SOURCE: "The Composition of the History of Herodotus," in Classical Philology, Vol. LII, No. 1, January, 1958, pp. 9-19.

[In the following essay, Lattimore, a noted classicist, explores a series of textual and structural problems in the History. He provides a detailed analysis of Herodotus's compositional methods and considers the constraints on writing and historical research that Herodotus faced.]

The general problem. Concerning the composition of the History of Herodotus, one must choose between two general propositions. Either Herodotus wrote his book so that the parts always stood substantially in the order in which we now have them; or he did not. If we believe he did, we shall be more persuasive if we can show that this was, a priori, the more likely way for him to compose his work, and if we can present internal evidence in favor of a continuous forward process. If we believe that the parts of the work originally stood in an order different from what we now have, then we believe in transpositions of written material, or in insertions, or in deletions, or other or all sorts of revision; and we should be able to point to passages which we consider indicative of such processes, and show how they force us to believe in changes made by the author.

I believe that the text of Herodotus as we have it is a continuous piece of writing which Herodotus set down from beginning to end in the order in which we now have it; that the sentence at the opening of the work is a true statement of the author's intention at the time when he wrote it, and that it was written down before any of the subsequent text was written down; and that the whole History is, substantially at least, a first draft which was never revised, nor meant to be, because the first draft was always meant to be the final draft.

Practical considerations. Such a method of writing— to begin at the beginning and write straight on to the end, never working backward—was, I believe, the natural way to write at the time when Herodotus was writing. It was, I believe, the method followed by the predecessors of Herodotus in the fields of narrative and exposition, by Homer, Hesiod, and the elegiac poets; also, I believe, probably by the logographers, philosophers, and tragic and lyric poets as well. There are two good reasons for this. There is the fact that composed literature comes first and writing comes second; that the written piece is not for the reader to read to himself but for the writer, or his representative, to read from to others; that the writer therefore naturally thinks of himself as a speaker who, when he has contradicted himself or got his parts in the wrong order, cannot go back to correct or transpose, but must make the correction as he goes forward. There is also the fact that writing straight ahead is mechanically the easier way to write. In the time of Herodotus, writing at its easiest could not have been easy. We do not know exactly what writing materials he had, but we certainly know some of the writing materials he did not have. He did not have what we would call good paper, good ink, good scissors, or a good eraser. He did not have a set of identical sheets of paper and so he could not, if he found something objectionable on page 8, take it out and write a new page, number it 8, and put it in the stack where it belonged. He could not, in case he needed to add to page 8, make a new page called 8a, and insert it. It was probably very bothersome to erase. To be more particular: he could not write about Cambyses first, and then about the topography and anthropology and early history of the pharaohs, and then switch them around—not, at least, without some surgery. To insert the Athenian and Lacedaemonian excursuses (1. 59-69 or whatever you think the insert is) into an already written sequence would have been no simple process but a bitterly difficult one. What has been gained to justify the labor and time spent? It was hard enough to write the History anyway. Further, Herodotus had at no stage, as we have, a printed text of his History, conveniently if arbitrarily divided into books, chapters, and subchapters, and supplemented by Powell's admirable Index. It was far harder for him to find his way from place to place in his own work than it is for us to find our way in it. Finally, if he composed a finished piece from notes, those notes were laborious to make and unwieldy to handle when made.

The natural way for a writer to compose a large work under these circumstances would be to use what might be called a forward, or point to point, or progressive style; one subject to irrelevancy and contradiction, but controlled to some extent by being thought out in advance, and capable, for short or simple pieces, of producing a perfectly articulate work; still, because of the necessity to shape material in transit, assuming the superficial appearance of a number of independently made parts which have been rather badly assembled.

Cases of the progressive style: Correction-in-stride. "This is what the Persians and the Phoenicians say, but as for me, I shall not go on to say whether it happened thus or otherwise, but I shall indicate the first man who to my own knowledge was responsible for wrong acts against the Greeks, and when I have indicated him I shall go on ahead with my account … Croesus was a Lydian born, and he was the son of Alyattes … this Croesus was the first barbarian we know of who overthrew some of the Greeks and made them pay tribute to him, and he did also take some of the other Greeks as friends … Before Croesus ruled, all the Greeks were free … The kingship, which had belonged to the Heracleidae, came around to the line (Mermnad) of which Croesus was born, in the following way. Candaules … was lord of Sardis … and as Candaules was going to bed Gyges came out of hiding and killed him, and held his queen and his kingship.… Gyges also, once he was king, led an army into Greek territory, against Miletus and Smyrna, and he took the city of Colophon.… I will mention Ardys the son of Gyges who was king after Gyges. He captured Priene …"

