Scientific research could doubtless, if it would, do much towards perfecting Pisé-building.

We know very little about the behaviour of different earths under compression, or of their several reactions to chemical treatment. Meanwhile, a few trifling mechanical modifications are all that distinguish our modern plant from that devised by the ancients. That said, a short description of the “Mark V” model may be of some interest, pending the future developments that may now be hoped for.

 
see caption

 
see caption

II

The chief desiderata in designing a satisfactory Pisé plant appear to be these:

All constituent parts should be reasonably light and easy to handle. The shutters should be rigid and not liable to warp, without being expensively constructed. The shutters, when clamped in position, should be firmly and positively supported, without deviation from the vertical.

The fairway between the shutters must be as little obstructed by the cross-braces as may be, leaving good room for the men on the wall to tread and ram.

The through-pins by which the shuttering rests upon the base wall or on a completed course of Pisé, must be easily withdrawn without injury to the wall.

The shuttering must be easily disengaged and removed from the wall, one side at a time.

The special corner-piece must have some means of rigid attachment to the ordinary shutters on the two meeting walls.

There must be some means of blocking off the shuttering at any desired point, for the forming of door or window openings at any level.

The whole apparatus must be as simple and as fool-proof as possible, and built to stand rough usage and exposure to the weather.

III

The author has attempted to construct a plant embodying these essentials, and the working drawing and photographs shown will give the reader a tolerable idea of his “Mark V” model.

The thing has, at the moment of writing, only been experimentally tested in one of the London parks. These trials were, however, sufficiently satisfactory to encourage a belief that the new plant will prove a very considerable improvement on the old. It has now been despatched to a site in Surrey, there to undergo the searching and very practical test of being used for the building of a small-holder’s house and homestead.

 
see caption

A Simple Mould for Pisé Blocks.

 
see caption

Block-Moulds, Large and Small. The Latter shown opened out.

IV

To the second edition of this book a postscript must be added. Since the last paragraph was written, the small-holder’s house has come into actual being at Newlands Corner, near Guildford, and has attracted a good deal of attention from the Press, both at home and abroad. It has been inspected by multitudes of people, including a great number of Colonials and prospective Colonists, and by many distinguished persons directly or indirectly concerned with the problems of housing.

That “Good wine needs no bush” may be a true saying, but a novel system of building assuredly needs demonstration, however great its merits. The success of the experiment at Newlands is admitted by all who have made the pilgrimage thither. Often would critics come to scoff and remain to pray. Specially prized amongst the converts is a foreman-bricklayer once openly scornful in his unbelief. Of enthusiasm, perhaps, there has been almost over much; and it has been difficult to restrain the zeal of would-be pisé-builders until the coming of spring, and the return of such weather conditions as the craft might reasonably demand.

For pisé is a “dry-earth” method of building, and, as at present practised, that means it is a summer job, so far, at any rate, as England is concerned.

The author is the last person to claim that pisé-building may be successfully and economically carried out in all places, and at all seasons. He merely suggests that in a great many parts of the United Kingdom, pisé offers possibilities of cheap yet permanent building that are very well worth exploitation.

A wide and thorough trial of the method now seems assured under a variety of conditions in a sufficient variety of places. Pisé is to be given its chance in Housing Schemes, in Government building demonstrations, on Ducal estates, and by ordinary private citizens in need of houses—by the rich (old and new), and by the poor.

If Reason Rule

If reason rule, pisé will make good and all will be well.

If pisé-building is attempted where the conditions are unsuitable and in defiance of its physical limitations, the misguided enthusiasts responsible must blame only themselves. But it is not self-reproach alone that they will have to suffer, for the author and all true friends of pisé will view their troubles with as much anger as sorrow.

Nothing could be so well calculated to bring discredit on a new movement as the failures of a few enthusiastic incompetents.

 
see caption

Sketch of a Pisé House in Course of Erection.
With acknowledgements to The Sphere.
Larger View
or hover your mouse over the text areas

THE FIRST DEMONSTRATION PISÉ DE TERRE HOUSE
AT NEWLANDS CORNER, NEAR GUILDFORD

With acknowledgments to the “Spectator”

Description.—The house has six rooms arranged on one floor, of areas and cubical contents as laid down in their higher “schedules of accommodation” by the Ministry of Health and the Board of Agriculture.

