§ XII. 2. Inlet windows. These windows may, of course, be of any shape and size whatever, according to the other necessities of the building, and the quantity and direction of light desired, their purpose being now to throw it in streams on particular lines or spots; now to diffuse it everywhere; sometimes to introduce it in broad masses, tempered in strength, as in the cathedral colored window; sometimes in starry showers of scattered brilliancy, like the apertures in the roof of an Arabian bath; perhaps the most beautiful of all forms being the rose, which has in it the unity of both characters, and sympathy with that of the source of light itself. It is noticeable, however, that while both the circle and pointed oval are beautiful window forms, it would be very painful to cut either of them in half and connect them by vertical lines, as in Fig. XLIV. The reason is, I believe, that so treated, the upper arch is not considered as connected with the lower, and forming an entire figure, but as the ordinary arch roof of the aperture, and the lower arch as an arch _floor_, equally unnecessary and unnatural. Also, the elliptical oval is generally an unsatisfactory form, because it gives the idea of useless trouble in building it, though it occurs quaintly and pleasantly in the former windows of France: I believe it is also objectionable because it has an indeterminate, slippery look, like that of a bubble rising through a fluid. It, and all elongated forms, are still more objectionable placed horizontally, because this is the weakest position they can structurally have; that is to say, less light is admitted, with greater loss of strength to the building, than by any other form. If admissible anywhere, it is for the sake of variety at the top of the building, as the flat parallelogram sometimes not ungracefully in Italian Renaissance. [Illustration: Fig. XLIV.]