The contradiction is evident. It can be explained but not explained away. The great achievements of Greeks and barbarians, above all, how they fought each other. Who began it? Stories are told from the heroic age, but that is not a time we have real knowledge about. When we look for the first real historical character who wronged the Greeks, and so caused the wars, we light on Croesus. An exception immediately occurs, the Cimmerians but the Cimmerians merely raided and ran. Exception dismissed. But who was Croesus and why was he on the throne? The question takes us back to the succession of Gyges, and from Gyges we can pass on down through the known succession of Mermnad kings until we shall come to Croesus once more. But immediately there is a slight embarrassment. The erga of Gyges included attacks, patently unprovoked, on Greek cities, and Gyges came before Croesus. We might say Herodotus, who has not yet wasted much work, should scrap his beginning and start all over again, with Gyges. But he has by now achieved his beginning, an immense gain, for Croesus leads to Cyrus, and Cyrus to the grand Persian progress which will be a frame to help support the main story and all its attached excursuses. And besides, there was a sense in which Croesus, who let the Persians loose on the Greeks, was far more responsible for the ensuing trouble than Gyges. Herodotus was right, but he did not choose the right language. Gyges actually had attacked them too.…

Herodotus has now made his correction and acknowledged the error. The reader who goes on reading, the listener who goes on listening, has no right to complain that he has been misled, because he has been set right. So we can get on with the logos.

We can trace the same method elsewhere. "Alyattes … had other achievements to show for his rule, notably this. He fought the Milesians, this being a war he took over from his father. For he [antecedent not stated] invaded that territory and besieged Miletus in the following way … In this way he [antecedent again not stated] fought them for eleven years … Now for six of the eleven years it was Sadyattes, still ruling the Lydians, who kept leading these expeditions into Milesian territory, for it was he, Sadyattes, who had started the war; but for the five years that followed the six it was Alyattes the son of Sadyattes.…" Here it would have taken only a minor operation to restore the first six years to Sadyattes, who has been dealt with immediately above, and then go on with Alyattes. Failure to do this makes a rather messy and self-conscious progress for a page or so. Herodotus did not cancel or erase. Did it ever occur to him to do so? He goes straight on writing, but does not leave the subject until he has left the reader with the right facts: six years for Sadyattes, who started it, and five years for Alyattes, who finished it.

"Besides, if Heracles was alone, and besides that, if he was only a mortal man, as they say he was, how could it be within nature for him to have killed many tens of thousands? Now that we have said as much as we have said, I hope neither the gods nor the heroes will be angry with us." This is a sort of overpass quite in the manner of Pindar. The trouble-some thing said is unsaid not by cancellation but by apology which follows on the saying.

"The Libyan nomads make sacrifice as follows … They sacrifice to the Sun and Moon and to them only. All Libyans sacrifice to these gods; but those who range … around the Lagoon of Tritonis sacrifice to Athene chiefly, and after her to Triton and Poseidon." The antecedent of "they" (Libyans or nomad Libyans?) is uncertain, nor is it clear whether the Libyans of Tritonis are included among the nomad Libyans or not.… It seems pretty plain that the sacrifice to Athene and the others was not in the mind when the phrase about sacrificing to Sun and Moon only was written; the exception occurred to Herodotus immediately and he set it down.

" … Aristagoras made an expedition against Sardis. Now he himself did not go on the expedition but stayed in Miletus." The first phrase by itself … would naturally imply that he did go, so this is a correction-instride.

"These nations all have cavalry, except on this occasion they did not provide cavalry, but only the following ones did." There is no real contradiction, but Herodotus writes as if he were correcting himself. The word "except" … shows that.

" … the biggest single national group he [Mardonius] selected was the Persian force, men who wear necklaces and bracelets: the next largest was the Medes; actually there were as many Medes as there were Persians, but they were not such strong men." We can now see that this is a classic example of correction-in-stride.

Dislocated additions. In the above cases, the sign of progressive style is the acknowledged or at least recognized contradiction and the means taken to prevent such contradiction from leaving the reader or listener falsely informed. There are also unrecognized contradictions, often noted, but these are only what we should expect to find in an early work of this magnitude, in whatever way the parts were composed, and whether or not the entire draft was revised. Correction-in-stride, on the other hand, cannot well indicate revision in the modern manner, for there is no point to it.

Alteration of content in the process of composition is also, I think, betrayed when an addition to the text is put, not where it logically belongs, but apparently at the point the author had reached when the thought of adding occurred to him.

