ON THE ATOMIC GROUPING IN ORGANIC BODIES. 
891 
Photographic arrangemen ts. 
Exposures. —The exposures were arranged in accordance with what experience 
taught us was the best time for different lengths of tube and differing liquids. 
In every case photographs were taken of each compound with varying lengths of 
exposure, in order to ascertain which gave the most trustworthy and measurable 
results. On the one hand, it was necessary to show the detail in those parts which 
had no general absorption but in which there was special absorptions ; and, on the 
other hand, it was necessary to show the special absorptions in those regions 
which had general absorption superposed. These details were often obtained by 
means of one or more photographs. It had also to be remembered that the silver 
salt employed, though sensitive, it is believed, to the extreme theoretical limit of 
the prismatic spectrum, yet has one maximum of sensitiveness, situated a little below 
the limit of the visible spectrum, from which point the “ curve of sensitiveness ” 
gradually falls away on each side. Thus, with an unabsorbed continuous spectrum 
it would require about three times more exposure for X 12,000 to impress the film to 
develop to a standard density than it would for X 8000. This difficulty is, however, 
met with at the other end of the spectrum, where the usual photographic plates are 
employed, and is necessarily present in all spectrum photography. 
Precautions to he observed. —There are a few cautions to give to other observers— 
first, to ascertain that the floor of the laboratory is free from vibration ; second, to get 
an accurate focus of the locality near X 900 ; third, to set the prisms at an angle of 
minimum deviation for that locality, since in that region the greatest variations in the 
absorption phenomena are to be found ; fourth, to use a good reference spectrum, such 
as that of chloroform ; and, fifth, to see that the tubes are accurately centered with the 
axis of the collimating tube. 
Mapping the results. 
In order to map the results in wavedengths, photographs of the solar spectrum 
were made with the same spectroscope as that employed in the research, and the 
Fraunhofer lines accurately measured and laid down to scale. The wave-lengths, 
also to scale, were erected as ordinates and a curve constructed. The value for 
every 1 0 in the wave-length scale was then found on the curve and fresh ordinates 
plotted from these points on the curve, and this was our “ prismatic scale.” 
When plotting an absorption spectrum the distance between two known points was 
noted off on a convenient ivory scale, and this applied to the ordinates of the prismatic 
scale at the known wave-lengths. All intermediate points could then be immediately 
read off in wave-lenoths. The wave-lengths between 900 and 1000 we consider to 
be accurate to within one or two, and beyond that point to within three or four; the 
shorter wave-lengths are accurate within one, we believe. 
5 y 2 
