beautiful and most perfect of the abbey edifices of France[588]?” And, coming to meaner structures, dare we say that a discrepancy of 5° or 10° is likely to be the result of mere carelessness? One might as reasonably argue that the existence of a “low side window,” with its deviations of size, shape, and position, indicates a clumsy arrangement of fenestration. Rebutting arguments against the “theory of error” may be urged on broad grounds. We have seen that in the greater number of instances, our undeflected buildings bear to the North of the equinoctial point, indicating, if we suppose a calendar alinement at sunrise, that work was begun either a little after the advent of spring, or, less probably, a little before the autumnal equinox. Consider, then, the case of a church of which one member is to be rebuilt. Under similar conditions of starting work, the assumed error in direction might be on either side of the axis. We might perhaps conclude that it would actually tend to be on the Northern side of that axis--a Northern deviation added to a Northern deviation--for, in spite of liturgical rules, a liking for summer orientations (North of East) appears to have been very common. But, not to press this probability, it is at least claimable that, on a balance of instances, the deflections, if due to clumsy workmanship, would range themselves in equal numbers North and South of the earlier alinement. But the fact remains that the Southern variations form the rule, and we are driven to the belief that they are Southern by reason of design.