The plan is an adaptation of the first type illustrated in the Board’s new manual “designed for the guidance of County Councils and their architects” in the matter of buildings for small-holdings.

The walls are of 18-in. solid pisé-work, the roof of red Bridgewater tiles, and the chimney breasts and stacks of brickwork.

The floors are boarded save for the back kitchen, which is tiled. The inner partitions are of 2-in. breeze blocks, the ceilings are plastered, and the casement windows are of steel.

There are two good lofts for storage, one entered from the barn, which is an extension of the house proper.

The pillars of the barn and the partition wall between scullery and veranda are of 18 in. by 9 in. by 9 in. rammed earth blocks; the angle pillar to the veranda is of similar blocks made from soft chalk.

The rest of the structure is of monolithic pisé, built up in situ without joints of any kind, either horizontally or vertically.

Cost.—The total cost of the whole of the outer walling of the house (in pisé) amounted to less than £20. Had the walls been built in brickwork the cost would, according to estimate, have been about £200.

see caption

NEWLANDS CORNER PISÉ HOUSE. THE PLAN.

The Newlands Specification

Specification.—The following is an abridged extract from the specification so far as it affects the pisé-builder:

(1) Excavate to a depth of 9 in. over the site, dumping the turf and surface humus where directed.

This soil is not to be used for building.

(2) Lay a 6-in. bed of cement and flint concrete 3 ft. wide under outer walls. Centrally on this, lay two courses of brickwork in cement, to a width of 18 in., or build up to the same extent in concrete.

Lay on this an approved damp-proof course; if of slates, having a further course of brickwork or concrete above it to prevent fracture when ramming.

 
see caption

The Newlands Corner Pisé Demonstration Building.

(3) Erect the walls according to the plan on the bases thus formed, carrying them up plumb and true and properly bonded by working round the building course by course, using the special angle-pieces at the corners to keep the work continuous and homogeneous.

(4) All stones and flints above a walnut size to be removed by riddling and reserved for concrete.

All sticks, leaves, roots, and other vegetable matter to be eliminated.

 
see caption

Newlands. The Cottage from the South-east.

(5) The soil immediately on the site to be used without admixture of any sort and to be thrown direct into the shutterings.

No water to be added without the express permission of the architect.

(6) The boxes are to be filled in thin layers of not more than 4 in. at a time, and well rammed until solid. The workmen are not to use their rammers in unison.

 
see caption

Newlands. The Garden Court.

(7) Rammed earth at box ends to be shaved down to a 45 degrees slope so as to splice in with new span of pisé adjoining it.

Where door and window openings occur, the special “stops” to be adjusted and firmly secured so as to withstand hard ramming. Two 4 in. by 2 in. by 9 in. plugs to be built in to each window jamb for the securing of the frames and three to each door jamb.

Special care to be taken in the thorough ramming at the corners and along the box edges.

(8) Insert below floor level, where directed, 24 3-in. field drainage pipes to act as ventilators through the thickness of the wall. Insert wire mesh stops to exclude vermin.

 
see caption

Newlands. The Backyard, showing Barn with Pisé Pillars.

(9) Set all frames square and plumb, and where in outer walls, flush with finished exterior plaster-face, the joint being covered by a 2-in. by ¾-in. fillet.

Where lintels occur, they are to be tailed in at least 9 in. on each side the opening.

Provide plain picture-rail round all rooms at window-head level, providing plugs for fixing where necessary.

Secure to floor round all boarded rooms a 2-in. by 1½-in. angle fillet as skirting.

(10) The smooth surface of the pisé walling to be hammer-chipped to give good key to the plaster.

Before rendering or plastering walls, any loose earth or dust to be removed with a stiff brush and the wall surface evenly wetted.

The rendering to be carried evenly round the walls—the minor square angles being roughly chipped down first so as to obviate sharp corners. The main corners of the house are ready-rounded off to a 9-in. radius by the special corner mould.

 
see caption

Newlands. Framing the Roof.

(11) Bond brick and slab work to pisé walls by driving iron spikes into the latter every few courses at joint level and bedding in.

(12) Colour-wash walls with tallow lime-whiting tinted with ochre. Provide 2 ft. skirting of pitch, applied hot, to form base-course round exterior of building.

 
see caption

Newlands. An Interior, showing Fire-brick Hearth Fire.

