§z6 Mr. herschel’s AJlronomical Obfervations, See. 
to fo great an error as one tenth part of the whole, and 
the figure ab cd contains that part wherein the meafure 
rn being taken inftead of on, the error will be lefs than 
the 20th part of the whole meafure. In a portion fome- 
thing more confined the error will foon vanilh, fo that 
the difference may be fafely negledted intirely. Thus 
in the fpace ab xy the error cannot amount to a hun- 
dredth part. Thefe figures may be conftrudted by taking 
the feveral points D, d, y, and c, c, x, 26°, 18% 8°, re- 
fpedlively, from the vertex, the curves ad, a d, a y, bc, 
bc, B x, being the loci of thofe points of the tangents 
which touch the feveral ellipfes of illumination that may 
be contained in the femi-difk. of the Moon, when thefe 
t 
tangents make thofe feveral angles of 26°, 1 8°, 8°, with 
their fubtangents. 
1 
