The trick in studying early Spartan history is knowing when to stop. The temptation is always to keep squeezing our scraps of evidence a little bit harder, and the danger is always that the resulting theories will persuade no one but the author. Andreas Luther's new book is a case in point: it contains much sound analysis as well as a good deal of attractively adventurous speculation on the development of the Spartan constitution, but it also contains some pretty wild ideas. Rather than follow his discussion step by step, I shall present his arguments in order of plausibility, which as it happens leads me to begin at the end.

Herodotus' story of how the ephors forced king Anaxandridas to contract a bigamous marriage opens the last section of this short volume (94-137) which examines the role of the ephorate from c. 550 BC until the end of the Peloponnesian War. A careful analysis of the evidence leads to the conclusion that this century-and-a-half saw no real change in the position of the ephors or in their relation to the kings and elders. The tale of Anaxandridas shows ephors and elders already threatening a king with legal action if he should defy their authority, just as they would do on many a later occasion down to the trial of king Pausanias in 403 BC. Stories about Cleomenes I may seem to portray a king with greater power than most of his successors enjoyed, but a closer look reveals that this was a matter of personal influence rather a different constitutional position (101-104, 117-119). The only attested innovation is a law of 506 BC which no longer allowed the kings to share command over a single army, but this did nothing to reduce the power of the monarchy (105-114, 138-139).

In this part of the book, the argument is persuasive, even if some troublesome bits of evidence are swept under the carpet. Herodotus' claim that the kings could declare war at will (6.56), for instance, poses a problem which surely deserves more than a bald statement in a footnote to the effect that it is 'exaggerated' ("überspitzt", 134 n. 472). Again, while it is no doubt true that stories about the ephor Chilon and about Lycurgus' alleged creation of the ephorate are unreliable, one should at least explain why they might have been invented rather than refuse point blank to discuss them at all (18-19). Conversely, despite optimistic assumptions commonly made about the three-generation reach of oral tradition, one might wonder whether Herodotus' account of mid-sixth-century events is really much more trustworthy than his or anyone else's stories about Chilon, set about a generation earlier.

For the period before 550 BC, Luther regards only two pieces of evidence for the Spartan constitution as worth considering: the Great Rhetra and Tyrtaeus' 'Eunomia'. In the first two-thirds of the book he develops remarkably radical views on these two notorious texts.

The primary argument is that neither source contains anything which implies constitutional arrangements different from those which obtained in the classical period, and that the absence of any explicit reference to ephors does not prove that these magistrates did not yet exist. The obscurity and corruption of both texts pose numerous problems which are clearly and comprehensively set out (35-41, 79-87), and the details of classical Spartan decision-making procedures are not much clearer, so that there is plenty of scope for Luther to make his case. Everything, he argues, points to a three-stage process whereby the ephors on behalf of the assembly put proposals to the Elders, who then vetted these, and finally sent back approved proposals to the assembly for confirmation by vote (48-54, 137). The chief support for this reconstruction is Plutarch's account of a crisis situation in the mid-third century (Agis 8-9) rather than classical or archaic evidence, but it is nevertheless a quite attractive way of reconciling the disparate information offered by our sources. As for the ephors, they may have been subsumed under 'the men of the people' in Eunomia (85-87) and under the assembly in the Rhetra (45). The latter is more difficult to accept than the former, since the Rhetra otherwise offers quite detailed regulations and is not the 'bare outline' ("dürres Gerüst") which our author makes it out to be. As a fall-back position, however, we are offered the - to my mind remote - possibility that ephors featured explicitly in the Rhetra's corrupt line, or else in a lost part of the text (44-45).

This is only the start of Luther's revisionism, for he goes on to advance two even more radical propositions. First, the main text of the Rhetra tells us nothing about actual archaic Spartan institutions because it was not a law but an account of Lycurgus' legendary legislation (42-43, 59). The so-called Rider to the Rhetra, on the other hand, was indeed a law, but need not have predated the fifth century (46-47, 58). Secondly, and most startlingly, Tyrtaeus' work tells us nothing about archaic Sparta either, because he was not an early poet but active in the second half of the Peloponnesian War (66-79).

