otherwise occurring between the completely ornate and completely plain member. The decoration is applied under each triglyph, because the same number and quantity of horizontal lines that suffice to support and bound the metope sculpture would not suffice to stop and contrast these groups of strong vertical lines. The principle is exactly that which led the Italians, whenever they had a string-course serving as sill to a tier of windows, always to attach something, it hardly matters what, under each window or each window-jamb. The sub-triglyphs are simply repetitions (with less projection) of the mutules; and this repetition serves more than anything else, except the cap and necking repetition, to give unity of style.

Descending to the column, we must observe that the profile obtained, as already explained by the generalized imitation of limbs, though perfectly proper for the support of a plane extending in every direction from the capital, (as a flat ceiling,) requires an addition to fit it for placing under a beam that extends on only two sides of it. Unless the architrave were as wide as the echinus (which would render the whole top-heavy), it would not press on the whole of that member which is essential to preserve the analogy with an animal extremity. The abacus, then, presents the simplest possible way of spreading this pressure over the whole capital, and its thickness is regulated by what is found by experience just to give the expression of sufficiency to this purpose. If too thin, it is apparently useless, and if too thick, unnecessarily massive.

The shaft, as already noticed, though required by convenience to be round, is, nevertheless, made to present square (right-angled) edges. Nothing could be so contradictory in principle, to everything else in the Doric order, as the sleek fatness of a completely rounded shaft, whose mass only gives it clumsiness without the slightest expression of power. Dorian entirely debarred from the use of flutings would have made his columns square, at whatever sacrifice of con

A

venience. The first improvement on the square would be by truncating its angles, to reduce it first to an octangular, and then to a 16-sided prism. But the contrast between two successive sides of this being very slight, and liable to be counteracted by the faintest weather-stain, this contrast was exaggerated to the utmost, by so hollowing out each face as to reduce the arris to a right angle. The same thing was done for the same reason by the Gothicists in many of their octagonal features (see ninth example in the parallel of neckings in p. 185). There is only one case, however, of the 16-sided shaft-that bold example crowning the promontory of Sunium (Cape Colonna), evidently designed for distant view. Everywhere else we find the sides increased to twenty, on account of the common-sense principle which requires that in every structure, as solids should be over solids, and voids over voids, so should projections be over projections, and recesses over recesses. Let the square in the annexed figure be the plan of the abacus, and a ba, part of that of a 16-fluted shaft. If a recess be placed as at a a, beneath the most receding parts of the abacus, (or those nearest the axis,) then a recess also, as at b, will come under its most prominent point.

But by increasing the flutes to twenty, one can be placed centrally under each face of the abacus, as at c c, and an arris (or greatest projection) at d, under the angle of that member. This could not have been obtained with any other number of

flutes, between twelve and twenty-eight, of which the former might probably be used with advantage in bold plain engineering works, but the latter would introduce too much of the principle of gradation, in the seven gradually diminishing quantities, from the visual middle of the column to its visual side.

It might be thought that contrast would be better consulted by making every recess or flute, a pair of planes meeting in a nook, as if the plan consisted of five superposed squares (as practised in some Egyptian works with a smaller number); but not only would this introduce unnecessary complexity by doubling the number of lines, but by drawing the outline elevation of such a column, we shall find, in going from the centre to the side, a breach of continuity-a sudden change in the law of gradation, at that recess where we first lose sight of the nook-line. Though gradation was to be avoided, it was felt that wherever it did unavoidably occur, it should be continuous. A sudden breach in any gradation is ugliness, because it is neither regularity nor irregularity. It is the same principle on which we condemn the sudden change of curvature in the Tudor arch, and any change from one curve into another, except the perfectly contrasted flexure, as noticed in Chapter II.

There are obviously only three simple modes of striating columns by convexities alone (reeding)-by concavities alone (Doric fluting), and by alternate concavities and convexities (scalloping). The last is the mode most common in

mwm

Scalloping.

Reeding.

Doric fluting.

nature, because regular striation is here confined to elegant (not grand) objects, and this is abstractedly the most elegant kind, being all gradation and no contrast. Of the other two modes, the Doric affords most contrast, for several reasons. First, all its lines (which are the only places where contrast of light and shade can occur) are visible,-while in the reeded column, only a few of the nook-lines can be seen at once. Next, only two of these nooks in the reeded example can so receive the sun as to have one side shaded and yet the other not shadowed by it as at a, b. In the nook c, both

sides receive light, though not equally; and in d, one casts its shadow on the other: now the edge of a cast shadow can never have the sharpness of contrast that an actual edge of a body has. Moreover, in concave surfaces, as already remarked, the cast shadow of the edge often (in sunshine) reduces great part of the concavity to equable shade, and thus obviates part of the gradation that is unavoidable on convexities.

The angular plan of the column ceases at the top of the shaft, because its continuation throughout the swell, or echinus, would introduce too many curved lines. It would be more elegant than the present capital, but less fit in a composition of which grand severity (not elegance) was the aim, and in which the curves were made as few as would just suffice to give greater value to the general rectangularity. The long fluting lines, then, being stopped suddenly, the same principle that called for the sub-triglyph, required here the contrast of strong and repeated horizontal lines. One was not sufficient to stop such long and strong lines as the arrises; so three, four, or five of these stopping lines (annulets) were made, according to the height of the column, and their profile carefully studied to produce the strongest alternation of light and shade. The Pæstans trusted to intensity instead of number, and substituted one deep, black hollow, but the leaves introduced therein show a great decline from Doric severity. Indeed, all the colonial examples are very impure.

The diminution and entasis, essential to the character of limb-columns, do not, as might be thought, interfere with the severe rectangularity of the style, but actually increase it when seen from a near point of view. To explain this, we must remember that the ocular images of objects are formed on the retina, which is not a plane but a spherical surface, and the most severely contrasted angle is not always an actual right angle, nor yet that which appears most so in perspective, but that whose image on the retina is most rightangled. Every designer should understand spherical perspec