N.B.—The exterior of the walls of the Newlands Corner house have been finished in several different ways with a view to determining the most durable and economical form of epidermis.

A trial pisé-building adjoining has stood for four years without any external protection whatever. It has suffered no damage and grows continually harder. For the sake of appearances, however, and for the better preservation of the wall from chance injury whilst still “green,” a coating of some sort may be deemed necessary.

A Swedish Contribution

THE THEORY OF PISÉ

The Swedish scientist, Mr. Karl Ellington, of Nossebro, who is basing a book on pisé (in his own tongue) upon the frail foundation of the present volume, has, in the course of a letter to the author, made some exceedingly suggestive “guesses at the truth.”

“I am very interested to hear that you are proposing to use an hydraulic rammer for making blocks. I have thought a good deal about this pressure business. I am trying to scrutinise the thing from ‘the inside,’ so to speak. I am trying to trace out how Nature makes rock. That helps us to understand pisé. Nature made all the stratified rocks out of what was once fine loose earth and mud. Rivers carried the mud out to sea. Waves pounded and gnawed the shores and got down some more stuff. The tides went forth and back and shovelled and levelled at the sea-bottom. Some more mud on top of that, and a few hundred or thousand feet of the heavy water on top of that—and Nature’s pisé was in its making. But why do these mud particles stick together for ever even after that stratum is raised up high above the sea and the pressure is discontinued? That is the counterpoint of the whole problem. What is gravitation? Is it some form of magnetic or electric energy? We don’t know. Do particles of mud grip and hold each other if they are forced together close enough to be united by some sort of magnetic or electric energy? Or do the particles only get a ‘mechanical’ grip on each other? However that may be, we seem to know now that we can make them grip by bringing them closely together. It would seem important, then, that we must bring as much of particle surfaces together within any given cubic space as we possibly can; that is, we must have as little of ‘holes,’ ‘empty spaces,’ pores and channels as possible in the mass, in the pressed wall. This, then, would in turn make it important that plenty of very fine (small) particles must be present in the mass—and so well distributed among the coarser particles as to be on hand close by wherever there can be one more chance for a small particle to fill a little chamber that the coarser particles would like to bridge over. We can think of how well Nature was fitted for this work of shuffling over all the particles at the sea-bottom and under great water pressure till she got every particle into the niche where it would exactly fit. She used waves, tides, and gulf streams as shovels and mixers and packers, and the water above as ‘hydraulic rammer.’ Looking at the pisé matter in this way, it would appear that both the mixing and the shuffling are of vital importance. And by ‘shuffling’ I mean in this connection only that the smaller and larger particles get a chance to shift over a little during the process of pressing the earth together to hardness, so that the pressure may not work only and exclusively in a straight downward direction, but in a sort of wavy zigzag direction as well—much as when a street-roller is working the macadam and gravel a little forth and back at the same time as downward. I have a great respect for old tools which are the outcome of long-time experience and handed-down wisdom. I suspect the presence of some of that sort of experience in the rammer described in your book, p. 59. That tool would do the necessary shifting while attending to its main intention: hammering the mass solidly together downwards. Now for your hydraulic rammer—is it advisable to make it blow or press only in a straight line downward? Maybe there ought to be two or three kinds of strokes alternating—one stroke with a rifled or wavy surface under the rammer—and the next stroke with a plane surface. . . . What sort of witchcraft enters into the effect of high frequency blows as compared with blows with a little longer intervals between? Do the strokes create also some ‘magnetic’ effect in the pounded earth-mass which helps to fasten the particles to each other? And does this magnetic charge or friction heat, or whatever it is, act more promptly if one keeps on ‘striking the iron while hot,’ instead of letting the charge ‘evaporate’ and sneak away between strokes? Two or three of my hairs are turning grey over these questions alone. You compliment me by insinuating that I might stumble across some fruitful idea for the forms or boxes if I speculate a little more on the key-problem. Well, the thing won’t leave me alone, so I have thought out several foolish variations and rejected them too. But the last one seems to have a little more vitality, so if it will live till I write my next letter I will tell you about it. One is so apt to follow the temptation of ‘perfecting’ an apparatus—at the cost of getting away from keeping it cheap, simple—and ‘fool-proof.’ By this time the idea has grown ripe in my mind, so that I ought to write out a little book on the pisé problem in Swedish and have it printed before springtime. Something ought to be done. . . . I have to ask you kindly to permit me to make use of the data contained in your book. To this I will have to add what special precautions we must observe as to foundations in a climate like ours. I intend to treat only the pisé method. Cob and chalk methods are not applicable here, as we have such materials only in a few unimportant spots.”