In each case, a small point of grammar leads to dramatic conclusions. The verbs in the body of the Rhetra are all either infinitives or participles in the accusative singular. Luther argues that the text needs a main verb and some indication of who is the agent, and he suggests that the text as transmitted was originally prefaced with a phrase such as 'The god told Lycurgus that he should ...' or 'It is said that Lycurgus did ...'. The things stipulated in the Rhetra would thus have been part of an early version of the legend of the lawgiver - older than the Rider (46) and therefore of archaic date - rather than a set of regulations actually in force in Sparta (33-35). This is ingenious but unnecessary. The text is perfectly grammatical and intelligible as it stands, without any supplement: the infinitives and participles clearly mean, as most scholars recognise, 'One must do X, Y and Z'. In a footnote (34 n. 106), Luther objects that this common interpretation is 'linguistically problematic', but does not explain why. Later, in discussing Tyrtaeus' 'Eunomia', he has no trouble accepting that the same construction is indeed used here to convey an order (82), and I would add that Hesiod's 'Works and Days', for instance, features several instructions which are grammatically exactly parallel to the Rhetra and where the meaning cannot be in doubt (592-594, 735-736, 748-749).

Luther is right to demur that the normal interpretation is 'only one of several possibilities'. The great advantage of the usual reading, however, is that it accepts the text as it stands, whereas his own hypothesis forces us to posit not only that the Rhetra was part of a longer text, but (1) that this text was an archaic prose narrative, when no other archaic literary prose is known, and (2) that somehow this literary account could later be misrepresented as a legal text. It would be hard to defend either point, and Luther does not really try.

The case in favour of a very late date for Tyrtaeus is even more tenuous. In speaking of the conquest of Messenia, Tyrtaeus used the first person plural on one occasion ('We captured Messenia thanks to our king Theopompus ...', F 5.1-2 West), and the third person plural on another ('The fathers of our fathers fought over it for nineteen years ...', F 5.4-8 West). From this, Luther infers 'on stylistic grounds' that the poet refers to two different Messenian Wars before his own time (66). The conquest by Theopompus is then dated to c. 600 BC (69-71), and the war of the 'fathers of our fathers' identified with the Messenian revolt of 464 BC (73-77), which puts Tyrtaeus himself two generations later, near the end of the Peloponnesian War, when the Spartans were also worried about Messenian raids and defections (78-79).

The 'stylistic' argument surely has no force at all, but one might concede that it is possible in principle that the two fragments referred to different wars. Unfortunately for this theory, Strabo tells us that the war of 'the fathers of our fathers' was according to Tyrtaeus the first conquest of Messenia (8.4.10; cf. 6.3.3) and thus to be identified with the conquest by Theopompus. Perhaps one could blame this on a misunderstanding, but the issue is not addressed. Strabo also reported (ibid.) that Tyrtaeus mentioned an alliance between Messenians and Argives, Arcadians and Pisatans in his own day, details impossible to reconcile with the Decelean War-scenario and apparently forgotten although they had earlier been noted (67 n. 220). The discrepancy between the 20 years of ancestral war in Tyrtaeus and the 10 years for the Messenian revolt in Thucydides is casually dismissed ('only rough indications of time', 77). The chronological complications of downdating Theopompus by about a century remain unresolved, and the big question of how Tyrtaeus came to be regarded as a seventh-century poet, if he was in fact a contemporary of Thucydides, Aristophanes and Euripides remains unanswered.

So far as I can see, Luther's main arguments about the lack of evidence for significant changes in the power of kings, elders and ephors can stand on their own, and it would be a pity if they were to be overshadowed by his more eye-catching subsidiary theories about the Rhetra and Tyrtaeus. He is perhaps rather too keen to press his case that there is no evidence for change - certainly on the last page where he gratuitously suggests that the ephorate may even have been part of a pre-state 'common heritage' of the Dorians (140) - but his work is an important corrective to the school of thought which assumes that an institution or procedure must have been introduced shortly before it is first attested. In view of the parlous state of the evidence for early Sparta, Luther is right to insist that some things may have gone unrecorded for decades, generations and centuries after they had come into existence.

Hans van Wees