admit that, while they suit certain circumstances, they are deplorable in many others. A mat finish can be obtained with great facility with all shades without exception, but there are three which have been adjudged worthy of serious consideration. These are the brown. mat finish obtained with rutile; the green obtained by combining rutile with cobalt oxide; the gray blue obtained in the same way as the green, but by the use of a less proportion of cobalt."

The introduction of mat finishes in ceramic decoration filled a longfelt want. They are necessary because they make possible the use of certain motifs of sculpture that lose much of their character with brilliant glazes. The best things should not be abused, however. The mistake has been made of applying mat finishes to everything, notably in architecture. The Sèvres manufactory has set the example, as, indeed, it may well do, since it is our great national establishment for ceramic study. That factory was responsible for the construction of the immense portico of stoneware with a mat finish seen to-day in the square of Saint-Germain-des-Pres in Paris. This sample of modern ceramic architecture bears witness to our inferiority to the Orient, where they use brilliant enamels exclusively with a skill that might well inspire our architects and ceramists. Paris to-day numbers several edifices of mat stoneware. Let us hope that they will not be multiplied.

The tones communicated to glazes and covering layers by metallic oxides are, from the standpoint of coloration, closely dependent on the nature of the atmosphere of the kiln. The flames that come from the hearth of a kiln and penetrate into the firing chamber, inclosing the articles to be fired, are possessed of different properties according to their composition. They may be oxidizing, neutral, or reducing. A flame is oxidizing when it is saturated with oxygen derived from the air and contains not only an excess of carbon but an excess of oxygen. Under these conditions it is blue in color. A flame is neutral when it contains neither carbon nor oxygen in excess. It still has a blue tint, but this is less intense than in the preceding

A flame is reducing when there is a lack of oxygen in it and it is saturated with carbon. It is then yellow in color. It is therefore easy to understand that these flames, having different compositions, will exert considerable influence on the colors which may be submitted to their action. The ceramic ware which we usually see is fired with an oxidizing flame.

Two kinds of decoration may be obtained with a reducing fire; one at a low temperature, the other at a high temperature. The decoration with a low temperature is represented by the faïences

@ All the formulæ for mat finishes have been published many times, and are known to-day to all ceramists.

45745°- -SM 1909 42

with metallic luster. To obtain these scintillating deposits, whose process of manufacture I have described several times, it is only necessary to apply to the glaze after it is fired a copper salt-oxide sulphide, oxalate, or the like. The piece is then heated to a dull red (about 550°) and afterwards submitted to the action of reducing gases. These iridescent tones may be varied by the addition of salts of silver and bismuth, but I have previously shown that the brown, yellow, golden, or red shades of metallic deposits are not due, as heretofore believed, to the nature of the mixture of oxides, but to the duration of the reduction.

The copper red, called rouge, flammé, or flambé, is fired in a reducing flame, not at a low but at a high temperature. At the present time, it is true, the public has been so misled by ceramic fairy tales that they are completely ignorant as to what these flamed pieces are. As a matter of fact, when a particular vase has little value, and is consequently difficult to sell, they embellish it with the description " flammé de grand feu" (flamed at a high temperature), demand quadruple the actual worth, and almost always then find a purchaser. Everyone seems to possess a peculiar mentality that leads him to see something wonderful in the simplest things. He is easily led astray by these ceramic fables, for the art to him is secret, of fearful temperatures, and of mysterious manipulations worthy of the most somber epochs of the middle ages. These fables are a relic of Bernard Palissy, who had better never have written his Art de terre.

Many imagine that these flame effects can not be obtained except on stoneware. This is again an error; they are even more beautiful on porcelain on account of the whiteness of the clay.

The term flammé applies exclusively to that beautiful red finish, with a copper base, which generally shows vertical striations due to the variable intensity of the coloring material. These variations of tone are a result of the action of the reducing gases, which contain zones in the interior of the kiln which are not all saturated in the same degree with carbon dioxide or hydrocarbons. This is clearly indicated by the fact that on a vase colored with this copper red are frequently found greenish or bluish parts, and sometimes even palpable greens or blues showing an oxidizing action, and brown and black parts produced by a more intense reduction than that causing the vivid reds. These peculiarities show that there evidently exist varied zones in the kiln, some oxidizing, some neutral, and some reducing, the latter of course being the preponderating element.

These rouge flammé coverings may be either brilliant or mat; the former, however, are incomparably more beautiful than the latter. To obtain a mat red a large proportion of alumina (in the form of kaolin) or lime (in the form of chalk) must be introduced into the

covering layer. By a large proportion I mean from 20 to 30 per cent. The covering should be applied to the piece in a very thin layer, however, 2 millimeters at most.

