Mr Herschel on the Deviations from Newton's Scale 
former case, is represented in Fig. 4. Its ordinate, as we see, 
decreases rapidly from the red to the yellow, where it is beyond 
the reach of the present observations, then increases again yet 
more rapidly, and is greatest of all for the violet rays. For the 
sake of comparison, Plate VII. Fig. 5. represents the curve si- 
milarly constructed for the ordinary variety, which has a maxi- 
mum where the variety last described has a minimum. The 
straight line r B inclined at 45 ° to the abscissa in all the figures 
represents the values of ~ for such crystals as follow Newton’s 
Scale in their tints. 
The apophyllite has furnished us then with no less than three 
instances of remarkable deviations from Newton’s Scale in crys- 
tals with one axis. It would certainly be in the highest degree 
interesting to subject them all three to chemical analysis, but as 
the total weight of* the specimen presenting these anomalies did 
not exceed 60 grains, of which nearly one-half consisted of the 
ordinary variety, I have not sufficient confidence in my own che- 
mical dexterity to enter on so very delicate an inquiry, which 
would obviously call for a degree of precision attainable only by 
consummate masters in the art of mineral analysis. It remains, 
therefore, to be ascertained, whether their different actions on 
light be owing to a difference in composition, or merely in their 
state of aggregation. Meanwhile, as we have seen that the 
union of two crystals differing in their scale of tints produces a 
scale differing from either, it may not be irrelevant to inquire, 
whether the alternation of laminae of two of the varieties 
above described, may not be capable of producing the remain- 
ing one. 
To this end, let t'\ &c. be the thickness of the 1st, 2d, 
8d, &c. lamina so superimposed as to have their axes coincident, 
and of the same refractive density : /, /", &e. the minimum 
lengths of the periods susceptible of being performed by a ray 
of any colour within these several crystallized plates, and 6 the 
angle with the axis, at which a similar ray traverses the system. 
Then, as M. Biot has proved, the number (n) of periods, and 
parts of a period, actually performed by this ray during its pas- 
sage through the first plate, is given by the formula, 
