498 MESSES. LOCKYEE AND EOBEETS ON THE QUANTITATIVE 
Curve 2. 
This comprises a series of alloys ranging from 900 ’0 to 922-5 parts of gold in 1000 
of the alloy. It was obtained by measurements made by placing the micrometer-wires 
in a vertical position and measuring the base of the lowest visible portions of the spectro- 
scopic lines under examination, a method which appeared to be better suited to wide 
ranges than observations made in the upper portions of the spectral lines ; these latter 
were, however, found more suited to the study of narrow ranges, such as that shown in 
Curve 3. We should remark that, at this early stage of our inquiry, our attention was 
mainly devoted to ascertaining whether the curves would be continuous, and to deter- 
mining at what point a curve deduced from any given lines in the spectrum would cease 
to be sensitive. 
Curve 3. 
The Curve 3 was constructed by means of observations made upon several speci- 
mens of gold-copper alloy, the composition of which had been determined to the 
roioo'o P ai ’t in the ordinary manner. Micrometer-readings for these specimens were 
taken, which, together with their known approximate compositions, furnished the coor- 
dinates to the curve. 
It will be seen at a glance that with regard to the lower part of the curve a change 
of composition of ro^Joo gives a much greater change in the micrometer-readings than 
it does in the upper part. It is also to be remarked that, as in this more sensitive 
portion a change of 10 qq 0 part is represented by a change of 200 in the micrometer- 
readings, a true mean curve can only be constructed when the actual composition of the 
alloy is stated with much greater accuracy than that afforded by the present method 
of assay. 
We should observe that as in several cases duplicate and even triplicate readings 
were made with the same specimens, the number of spots placed opposite any 
figure in the vertical column does not necessarily indicate the number of pieces of that 
composition under examination. If we examine the Mint specimens, of which the 
composition is stated to be 916*4, it is seen that the maximum departure from the curve 
is less than 2 -qJtoo part ; and in this case it is impossible to say that any error has 
been made by the spectroscope, because, were the composition of all the alloys in 
question to be what the spectroscope states them to be, they would still be called 916-4 
on the parting assay. Again, if we consider the specimens 916-6, it will be seen that 
while in those two with micrometer-readings 150 the departure is not more than that 
previously stated, 2 o,oo o part, on the other hand the composition of the two the micro- 
meter-readings of which were 250 would carry them into a region of the curve which 
would spectroscopically indicate their composition to be less than 916-5. The departure 
in this case was so great that it was considered advisable to make a fresh determination 
of their composition by the ordinary method ; and it was satisfactory to find that the 
parting assay determination, as revised, was almost identical with that given by the 
spectroscope. 
