1920 ITINERARY OF ANTONINUS 381 a source other than that from which the numbers of the inter- mediate stages were derived and that he left many of these incongruous numbers as he found them.-'^ It is possible, therefore, that he made no effort to harmonize his authorities. Such efforts were, however, still being made by redactors even intheCarolingian period, as the marginal notes of our older manuscripts (the Vindobonensis L and the Parisinus B) abundantly testify.^ There is another passage in which the Tabula helps us to emend the text of the Itinerarium. At page 472, line 2, in the second Iter, instead of Vagniacis mpm xviii, we must read Vagniacis mpm viii. The corrupt reading (xviii) makes the distance from Londinium to Durobrivae 37 miles. The actual distance is nearly 29 miles, but both the third Iter and the fourth agree in making the distance 27 miles and, accordingly, we may presume that this was the ancient calculation. Moreover, the corrupt reading makes the distance from Noviomagus to Duro- brivae 27 miles, which is 10 miles in excess of the number (xvii) given for this stage by the Tabula. Our correction brings all our four authorities into agreement. The stages are as follows : Itinerarium. Modern name. Mileage. Londinio mpm xii Noviomago mpm x Vagniacis mpm viii Durobrivis mpm viiii London Bexley Heath Springhead Rochester 11 9 9 Similar cases of a superfluous x are not wanting in other parts of the Itinerarium, but they are apt to be obscured in Parthey and Pinder's edition, since these editors, naturally enough, adopted the critical principle that omissions are more likely to happen than interpolations, and, consequently, where the numbers given by their manuscripts differed, usually adopted the higher number for their text. At p. 260, line 3, however, they read xxiii for the. distance from Siscia to VariaTW, in spite of the reading xxxiii given by the Escorialensis P. In this case the lower number is confirmed by the parallel passage at p. 265, line 6.^ After all there is no great critical principle involved. The manuscripts were not usually written line by line, but first the names were written down in one column as far as the bottom of the page, or the end of the Iter, and then the numbers were written ^ See Corpus Inscr. Latin, xiii, vol. i, pt. 2, p. 587. Where an Iter was derived from Kubitschek's road-map, the headings would necessarily either have originated with the compiler or been taken by him from an independent source.
- These notes have R prefixed to them. See the quotation from Ekkehart above
p. 377, n. 3.
- Other instances will be found in the Danubian Iter, pp. 242-9.
