This section is from the book "Modern Buildings, Their Planning, Construction And Equipment Vol5", by G. A. T. Middleton. Also available from Amazon: Modern Buildings.
As a general rule, a Contractor sets out his angles by means of a large square made in his workshops, by setting two pieces of timber at right angles to one another, and binding them across the angle. The accuracy of this is easily tested. It should be laid down, say, on a flat piece of ground, and a line marked along each edge. One of these lines should be continued by means of a straight-edge, and the square should be turned over so that one of its sides rests on this continued portion. The other side should still lie alongside the line at right angles to it, as before. The Foreman's attention should be called to any incorrectness, but, as in everything else, it is not the Clerk of Works' duty to put it right himself. Any angles other than right angles should all be checked by cross measurements.
In the matter of the heights, which involves that of the depths of foundations, a Clerk of Works' responsibility is great. It is always possible upon sloping ground for a Contractor to make a slight mistake greatly to his advantage in the matter of digging. The first thing for the Clerk of Works to determine is what spot upon the drawing represents the present ground level, both in plan and section. This he should take as a datum, and as the level on any given point near the building will be altered again and again he should set up from it another datum at some point which is not likely to be changed, and preferably at two or three such points. Say, for instance, that the ground in front of the principal entrance is to be taken as datum. This point is located on the site, and the level staff is held there, a back-sight reading being taken on to it, say, of 4.23. The level is then directed towards a fence, and the staff is held so as to rest upon the bottom rail of the fence, the exact point being marked first in pencil, and subsequently with paint and a saw-cut, so that it may not be obliterated. Say that the reading here is 2.12. Assume that the first point is 10 feet above datum, this being convenient if none of the excavations are to be carried deeper than 10 feet. The collimation line of the instrument will then be 14.23 above datum, and the level of the bench-mark determined upon the fence will be 14.23 - 2.12=12.11 above datum. This is recorded in the level book, and may very well be painted on the fence. If practicable, and it generally is practicable, the level should be turned round to some other object on quite a different part of the site and clear of the building operations, such as a tree or a gatepost which is not going to be removed, and the staff should be held against it, and moved up and down by the holder, until it again reads 2.12, when a pencil mark should be made at its foot, and this again should be subsequently made permanent by saw-cut and paint, or by paint only in the case of a growing tree, which might be injured by cutting. Thus a second bench-mark is set up, at the same height above datum as the first, so that if either be accidentally destroyed the other can be worked from; while, further, similar bench-marks can in future be determined from them if necessary. It is now known that the point on the ground line, shown on section in front of the main entrance, is 10 feet above datum, and from this it is easy to calculate what should be the height of the bottom of the trenches and of the top of the concrete, as determined off datum, whether the building be a simple one or a complex one, whether the foundations be all carried down to the same level, or whether there be many different levels, with cellars of varying depths; while the drain trenches, with their gradual fall, can be equally well determined. Then, when the trenches are excavated, it is not at all a difficult matter to anyone who is accustomed to levelling to check whether the correct depths have been reached. Say that a Clerk of Works is asked to pass the depth of a certain trench. He consults his section and finds that the bottom should be 3 feet 6 inches, - that is, 3.5 feet below the gravel path before the front entrance. In other words, it should be 10 feet-3.5 = 6.5 feet above datum. He sets up his level in such a position that from it he can see one of his bench-marks in one direction and the trench in another. Sighting on to the bench-mark, upon which the staff is again held, he this time obtains a reading of 2.27, and this added to 12.11, the height of the bench-mark above datum, determines his present collimation level to be 14.38. Subtracting 6.5 from this shows him that the staff ought to read 7.88 when it is held in the trench. He instructs the holder of the staff to take it to the trench, and sees him place it there. He finds, perhaps, that it reads 7.90, and with this he is satisfied, as it shows that the excavation has been carried a quarter of an inch too far; and, of course, he would be equally satisfied if the error in the other direction were as slight. In order to test the level of the bottom of the trench as excavated, he now has the staff held in it again at another spot, It should again read 7.90, or very nearly so. If not, the trench has not been properly levelled; but unless there is a material error which would result in the concrete thinning out too much, this may be passed, as absolute accuracy at a trench bottom is not to be expected.
It may be said that it is quite a different matter with drains, which must be kept absolutely true to their proper inclination. The top of the concrete is tested in the same way as is the bottom of the trench; but when drain levels are tested it must be remembered that there is the necessary fall to provide for, and that this should be regular.
Instead of first setting out a front wall or central axis, it is sometimes advisable to lay out a corridor, and, in fact, a line through a building, such as the side wall of a corridor, will often form a better basis for measurements than any other. On a very large building it is well to determine two or three main walls, which run through or across the building, wholly or partially, and if two walls are chosen these are best at right angles to one another.