The reduction should be an energetic one, and in this connection I have observed that this can be controlled by the proper regulation of the draft damper. The constructors of kilns have gotten into the bad habit of placing this directly over the grates, and I have noticed that such a practice is injurious to the proper operation of the kiln, while if the regulator is placed at the base of the chimney, which surmounts the dome, the draft is perfectly even; it can then be regulated in such a way that the interior atmosphere may be rendered at will either strongly oxidizing or strongly reducing.

In an oxidizing fire a lead glaze with copper acquires that vivid green tone which has always been very popular among primitive peoples, and which is still in general usage in the north of Africa— for example, in Algeria and Morocco, in spite of constant contact with France-in Italy, and in Spain. An alkaline covering (rich in potash and soda) acquires a remarkable blue tone (Egyptian blue), to which may be given a greenish tinge by the addition of lead. This gives a turquoise tone.

In a reducing fire the green glaze, the blue glaze, or the turquoise become red, but the last is alkalino-plumbous and will always give a much more beautiful red than one containing only lead.

With this red I have associated the titanium blue. We have seen that rutile, natural titanic acid, gives a yellow or brown-yellow in an oxidizing fire. Under the influence of reducing gases, the titanic acid passing over into the blue sesquioxide of titanium is an important feature in the decoration of the flamed pottery. This titanium blue has the remarkable peculiarity of preserving its blue tone in artificial light, instead of appearing brown like the cobalt blues, or green like the copper blues.

When the copper red and the titanium blue are combined on the same piece, violet would ordinarily be expected, but each color keeps its own shade. Violet can nevertheless be produced as I have observed in extremely rare cases, an effect due, perhaps, to particular conditions of reduction that escaped my attention in the firing. Violet tones in the flamed ware can be easily obtained, however, by combining the copper red with the blue of cobalt if the latter is not added in too great excess.

Finally, I have obtained some curious modifications of tone by combining copper, sesquioxide of titanium and glucinum. (I say copper and not oxide of copper because, according to the theory of Ebell, the coverings are colored red, not by the oxide but by the metal itself.) I introduce the glucinum in the form of silicate, conse

quently as emerald, using the emerald of Limousin, which is nearly white because it is free from the chromium oxide present in the green emerald used in jewelry; an oxide which is really an impurity.

To sum up, ceramic decoration in a reducing fire makes it possible to obtain effects which are totally unknown in decoration by an oxidizing firing. The colors obtained by reduction possess a considerable power; they are not due to chance, as certain ceramists declare, but they are determined by true chemical reactions between the metallic oxides and the products of combustion. We have not yet mastered them, it is true, for the study of them is only roughly sketched out, and if I have been able to determine the exact conditions under which copper red can be obtained at will it is only the result of a considerable number of observations."

Decoration with a reducing flame offers a vast field for research which I believe will reveal some absolutely new facts.

In this brief study of the evolution of ceramic decoration I think I have demonstrated sufficiently the truth of the statement I made at the beginning; that is, that this art has been intimately connected with the history of races since the origin of humanity.

I can not enter into all the details of the production of copper red. I have recently written an article on this subject which appeared in the Transactions of the English Ceramic Society, vol. 8, p. 71 (on the development of copper red in a reducing atmosphere).

SOME NOTES ON ROMAN ARCHITECTURE.

[With 4 plates.]

By F. T. BAGGALLAY, F.R.I.B.A.

Roman buildings, after remaining for three centuries the sole inspiration of the architects of all Europe, have for a long time now received far less attention from English students than is their due, whether they are judged on their merits or by the fact that they are the direct ancestors of all such modern architecture as can claim ancient descent.

Our attention has lately been directed once more to Roman work— first, by the fact that when last English architects examined it they were content to look at a part only, and hardly went below the surface even there; secondly, I think, by our growing acquaintance with M. Choisy's illuminating work on Roman building methods; and, lastly, by a half hope that, as the Imperial Roman system of construction was largely a monolithic concrete system, it may contain some suggestions for dealing with ferro, or reenforced, concrete: all the more since Roman concrete was almost always itself in a sense reenforced with brickwork. It is true that to call the brickwork used with concrete by the Romans "reenforcement" is somewhat misleading; at best it has but little resemblance to what we call reenforcement now; and much is merely centering or facing. On the whole, however, the cause for paying special attention at the present time to Roman architecture is strong enough, I hope, to excuse me for bringing the subject before you, although I have no fresh information and little new to say about it. I should like to treat it historically; for, although the dry facts connected with an architectural style or system of building can be learned without reference to the historical point of view, they can only be really interesting or fruitful when seen in chronological sequence and in connection with the circumstances that molded them. Without the help of history we see only the effect and not the cause,

a Read before the Birmingham Architectural Association, February 19, 1909. Reprinted by permission, after revision by the author, from Journal of the Royal Institute of British Architects, London, third series, vol. 16, No. 20, October 16, 1909.