How were these great straight limbs plotted out? That is a question which has been fully debated but not yet settled. Where two ends of a trajectory were in sight one from the other the matter was simple enough; but what was the method used when the straight line exceeded the horizon: when it was carried on, for instance, for more than thirty miles, as is the case over and over again in the great north-eastern road from Paris to Cologne, and in the road from Amiens to Boulogne? What was the method when, even for lesser distances, one end of the trajectory could not be seen from the other on account of intervening hills, or where in flat land forests were sufficient to impede the view? One theory has been that of smoke signals, a method which has been found of use, I believe, in barbarous countries in our own day. We must, I think, certainly reject it in such a climate as ours. Such signals could only here be used upon a few days in the year, specially picked, and the Roman engineers would not have depended upon the caprice of the weather. There has also been suggested (I adopted that suggestion myself in the monograph of which I have just spoken) the use of high movable platforms, but I now think that this also should be rejected on account of its clumsiness, and of the fact that in an uncleared country it would often be quite impractical. The most probable method was suggested to me by a correspondent some years ago, based upon his own experience in the planning of roads in new countries. It is the method of _odds and evens_, and requires some description with the aid of a simple plan.