Now, of the extreme care with which this preliminary question of level was considered by the builders of the great pyramid, we have singularly clear and decisive evidence. For all around the base of the pyramid there was a pavement, and we find the builders not only so well acquainted with the position of the true horizontal plane at the level of this pavement, but so careful to follow it (even as respects this pavement, which, be it noticed, was only, in all probability, a subsidiary and quasi-ornamental feature of the building), that the pavement "was varied in thickness at the rate of about an inch in 100 feet to make it absolutely level, which the rock was not."[43] But now with regard to the true north-and-south direction, although the shadow method, carried out on a truly level surface, would be satisfactory enough for a first rough approximation, or even for what any but astronomers would regard as extreme accuracy, it would be open to serious objections for really exact work. These objections would have become known to observers long before the construction of the pyramid was commenced, and would have been associated with the difficulties which suggested, I think, the idea itself of constructing such an edifice. Supposing an upright pointed post is set up, and the position of the end of the shadow upon a perfectly level surface is noted; then whatever use we intend to make of this observation, it is essential that we should know the precise position of the centre of the upright's base, and also that the upright should be truly vertical. Otherwise we have only exactly obtained the position of one end of the line we want, and to draw the line properly we ought as exactly to know the position of the other end. If we want _also_ to know the true position of a line joining the point of the upright and the shadow of this point, we require to know the true height of the upright. And even if we have these points determined, we still have not a _material_ line from the point of the upright to the place of its shadow. A cord or chain from one point to the other would be curved, even if tightly stretched, and it would not be tightly stretched, if long, without either breaking or pulling over the upright. A straight bar of the required length could not be readily made or used: if stout enough to lie straight from point to point it would be unwieldy, if not stout enough so that it bent under its own weight it would be useless. Thus the shadow method, while difficult of application to give a true north-and-south horizontal line, would fail utterly to give material indications of the sun's elevation on particular days, without which it would be impossible to obtain in this manner any material indications of the position of the celestial pole.