^{page 299 note 1} La Trace de Rome dans le désert de Syrie, Paris, 1934Google Scholar. See my review of the two fine volumes in Geographical Journal, January, 1936.

^{page 299 note 2} JRAS., 1936, pp. 423 sqq.

^{page 300 note 1} The sketch-map (see p. 301) shows essential topographical features reproduced from Sheets I 38 and J 38 of the International 1 · 1,000,000 Map, with Roman posts, ruined sites, and old caravan routes added as located in the coarse of my surveys.

^{page 300 note 2} ProfessorHerzfeld, E. has reviewed available data with much critical care in his exhaustive article “Hatra”, ZDMQ., 1914, pp. 665 sqq.Google Scholar

^{page 302 note 1} Dion's, Roman History, lviii, 31Google Scholar, Loeb Classical Library, viii, pp. 419 sq. That Hatra is placed here in the territory known to the Romans as Arabia, i.e. the central portion of the Jazīrah, fully agrees with a previous passage of Dion Cassius, loc. cit., lviii, 22. This mentions Mannus as “the ruler of the neighbouring portion of Arabia” and Singara as taken from him by Trajan's general Lusius Quietus.

^{page 302 note 2} Dion, loc. cit., lxxvi, 11 sq.

^{page 302 note 3} Dion, loc. cit., lxxx, 3.

^{page 303 note 1} Aminianus Marcellinus, XXV, viii, 4.

^{page 304 note 1} Ibid., XXV, vii, 14.

^{page 306 note 1} The difficulties here presented for traffic are illustrated by the fact that the railway which for the sake of Mosul has been carried over this ground could cross it only with the help of a fairly expensive tunnel completed quite recently.

^{page 307 note 1} Herzfeld, , ZDMG., 1914, p. 676Google Scholar.

^{page 307 note 2} Poidebard, , Trace de Rome, pp. 118 sqq.Google Scholar; Geographical Journal, June, 1940, pp. 430 sqq.

^{page 307 note 3} Geographical Journal, July, 1938, pp. 62 sqq.

^{page 311 note 1} Kiepert, , Forma Orbis antiqui, v, p. 6Google Scholar. I owe this reference to Père Poidebard who before the start on my survey was kind enough to call my special attention to the need of tracing the direct route of the Tabula from Singara to Hatra.

^{page 311 note 2} Loc. cit., ii, pp. 306 sq.

^{page 312 note 1} Sarre-Herzfeld, , Archaeologische Reise im Euphrat und Tigris Gebiet, ii, p. 307Google Scholar.

^{page 313 note 1} Geographia, V, viii, 1; VI, iii, 4; see also Herzfeld, , Archaeologische Reise, ii, p. 307Google Scholar.

^{page 314 note 1} Ammianus Marcellinus, XV, vi, 9–11.

^{page 314 note 2} Ibid., XXV, viii, 5. The translation of C. D. Younge (G. Bell and Sons, 1911) is quoted here and elsewhere with such slight modifications as the original wording justifies.

^{page 314 note 3} Ibid., XXV, viii, 15.

^{page 315 note 1} Sarre-Herzfeld, , Archaeologische Reise, i, p. 66, note 1Google Scholar.

^{page 315 note 2} Thus it has been assumed that the indication of six days in the text refers to the whole march from Hatra to Nisibis, an impossible performance for an exhausted force over a distance of at least 220 miles (see Archaeologische Reise, iii, p. 306, n. 3).

^{page 315 note 3} Ritter, , Erdkunde, xi, p. 467Google Scholar.