^{page 37 note 1} Archaeologia, lv, 186.

^{page 37 note 2} Röm. Funde aus Heddernheim, ii, 40.

^{page 39 note 1} Lindenschmit, Alterthümer unserer heid. Vorzeit, Band III, Heft ix, Taf. 3.

^{page 41 note 1} In a lesser degree it appears elsewhere, e.g. on a brooch (fig. 3d) of the trumpet type, sub-group R (i), at Traprain Law (P. S. A. Scot., 1915–16, p. 97, fig. 22, no. 1). This brooch might, for all I can see, belong to the late first century.

^{page 41 note 2} Until 1929, it was usual to date this destruction c. 180 (Dio, lxxi, 7, § 1). Evidence found at Birdoswald in that year seems to demand moving its date to c. 197 (ibid., lxxv, 5, § 4). See Cumb. and West. A. and A. S. Trans., N.S., XXX, p. 200; Arch. Aeliana, ser. 4, vii, 164.

^{page 43 note 1} In using the letter R for the entire group of trumpet-brooches, I am anticipating the classification of brooches in a work on the Archaeology of Roman Britain, shortly to be published.

^{page 55 note 1} How debased these imitations are may be seen when examples (e. g. W.A.M. xliii, p. 181, D; ibid., p. 390, B; or the Woodeaton specimen, fig: 12 g, at the Ashmolean) are compared with those figured in e.g. the Newstead, Traprain, and Corbridge reports.

^{page 56 note 1} Flawed castings of trumpet-brooches are fairly common at Brough-under-Stainmore; but though head-stud brooches were made there, I have never seen a flawed casting of one.