Mr. Ellington has long been an admirer and a firm friend of England, and he is good enough to regard his country as indebted to ours for the introduction of pisé-building:

“Let me tell you that the help you are giving me now—not me, but my nation—will work as an additional bond that draws us more closely towards each other. . . . Some of our people here have always looked too much towards the South and too little towards the West.”

A Pisé-Builder’s School

PISÉ, PRACTICE AND PLANT

Now that so many able architects and enterprising bodies are seriously taking up pisé-building, the improvement in plant and technique should be both rapid and considerable. The School of Pisé Building established at Hornchurch in Essex, by the Imperial Ex-service Association, should alone provide us with much new and valuable knowledge of a highly practical kind.

It is there, for instance, that various types of shuttering and rammers are being experimentally tested side by side, and their relative efficiency under varying conditions ascertained. Under some conditions it is probable that the floor and roof timbers (destined for use in the house under construction) will be found the most economical and satisfactory form of temporary “shuttering” for the making of the earth walls.

The pisé “Test-House,” built by Messrs. Alban Richards at their Ashstead works, was built in this way, and proved highly satisfactory.

Another effective and more generally applicable form of shuttering (designed and manufactured by the same firm) is illustrated in the diagram reproduced below. It should be observed that wedges intervene between the movable shutters and the uprights.

The method of employment of the “Mark V” shuttering is well illustrated by the bird’s-eye view showing the Newlands cottage under construction.

Alternative Shutterings

In this matter of shuttering there is still, however, great scope for improvement, and it may be hoped that soon ingenuity and experience will jointly produce a complete pisé plant perfectly fulfilling all the many conditions enumerated earlier in the book.

Shuttering made by riveting plain galvanised sheet iron to one side of a corrugated sheet has the qualities of lightness, smoothness, cheapness, and rigidity, and the claims of the inventor and patentee are now being put to the test in actual building.

see caption

PATENT SHUTTERING FOR PISÉ DE TERRE
By W. Alban Richards and Co.

There now seems little doubt but that pisé blocks will be largely used for partitions and chimney stacks where the soil is good enough, and experiments are being made with a view to discovering the best and cheapest way of making earth slabs similar to those of coke-breeze and concrete.

The size aimed at is 18 in. by 18 in. by 3 in., the edges to be tongued and grooved.

Certain “concrete” machines seem to lend themselves to adaptation for the making of earth blocks, but it is necessary to remember that sharp blows are required rather than a steady pressure, and also that we are working with a dry material. The ordinary primitive way of making pisé blocks is indicated below.

The hand-rammers are undoubtedly worth study and careful design. A set of three seems to meet all ordinary requirements, and those shown on p. 101 may be taken as typical. They should be of hard-wood, smoothly finished, and provided with long handles. They should be 9 in. to 12 in. long, and about 5 in. by 4 in. at maximum cross section.

In the sketch they are shown “narrow-ways-on.” No. 1 is used for preliminary pounding and final finishing, No. 2 for general consolidating, and No. 3 for working along the edges, against window stops, and under cross-ties.

A South African correspondent, Major Baylay, makes interesting comment as regards rammers and local pisé practice:

South Africa

“My experience of all black labour is, that they won’t put any ‘guts’ into it. They therefore want fairly heavy rammers, which they can lift and drop, say a foot, and which will do the rest for them. The heat of the sun and extreme dryness of atmosphere out here make it advisable to cover up completed courses at once with sacking, moist for choice, otherwise it is liable to dry out too quickly and crack. It dries out uncovered at night very well, when there is no rain.

“The red loams of South Africa, where not too sandy, make excellent pisé. They or their equivalent are found almost everywhere. In the dry state they set so hard that moisture added just before ramming is useless. A large heap must be made, well damped and covered over with moist sacking, and left until the moisture is distributed throughout the mass. When about four or five days old, in ordinary weather, the earth is ready to use—viz., just wet enough to bind when gripped in the hand. It should be passed through a sieve. I use a sort of ‘chicken run,’ 8 ft. long, and throw the earth on to it before using. Six feet of it is ½-in. mesh, and 2 ft. ¼-in. mesh; the reason for this is that, if the earth is a little too dry, it does not always bind well with the previous layer. Therefore, put a few petrol tins of the fine earth into the shuttering first in order to ensure good bond, and throw the coarser stuff in after.”

