¹ This explanation of amúm is due to Thumb, Handb. 254. Brugmann, Grundriss² 2.2.343, regards amúm as double accusative of the stems *e- and *u-.

^1a Wackernagel-Debrunner's formulation is somewhat less positive than mine.

² But cf. §7 end.

^2a The same extension of iyaṁ to the masculine appears in East Asokan.

³ Walde–Pokorny 1.187–8; Brugmann, Kurze vgl. Gramm. §825 and §837.

^3a E.g. Wackernagel 1.338, 2d paragraph, for pracetā.

⁴ Trautmann, Baltisch-slavisches Wörterbuch 20 sub au̯a-.

^4a In *amúi, morphologic innovation would have created a new sound combination, *ui, which did not exist in the language before. The inherited combination of u + i was of course only v i (e.g. bahú- ‘much’, fem. bahv i; madhv iva 'like honey'). This innovation has an analogue in the pre-consonantal ur of RV catúrbhiḥ (1st and 2d books), caturṇ am (8th book), caturdh a 'fourfold- (4th book) etc., which has replaced older *vṛ (*catvṛbhiḥ etc.) after the analogy of catúraḥ (before consonants only ūr is original). Still older is the Indo-European nasal-infix type *bhind- ‘to cleave’ etc. for regular *bh iṇd-; cf. Wackernagel §67 and Edgerton, Lang. 19.96–7 with fn. 36; 109 §64.

Analogy was even able to produce a new vowel: the r of the (RV) acc. pl. pit rn and gen. pl. pit rṇ am. Long r existed perhaps for some time (only before ḍ and ḍh) also outside these forms, if the type mṛḍīká- ‘merci’ = Avest. mərəždika- had originally long r, as Wackernagel holds (§28). But even in this case the r of pit rn is probably older than that of *m rḍīká-.

⁵ Wackernagel–Debrunner 3.§248d.

⁶ Franke, Pāli und Sanskrit 110, 129⁷.

⁷ J. Bloch, Langue marathe 355b.

⁸ Cf. Turner, Bull.SOS 3.401; the present writer, Lang. 19.14, and (for Hi. bhejnā) JAOS 65.161.

⁹ So J. Bloch, Langue marathe §124 and §64.

¹⁰ Franke, Pāli und Sanskrit 110.

¹¹ W. Geiger §16 and §17; Pischel §177.

¹² True, Latvian has only dêls ‘son’, while ‘daughter’ is meîta.

¹³ Bloch, Langue marathe §99, assumed IE gh.