412 
SIR W. CROOKES ON THE PHOTOGRAPHIC SPECTRA OF METEORITES. 
A 7000 A.U. ; the two main aims in the whole construction were, first, that the 
definition of the lines should be pushed to the highest stage of perfection; secondly, 
that the mechanical construction should be such that the adjustments once made any 
number of spectra could be obtained without uncertainty in the excellence in definition. 
When I commenced to build the instrument, I was greatly assisted by the advice and 
suggestions of the late Victor Schumann. Although the instrument is very different 
in design to any constructed by him, still the experience he had gained in his researches 
upon the spectroscopy of the ultra-violet region, which he freely placed at my disposal, 
was of the greatest possible value. 
Having decided upon the general plan, the instrument was built up, first temporarily 
in wood, afterwards permanently in iron and gun-metal—and set to work. This was 
in February, 1899, and since then it has been in continual use ; some thousands of 
photographs having been made. During this period each detail of the instrument 
has been subject to a process of evolution and reconstruction, each step leading to 
some advantage demanded by actual work. 
The parts of the instrument are all assembled upon a large truly surfaced cast-iron 
table measuring 31x51 inches ; the dispersion is given by five excellent double prisms 
made of right and left handed quartz, each double prism consisting of two halves 
made respectively of left and right handed quartz, prisms such as were first used by 
Cornu to avoid double refraction. 
The objective and collimating lenses are simple planoconvex, 40 mm. in diameter 
and 700 mm. focus. The spectrum is thrown upon a celluloid film so curved that all 
the lines are at their true focus. 
The dispersion by the five prisms is considerable, and with the collimating and object 
lenses used would involve a spectrum of some 24 inches in length. As there are both 
mechanical and optical reasons against the use of such a large sized plate it was 
decided to “build up” the complete spectrum in eight successive plates, involving a 
length of 3 inches for each step. The reason for this decision is simple ; to obtain 
perfect definition of any line it must be in or very near to the “ optical centre ” 
of the instrument, i.e., the imaginary line passing through the slit, the centre of the 
two lenses, the centres of the prisms, and the centre of the curvature of the plate : 
and it was found by many experimental photographs that the best definition over each 
three-inch film was obtained when the least refrangible line shown was in the position 
of minimum deviation. The many devices that from time to time have appeared in 
spectroscopic construction for rotating the prisms by link motion were rejected as 
uncertain and complicated—experiment having shown that the excellence of definition 
was maintained within a range of several inches on either side of the optical centre 
and position of minimum deviation. The exact “ data ’ involving the focus, position 
of the spark, angle of the plate, and position of prisms and lenses to give the finest 
definition of the photographed lines were found experimentally, and when found their 
positions were engraved upon the instrument and recorded so that for future work 