Second Note by Major Baylay, Peter Maritzburg, Natal, South Africa

“I have completed a small building, and though weather conditions have been as bad as possible, it is sound and very satisfactory.

“In my opinion, pisé-building should not be attempted in the rainy season in Africa. Earth contains too much moisture, and the power of the sun dries it out too quickly and causes cracks.

“Re plastering. I covered the outside and inside with a mixture of 6 earth, 2 sand, 1 blue (Hyd.) lime, the earth being the red, rather ‘fat’ earth found everywhere, and the same stuff the house is built of. It is put on thin with a trowel, after damping the wall. When it dries and cracks, rub all over with a sacking pad covered with the plaster mixture, but wetted to a thin cream consistency. It may sound an odd method, but the natives do this work well, and the result is as good as one can wish for. You can put tar or any wash (No. 6) on this.”

Soils

SOILS

Were it not for the fact (often somewhat embarrassing) that soil quite incapable of making good pisé will none the less produce enthusiastic pisé-builders, a warning as to the vital importance of the earth being really suitable might seem superfluous.

The author has found some of the staunchest champions of pisé-building living on and valiantly struggling with stiff glutinous clay and almost pure sand.

Even the most vigorous optimism can achieve little under such adverse conditions unless soil-blending be resorted to, and even so, pisé-building begins to lose points in the matter of economy directly complications of this sort are introduced.

Fortunately, however, England is well off in the matter of pisé soils, the red marls being amongst the very best.

A study of the country, or, failing that, of the geological maps, will reveal a great tract of this earth extending diagonally right across England, from Yorkshire down into Devonshire, where it ends conspicuously in the beautiful red cliffs about Torquay.

There is a large area of the stuff in the Midlands, notably in Warwickshire, with lesser patches here and there about the country.

Second only to the red marls come the brick earths, which, fortunately, are also widely distributed.

“Brick earth” is merely clay that has been well weathered and disintegrated under the action of wind, rain, frost, and organic agents, the sulphides having become oxides, and what was a cold intractable slithery mass having become merely a “strong” and binding earth.

It is probable that even stiff clay, if dug in the summer or autumn, and left exposed for a winter, would prove sufficiently reformed to be quite amenable for pisé building in the spring.

After the marls and the brick earths there is an endless variety of soils that will serve well for pisé-building—some, of course, better than others, but all, save the extremes (the excessively light and the excessively clayey), capable of giving good results under proper treatment.

Before putting pisé construction actually in hand, however, the intending builder will do well to submit samples of his earth to some competent authority, that they may receive his blessing.

A fistful taken from a depth of 9 in., and another from say 2 ft. below the surface, should give sufficient evidence as to the soil’s suitability or the reverse.

III
CHALK

III
CHALK

§ I. General

Chalk, as a source of lime, has always been of high importance to builders, and, until improved transport brought alien materials into its old preserves, chalk was in general use for walling in the form of roughly squared blocks.

Chalk again forms the basis of a compost that, used in the form of a stiff paste, has been largely employed for building from the earliest times down to the present.

“Pisé de Craie,” or chalk consolidated by ramming within a casing, is a form of building that has been long held in high repute in France and elsewhere, but which has only recently been given a serious trial in England.

Chalk in all these forms, if fairly dealt with and reasonably protected from the weather, is a most amenable and satisfactory material to build with.

The last-named method particularly seems to promise results that should satisfy the most exacting critics of the unconventional, as it assuredly does those who inhabit the cottages so constructed.

The several systems of chalk construction are fully dealt with in the pages that follow.

Chalk Compost: Historical.—At the Ancient British village on West Down, Chilbolton, some five miles south of Andover, delving archæologists have brought to light undeniable fragments of chalk “Daub,” with the wattle marks still clearly showing upon them.

This discovery is chiefly of academic interest, though it is a pretty refutation to those who regard any building material save brick and stone as “new-fangled,” and it should also serve to hearten the doubters and the timid amongst us who seek historic sanction for any departure from current building practice.

