method of giving the vapour the required form. I had observed that in glass tubes held before sodium flames the black vapour retreated before the approach of a cold obstacle. This suggested to me that it might be possible to do away with glass surfaces entirely, moulding the vapour into the required prismatic form by the proximity of cold bodies. Experi- ments along these lines were partially successful. Two pieces of thick walled iron tubing, the ends of which had been cut off at an angle, were introduced into a glass tube, and the sodium placed in the clear space between the bevelled ends, as is shown in fig. 5. The ends of the

FIG. 5.

tube were closed in the usual manner and the whole mounted between the collimator and telescope, in such a position that the prism formed by the sloping ends of the iron tubes stood with its refracting edge vertical. This was necessary, for in any other position the refraction due to the non-homogeneity of the vapour would have made itself felt. With a vertical slit and a vertical prism no lateral deviation could result from this cause. It was hoped that the sodium vapour formed by sudden and rapid heating of the glass tube would refuse to enter the colder iron tube, and that a prismatic form would result from the bevelled ends. The slit of the spectroscope was illuminated with monochromatic light obtained by prismatic analysis, the wave-