! 
requires a great deal of thought and knowledge to produce one, even though 
it consist of nothing but flowers. 
The first thing to be done at the commencement of a picture, is to con¬ 
sider what your materials consist of, and what effect you can make them 
produce : it is not often that a painter has all he wishes; ingenuity must 
often make up for the want of some, and the imperfection of others; and 
it is no small difficulty sometimes, when a strict attention to nature is re¬ 
quisite. But the best way to get over the difficulty, after you have made 
your drawing, and considered what parts you mean to come forward, and 
ijrhTr retire, is, to make a blot, of such colours as your subject allows 
you to introduce. If the whole range of colours is allowed you, dispose 
them as has been already mentioned, placing a prominent colour on a pro¬ 
minent part, and the retiring ones in their respective situations. If you 
have only part of the pure and brilliant colours at your disposal, then you 
must lessen the quantity of the other colours, rejecting the contrasting tints 
to those you leave out. If there be no pure yellow in it, there wants no pure 
purple; if no pure orange, then no pure blue is necessary: but if obliged to 
have pure blue, &c. in short, to form a picture out of discordant colours, 
then is the time to exert ingenuity, and by some means or other to har¬ 
monise them together, which is seldom impossible in works of fancy. 
I must now speak a little further upon the compound colours, lest I 
should be misunderstood in what I have said respecting their situation in a 
picture. We must first consider which are the compound colours; and it 
appears that the Prismatic Spectrum itself presents you with an alternate 
range of pure and compound* colours, by the blending of one with another ; 
* At least apparently so, though there are some who consider them all as elementary ones; 
whether they are so, or not, I cannot take upon me to say. But I cannot help considering violet as 
one, instead of indigo, and the latter as a compound; and I am inclined to think so the more, because 
a pure and a compound tint alternately are visible, in the prismatic range, as low as blue : and it 
does not seem probable, to me, that their order should change there, and for which there appears no 
other cause, but that indigo and red (it is said) will produce violet, if it may be called one, for it is 
many degrees inferior to the real one; whereas violet and blue will produce a much nearer imitation 
of the prismatic indigo, therefore, in that respect, I think there is stronger reason to consider it as a 
compound colour than violet. 