Composition and Uses.—In the Andover district Chalk Compost or “Chalk Mud,” as it is called locally, is prepared and used as follows:

The chalk is dug out in the autumn, and the frost allowed to play on it during the winter. In the spring building starts, and the weathered chalk is spread all around the outside of the walls. Straw is sprinkled on it and it is then well trodden, usually by the workers, but sometimes by horses. Sometimes chopped straw is added, sometimes unchopped straw is sprinkled on. The quality of the walls depends very largely on the preparation—that is, in getting the mud to the right consistency—and the old hands know by experience when it is ready.

The compost is lifted on the wall by a fork and another man stands on the wall and treads it in. It is then chopped down straight with a spade. Some of the naked walls at Andover show traces of the courses, which are usually something under 2 ft. in height.

Where a course has to be left unfinished it should be ended with a diagonal ramp so as to splice in with the work that follows.

Some of the old builders seem to have been somewhat catholic in their conceptions as to what constituted “chalk,” and vague patches of earth, loose flints and other stray substances not infrequently mar their work and sometimes seriously reduce its strength.

As a general rule, the finer the chalk the stronger and more durable is the walling.

What is aimed at is a conglomerate of small chalk knobs cemented together by a matrix of plastic chalk and straw, the whole forming as dense a mass as possible.

Grinding in a mortar-mill would probably reduce all the chalk to an amorphous powder, which would not be desirable, and in any case such mechanical mixing is quite unnecessary.

Building by ramming the moist compost between timber shutterings does not appear to have been practised in the past, though there is nothing against the method except its tendency to delay the drying out.

The drying of each course takes several days, depending on the weather. A course is usually laid right round the building. It must be covered up at night in case of rain, and when it is hard another course is laid on, and so on till completion. The aim is to build during the summer and autumn, and when the moisture has dried out, to render the exterior.

Where brickwork is used with chalk compost it is generally bonded in in the ordinary way, but block-bonding the depth of a chalk course is a better way of doing it.

The exterior corners of chalk buildings are the vulnerable points, and these should therefore be well rounded off.

Timber.—In the old work nothing seems to have been done to prevent woodwork built in to the compost from decaying, though in many cases it has survived surprisingly. In any new work, however, proper ventilated air-spaces should be contrived or the timber ends treated with some preservative.

The door and window frames are fixed to fairly large pieces of wood built in across the thickness of wall, and other woodwork is fixed to wood blocks built in in a similar way.

Picture-rails should be provided in all rooms, as chalk walls are apt to flake and chip if nails are driven into them.

Lintels are usually of wood, and when plastering is carried down over these some form of key must of course be provided to hold it.

Winter Work Barred

Frost.—New work must not be exposed to frost or there will be danger of collapse, and winter work is barred out for this reason.

Repairs.—Chalk compost walls are not easily repaired in that material, and bricks are generally used, well bonded in.

Chimneys.—Chimneys, too, are usually of brick, though there would seem no reason against the flues being carried up in chalk, especially if clay pipe linings were used.

The chimney-stacks above the roof might well be built in flint, the corners being rounded off in deference to the peculiarities of the material.

External Rendering.—It is of the first importance that a good weather-tight skin be maintained, and many old buildings have suffered through neglect of this precaution.

The rendering was often of the poorest quality, more mud than lime, and the constant repairs that the indifferent materials necessitated has resulted in many of the old cottages becoming patchworks of variegated plaster blotches, when not whitewashed over, which give an impression of dilapidation by no means warranted by the facts.

Rendering.—Given a good skin, however, of cement or cement and lime, a chalk conglomerate wall will last indefinitely. So vital is the skin that it is as well to put it on in two good coats—rounding off all the corners and finishing it either with slap-dash or rough from the wooden float.

Also, to ensure its proper adhesion throughout, wire-netting may be used as reinforcement—being secured to the face of the chalk wall by means of cross netting or wires laid on the wall as the building rises.

If the netting be of a fine mesh it also serves as an absolute barrier to vermin, though pounded glass incorporated in the base of the wall is equally effective.

Strength.—Provided the wall has dried out thoroughly, any of the ordinary loads occurring in a two-storied house can be borne with ease.

Chalk conglomerate walling, however, has no great lateral strength, and it should not be asked to stand up to thrusts.

The roof, therefore, must be well tied, and should sit on the building merely as a lid.

 
see caption

 
see caption