Although in classical Arabic all short vowels, as a rule, are preserved, a is more persistent than i/u: in prose, pausal -in/-un are elided, yet -an shifts to -ā. In many modern Arabic dialects too (dubbed by J. Cantineau diffeérentiel) a tends to be sustained in phonetic environments in which i/u are elided. This is, it seems, the reason that in the Bedouin dialects of northern Arabia and the Syrian-Iraqi desert it is the historical tanwīn -an, rather than -in/-un, that is preserved, especially when preceding an indefinite attribute; even phonetic -in has, it seems, to be derived from original -an. The same applies to medieval Judeo-Arabic 'n (spelt as a separate word) in this position.

Readers of Father Serruys's two articles on the Ordos salt lakes may be interested in the evidence relating to the episode referred to in his documents, which is provided by the reports, 1917–22, of the Sino-foreign salt inspectorate. The evidence is twofold: direct, relating to the episode itself, and indirect, relating to the context. In preliminary clarification it should be stated that both Chinese and Sino-foreign sources refer to three Ordos salt lakes at this date: Pei-ta-ch'ih, the northern great lake, ‘200 li due north of Hua-mach'ih’; Wo-po-ch'ih, evidently a transliteration; and Kou-ch'ih, the dog or petty lake. These would seem to be equivalent, respectively, to Father Serruys's Great Lake, Lake of the Cairn or obo, and Little Siker Lake.

Although any lingering doubts about the lexical value of the Semitic particle bal ought by now to have been dispelled by the observations of Labuschagne and Aartun, a few further remarks on the problems involved in contextual analysis might not be altogether superfluous. While this particular instance of etymological overkill is not included there, the several pitfalls encountered in the practice respectively of comparative philology and textual emendation are meticulously documented in Barr's monograph on methodology. And the dangers of ‘pan-Ugaritism’ have been signalled more than once. The purport of these caveats is clear: textual problems will not always (or even often) be solved by reference to cognates. Occasional solutions may be lexical, seldom morphological, and even more rarely syntactic, since ‘meaning’ is, in the final analysis, a product of context.

Scholars working with problems in ethnology, languages and history of the Manding-speaking peoples and some of their western Sudanic neighbours have long been aware that while much in the impressive range of works by Maurice Delafosse demands consultation, the writings of this prolific but fallible pioneer in West African studies must be treated with an extra measure of caution. The more one makes use of what is perhaps Delafosse's best-known work, Haut-Sénégal-Niger, the more apparent it becomes that a critical reassessment of his writings would be desirable.

Sogdian has two words for ‘thief, robber’, both of good Old Iranian pedigree. From Av. gaδ- there is B. γδ- /γaδ-/, attested in the pl. γδ'yšt, P 6, 14, SCE 312, and in γδ'wny ‘theft’, SCE 331, 426. More common, from SKr., Av. tāyu-, is B. t'y(h), VJ 239, P 2, 1113; 7, 138;24, 2, Chr. ty, C 2 51R.31, pl. S. t'yt (Man.) T i α 2 (Tales, 478), also L 61, 3; 103 R 2, and B. t'y'wny, P 9, 15, L 98a 1, Man. ṯ'ywnyy, BBB e 6‘theft’, B. t'y'yck', VJ 339 ‘stealthily’. But there are more Buddhist thieves, not recognized by the editor of the Paris text P 2, E. Benveniste. Where t'y there occurs, in 1043, 1091 (t'yt)-95, 1100, he adopted R. Gauthiot's earlier interpretation ‘āyatana’, ignoring the difficulty that these—whether ‘abodes’ or ‘senses’ —can hardly 'škr- ‘pursue’, and justifying it by a paper etymology: ‘Le sens est proprement “entrée” (cf. t'ys- “entrer” t'y < t-'y- comme tγ'm “guè” VJ 64e <t-γ'm); ce doit être un calque de chinois jou “entrée”. La preuve [sic] en est que dans Vim. 15, tys “entrée” signifie āyatana d'après le chinois (Weller, As.Maj. X, p. 323).— Différent est naturellement t'y “voleur” 1113’. An *ᾱy- ‘to come’, linked with gam forms, is not found in Sogdian, however, nor for that matter in any other East Iranian language. ‘Āyatana’ is only (once) tys in Sogdian, and *t'y ‘entry’ is a ghost word.