Royal Society. 141 
If an interpolation of the second degree proves to be insufli- 
cient for representing the observations, the series of values given 
is divided into several intervals, and each of these is then treated 
in the manner described. 
In some cases it will be advisable to represent, not y, but some 
function of y like y”, logy, &c., on a scale with equal divisions. 
The advantages which the method of graphical interpolation 
described appears to me to possess, as compared with the usual 
one of drawing a curve in a rectangular system of coordinates, 
are the following :— 
1. All the lines drawn are straight lines; the personal error in 
the drawing is therefore reduced to a minimum. 
2. The drawing can be executed with less trouble and greater 
exactitude, and it takes up less space than in the ordinary way. 
3. When the drawing is finished, the value of y belonging to 
any given z may be read off at once, and vice versd. 
XXIII. Proceedings of Learned Societies. 
ROYAL SOCIETY, 
{Continued from p. 79.]| 
March 22, 1860.—Sir Benjamin C. Brodie, Bart., Pres., in the Chair. 
HE following communications were read :— 
** Qn the Theory of Compound Colours, and the Relations of 
the Colours of the Spectrum.” By J. Clerk Maxwell, Esq., 
Professor of Natural Philosophy, Marischal College and University, 
Aberdeen. 
Newton (in his ‘ Optics,’ Book I. part ui." prop. 6) has indicated a 
method of exhibiting the relations of colour, and of calculating the 
effects of any mixture of colours. He conceives the colours of the 
_ Spectrum arranged in the circumference of a circle, and the circle so 
painted that every radius exhibits a gradation of colour, from some 
pure colour ef the spectrum at the circumference, to neutral tint at 
the centre. The resultant of any mixture of colours is then found 
by placing at the points corresponding to these colours, weights 
proportional to their intensities; then the resultant colour will be 
found at the centre of gravity, and its intensity will be the sum of 
the intensities of the components. 
From the mathematical development of the theory of Newton’s 
diagram, it appears that if the positions of any three colours be 
assumed on the diagram, and certain intensities of these adopted as 
units, then the position of every other colour may be laid down from 
its observed relation to these three. Hence Newton’s assumption 
that the colours of the spectrum are disposed in a certain manner in 
the circumference of a circle, unless confirmed by experiment, must 
be regarded as merely a rough conjecture, intended as an illustration 
of his method, but not asserted as mathematically exact. From the 
