were, one and all of them, constructed, and which is the very opposite of the ordinary tie-beamed roof of the present day. Where stone vaulting was used to finish the building internally, the external roof, though an important feature in the outline of the building, was, in fact, a mere protection to the vaulting, and nothing more; but the open roofs which succeeded them, took the place of the vaulting quite as much as it did that of the outer roof, and were constructed on the same principles, as far as their connexion with the side walls is concerned. They required a similar, though, for obvious reasons, not so strong an abutment; and all the mass of wood-work connecting the principal rafters with each other in pairs, so far from acting as ties, only served to increase the thrust. Even were there room for doubting this, it is well known to every builder, that, in practice, a tie can only act as such when placed at the very feet of the rafters; even placed a short distance up, the strain becomes too great to make it a desirable position for a tie. The collars, therefore, which connect the principal rafters from purlin to purlin, so far from acting as ties, only serve to keep the rafters sufficiently apart, and to prevent them from bending inwards by their own weight, and what they have to carry. Some examples, which have two purlins on either side, have but one collar, and that so placed as to connect the principal rafters at the top pair of purlins only. In others, where, as before, there are two purlins, on either side, with a collar to each, to connect the principal rafters to each other, the lower one might be supposed, theoretically, to act as a tie; but many have not even that, and the principal rafters have but one collar, and that placed at their junction with the higher purlins. The side walls may, therefore, be said to receive the whole weight and thrust of the timber roof, just as they used to do the stone vaulting, the weight in most instances being applied to the walls by the spandrils below the hammer-beams, abutting against them between every two windows of the clerestory. Flying buttresses were dispensed with only because the walls themselves were thought to be sufficiently massive to form an abutment; the timber roof, heavy, as in many instances it was, being incomparably lighter in its application to the walls than the stone vault. But to suppose that there was any resisting power in this form of roof to counteract the thrust, is acting on an assumption which has been disproved in every case where the side walls have at all given way. Westminster Hall has had flying buttresses built in places where the side wall has bulged fearfully. In Castle Headingham church, Essex, the walls have been thrust out sufficiently to spoil the effect of a roof which once must have been very beautiful. In other roofs of this kind, it has been found necessary to connect the hammer-beams with iron ties from side to side of the building; all this tends to prove how completely this form of roof is dependent upon the side walls for support.

Some appear to think that a roof of this character, when applied to a short building, (one of two bays only, for example,) requires less abutment than the same would if applied to a long building.

Now, except in so far as the ridge and purlins could not sink at the ends, being supported by the gable underneath, and so could not at that point thrust out the side walls, the abutment is not stronger in a short than in a long building. For example, the east end of a chancel has usually a large window so placed that the impost of the arch is very little below the top of the side wall; and the space between the window jambs and the side wall being a very small portion of the entire distance between the angles of that part of the building, it is clear that those portions of the wall have enough to do to resist the thrust of the arch of the east window without having to act as a tie to keep the roof from thrusting out the side walls. Of course, the angle can be well buttressed, but then it is the buttress and not the wall that does the work of resistance. Be this as it may, however, it would hardly be safe to overload the side walls, because they happened to be short in the direction from east to west. In vaulted buildings, where the side walls rose above even the apex of the east or west window, the wall might, and doubtless did, act as a tie, to a certain amount; but in open roofs of a high pitch, the case is usually otherwise. It seems that this particular kind of roof became less common for churches, but was retained for halls, the lowpitched roof having been adopted very generally in the later perpendicular churches. The safety and durability of this style of roof, the lapse of nearly four centuries goes far to prove; and though it may be demonstrated to the satisfaction of some persons that they are altogether unfit for our climate, and incapable of bearing the weight of snow that in winter must often rest upon them, yet they have most provokingly stood firm four hundred winters, more or less, and promise to last for many more. The structure of this kind of roof is extremely simple, and well adapted to a small church. Here, the side walls do not usually form abutments to the roof, which, in most cases, merely rests its weight upon them, and in theory connects them with each other after the manner of a tie. The mode of construction may be thus described:-Strong stiff beams are laid across, at intervals from wall to wall; and across these, and at right-angles to them, are laid others, the middle piece acting as ridge piece, the side pieces as purlins. Upon these are laid the rafters, which receive sufficient inclination downwards to the side wall, by the ridge piece having a block, or a short length of timber, laid between it and the beam below it. On the rafters are laid, longitudinally with the building, the boards upon which the lead cover is laid. On this simple principle the roofs of a large portion of our village churches are constructed. It is not always easy to discover what the use, real or intended, of the spandril is; whether it be to stiffen the beam above it, or the walls against which it is applied. Where it is very large, and the clerestory wall very lofty, as is the case at St. Mary's, Oxford, in the roof of the nave, it may very well act as a stay to the side walls ; in this particular instance the spandrils meet each other midway below the beam; but this is not one of the class we are describing.

In general, however, the spandrils do not extend beyond the purlin. We may take as an example the roof of the nave of All-Saints' church, Stamford. It has to cover a considerable space, the distance between the walls being twenty-three feet.

