176 PROCEEDINGS OF THE AMERICAN ACADEMY. 



hardly necessary to assume textual corruption here. As in 1, 137, the 

 conclusion of the temporal clause is not definitely expressed, but is loosely 

 indicated by what follows. It is a good example of the fine disregard 

 of rigorous logical structure, which is thoroughly characteristic of 

 Herodotus. 



Customary or repeated action is expressed in the main clause by the 

 present or imperfect indicative, or an equivalent expression, in the tem- 

 poral clause by the subjunctive after a primary tense, and the optative 

 after a secondary. The adverb av is regularly used with the subjunctive. 



A noteworthy example of a clause of generic action with rjpos is found 



in 4, 28 : (Spovral re rjpoi tjj aWy ylvovrai, TTjvucavra pev ov ylvovrai, depeos 8e 



dfKpikacpees. Here the action of ylvovrai is general and the subjunctive 

 with av would be expected in the protasis. But in this case the indica- 

 tive is used because the action described by fipovral ylvovrai is regarded as 

 sure to take place and the time, though not expressed, is nevertheless 

 felt as definite. A similar usage is found in clauses with is '6 in 3, 104 : 

 iiii paXXov yj/v^fi, es o eir\ 8vo-pjjo-i Icov Kai to Kapra \|/-uy«, and 4, 181.* 



Future action is expressed in the main clause by any form which ex- 

 presses future time ; in the subordinate clause it is invariably expressed 

 by the subjunctive. When such clauses stand in indirect discourse 

 Herodotus always uses the principle of repraesentatio . But in 9, 51 the 

 subjunctive of a clause of purpose, though it is in the same sequence as 

 a clause with eireav and the subjunctive, is changed to the optative (eneav 

 . . . 17 . . . is hv fifj IdoiaTo). A similar case is found in 8, 61. 



B. Subsequent Action. 



The law stated above for antecedent and contemporaneous action holds 

 true in general for clauses of subsequence ; but clauses introduced by 

 npiv and TTp\v ij present certain marked peculiarities. After negative 

 clauses npiv and rrp\v rj are followed by a finite verb except in 1, 71, and 

 1, 1 (')•"). In 1, 71 : Tlepo-rjo-i yap, irplv Avdovs Karao-Tp(\j/ao-8ai, r/v ovre aftpbv ovre 



ayaBbv ovftev, the position of the clause with irplv f and the context show 



* iiixos is found in a similar passage in CM., 4, 400 : 



■tlfxos 5' T)4\ios /aeaov ovpavov ap.(pi[iff37iKei, 

 (a.fxi>L0ej3riKr] MS. Q and La Roclie). On the analogy of the passages cited from 

 Herodotus the indicative would be expected since the action of the temporal clause 

 is in no real sense contingent and the time is not indefinite but definite. Further- 

 more, -fifjiof. is elsewhere always found with the indicative except in later Greek. 

 Cf. Merry and Riddell, Homer's Odyssey, 2d ed., Oxford, 1886, note ad loc. 



t irplv with a finite verb stands before the main clause in Herodotus only once. 



