This illustration, external to our zone of study, is however quite revealing of a certain state of mind with regard to the metric one of the ways. In J Naveau 1997 (Bib.) one finds page 22, with Milliary columns 13, concerning a milestone of Jublains: " the ford is with more than four miles of Jublains. The monolith had to be moved approximately 1200 m to be thrown in the river ". Not very logical explanation... Why not simply not to consider that this milliary remained well places from there at the edge of the ford of St Léonard, on Mayenne, and that the miles in questions are Gallic leagues! Indeed 4 X 2,222 = 8,9 km, out distance estimated too short of 1,2 km. Let us correct to have the real distance 8,9 + 1,2 = 10,1 km. And 10,1 / 4 = 2,52 km, value of the Gallic league in this point. There is always a margin of uncertainty due to the round-offs, but even in this case the value would be included/understood between 2,45 and 3,00 km, completely compatible with the Gallic mile, but surely not with the romanized mile of 2222meter.

The milliary column 14, of Launay in Châtillon-sur-Colmont, discovered in 1981, is described by the author on the same mode: "the site of the wayside cross to the foot of which the column was lying exceeds approximately 1700 m the 9 miles of Jublains". Let us run the counter to this reasoning and suppose more logically than the wayside cross was indeed built with the site of a milliary column (of which anteriority is obvious). By applying same calculation that above, we can determine one Gallic league average of 2410 m, the inaccuracy being tall on so short-hauls, and the VIII leagueses can then be appropriate perfectly.

The precision of conversions increases with the number of leagues to convert, by reduction of the rounding round-off errors. On a determined route, it is interesting to represent the margins of uncertainty of round-offs according to the number of leagues. The double exponential obtained gives a good indication of the evolution of the precision, however that the central value gives the most probable value of the league used.