From a slight deviation from the original form, perceptible in some of the principals, it would appear that there the spandrils serve, as it were, to help out the length of the beam, rather than to combine with it the form and use of an arch. If the latter were the case, it would act as a powerful lever against the walls to thrust them outwards. But this, if we may so say, is the proper office of the spandril, and therefore the wall should always, in these cases, be of sufficient thickness to resist the thrust. In the chancel of this church, that office seems in a fair way of being performed. There, the pitch of the roof is much higher than in that of the nave, and the span less by seven feet, yet it has been found necessary to furnish the principals with an abundant supply of iron bolts and plates, and to connect the side walls near the east window with an iron tie.

But if the spandril does not combine with the beam above it the form and purpose of an arch, it is in the wrong place to stiffen it merely. A better arrangement may be seen in the chapel on the south side of the chancel. The space covered is, as in the last example, only sixteen feet, which is not much; but the beam does its work effectually.

In the chancel roof of Cherry-Hinton church, Cambridgeshire, the beams are constructed on this principle, on a much larger scale. It is in the middle of the beam, not at the ends, that the greatest substance is required. In the majority of instances, however, whether with or without spandrils, the principle of construction seems to be the simple one of connecting the side walls to each other by strong beams, and crossing them with others so arranged, as to carry the entire covering of the building. There is no attempt at mechanical contrivance. The principle of the modern trussbeam seems to have been unknown* to the architects of those days. They made no attempt to place each piece in such a manner that it might act to the greatest advantage, that its dimensions would admit of. The strain upon it was mostly in a transverse direction, not in that of its length. But the waste of material is, perhaps, more than compensated, by what is gained through its depressed form, and the capability of oak to resist lateral pressure.

As far as decoration is concerned, the interior may be made very beautiful, at a comparatively trifling cost; trifling, when we take into the account the quantity of timber and workmanship to make an equally handsome high pitched roof, and the increased thickness of the walls to aid the abutment. But such roofs are usually too flat for blue slate, and lead is too expensive an article for modern church builders. But if lead of sufficient thickness be found impracticable, on the ground of expense, zinc has the advantage of being both lighter and cheaper, considering the thickness sufficient to form a covering; and its temper and ductility are better known now than when it was first brought into use.

The flat roof has not been a favourite of late years, owing to the rage for early English, the prevalence of which seems to have been, and to continue to be, almost universal. Perhaps some persons may be bold enough to suggest the possibility of this roof being as suitable to early English, as are the roofs now usually placed upon such buildings. Many, perhaps most of the early English churches that remain, are covered with them, and what kind of roof they might have had originally seems difficult to determine.

In Rickman it is stated (and he still continues to be an authority in these matters) that "there do not appear to be any early English wooden roofs, which can clearly be distinguished as such." The south transept of Castor church, near Peterborough, is early English, with a high pitched roof, very possibly the original one; but it may

To whatever age the introduction of the truss-beam may belong, it will be admitted that the general character of roofs constructed before the sixteenth century, would lead us to the above conclusion. Even the tie-beam, as such, does not appear to have been thought of any great advantage.

We speak of this roof from recollection. It was one of those called by Britton, in his Architectural Dictionary, a compass roof. The whole of this church is very interesting, and will repay the trouble of a visit. The greater portion of it is Norman, but with the exception of that of the south transept, the roofs are, if we remember

be doubted whether an architect could be found who would be willing to copy it; and it is unlike any of the productions of the present age. Mr. Rickman, though he mentions this church in his Northamptonshire list, could not have visited it. In many a village church of that period, the low pitched roof harmonizes well with the massive Romanesque-like character of its masonry; and we can scarcely fancy the internal view would be improved by its having a new open roof substituted for the old one. However, such examples of the architecture of the thirteenth century are never copied now-a-days, be the church to be built great or small. Nothing but fragments of Beverley or Westminster will satisfy the age in which we live. The miniature cathedral has superseded the rectangular box with which we were contented ten years ago, and will, in its turn, it is to be hoped, give place to something as different ten years hence. Of all the styles of pointed architecture, that of Westminster Abbey may justly be considered as the most complete; and perhaps for that very reason, it is the worst to follow, unless we are prepared to use it in all its glory, with its noble and graceful proportions mounting upwards from base to boss. It is the last of all to bear mutilation. Much is said in these days of the importance of producing a fine development of the style, but this appears to be forgotten when the building is to be covered in. From a pier of elegant, and at the same time sufficiently massive proportions, flanked with four graceful detached and banded shafts, capped with the most correctly copied foliage-from such a pier, there spring arches which carry a meagre clerestory wall, from the top of which rises a roof, made after the fashion of those of the fifteenth century; a style of roof, moreover, which was far more common for halls than for churches: every thing in character till you are half way up the pier arches, when you are gradually prepared for the disappointment of not finding a splendid vault, crowning a building, which in many respects might be worthy of it.

Whatever may be the improvement which the attentive study of the principles of pointed architecture has produced within the last few years, it can scarcely be said to have extended to roofs. Much ingenuity may have been displayed in some cases, and very successful copying in others, but hitherto, early English in the hands of the moderns may be said to have fairly withstood every effort to make it accommodate itself harmoniously to an open roof. The later the style of pointed architecture, the more does the timber roof become it. It seems almost a question whether the peculiarities of the

correctly, all of perpendicular date. The author of the Glossary of Architectural Terms, speaks of early English roofs, as though there were some still existing, and he instances Old Shoreham church, Sussex. He may be, and probably is right, but we have not heard of any attempt being made to investigate the question, with a view to building roofs in the true early English style, whatever that may be. *This may, however, be partly attributed to the comparatively smaller width of churches.