"Next came a thousand spearmen, who were also a force picked out of the entire army. They carried their spears with the points held down toward the ground (7. 40. 2) … Immediately behind Xerxes came a thousand spearmen, the bravest and highest born of the Persians. They carried their spears in the normal position. After these came another picked force of a thousand cavalry. After the cavalry came ten thousand men chosen out of the Persian forces. This was an infantry force. A thousand of these had golden pomegranates instead of spikes on the butts of their spears, and these were formed so as to enclose the rest of the force; and the other nine thousand who were thus enclosed had silver pomegranates. Those who carried their spears with the points held down toward the ground also had golden pomegranates on their spears, and those who immediately followed Xerxes had apples" (7. 41. 1-2). This is obviously akin to the correction-in-stride, and might be called an addition-in-stride. A systematic revision made later than the original draft might better have put this information in the place (so near!) where it belonged.

"Croesus was dissatisfied with the size of his army, for the force he had used in the battle was far smaller than that of Cyrus; and being thus dissatisfied, when the next day came and Cyrus made no move to attack him, he marched back to Sardis. He had it in mind to call up the Egyptian forces in accordance with their sworn agreement (for he had made an alliance with Amasis, king of Egypt, before he made one with the Lacedaemonians), and to call for the Babylonians too (for he had an alliance made with them also, and the ruler of the Babylonians at this time was Labynetus), and to request the Lacedaemonians to be there at a stated time.… (1. 77. 1-3)." The narrative struggles through a series of after-thoughts hung on gar-clauses. They could be marginal corrections, but if so why are they here? The note on numbers belongs just back at the account of the battle (76. 3 or 4); and the alliances should come at 53. 3 or somewhere in 69-70, where alliances are under consideration. From analogies observed, it is more likely that the Egyptian and Babylonian alliances were hitherto forced from Herodotus' attention by the interest inherent in the Lacedaemonian alliance, and recalled to mind only by the context of Croesus' proposed winter withdrawal. (As for Croesus, he seems from this to have fought in pretty much the same way Herodotus wrote, meeting the occasion as it came along. Ce n'est pas la guerre. Croesus did not win.)

"As the Lacedaemonians say, they made the expedition not so much because they wanted to help the Samians in their need as because they wished to avenge themselves for the theft of the mixing bowl which they had tried to convey to Croesus, and for the corselet, which Amasis the King of Egypt sent them as a gift. For the Samians had stolen the corselet too, the year before they stole the mixing bowl … There is another one like it which Amasis dedicated to Athene in Lindos" (3. 47). The story of the mixing bowl has been explicitly told at 1. 70; and the language in this passage makes it pretty clear that Herodotus remembers dealing with it, whether or not he had his finger on that place in his own manuscript. The technique of "Schluβredaktion" would have shifted 1. 70 (in an adapted version) to the Samian passage or (less happily) 3. 47 to the Croesus passage; the technique used is that of oral discourse, and the thought is something like: "you remember the mixing bowl lost by robbery on its way to Croesus; well, they stole a corselet too."

This type of addition with gar-clause may thus, I think, be taken as supplementary evidence, less cogent than the true correction-in-stride, for a process of composition which worked always forward. The passages have been taken in isolation and in part as negative check cases against revision, but this keeps us from seeing how the progressive manner builds a positive style, which is to be seen only in an extended sequence. One aspect of this is a tendency to follow out the implications carried in single words, so that we can pick out guide words that direct the progress for extended duration.

Consider, for instance, the stretch from 3. 4. 3-3. 9. 4. When Cambyses was resolved to conquer Egypt, he was puzzled by the problem of crossing the waterless stretch on the way. Phanes of Halicarnassus advised him to win safe conduct from the King of the Arabians, through whose domain lay Cambyses' only entrance to Egypt. This entrance is then indicated, and the account closes with a note that a part of the route, "no small distance but three days' journeying," is dangerously waterless. This leads to a description of the system by which the Persians now keep this area watered; but the system was not in operation until the Persians came, and now Cambyses must acquire the good offices of the Arabian king by giving and taking pledges. The Arabian ceremony for pledges is then described. It involves a ritual of bloodletting, and the invocation of Dionysus and Ourania. "These are the only gods whose existence they believe in, and they say that their way of cutting their hair short is after the way Dionysus has his hair cut. They cut it so that it is round on top like a wheel, shaving up to the level of the temples. Their name for Dionysus is Orotalt and for Ourania, Alialat." We return to the measures taken by the Arab after the pledges were given and taken, and end with an appendix devoted to a "less easily believed version"; and can now (3. 10) resume the account of the war.

This passage has been called the Arabian Logos; and if there is any such thing as an Arabian Logos, this must be it. But it is no organized, potentially freestanding anthropology of Arabia or the Arabians, rather a sequence of notices which grows organically out of its place of occurrence in the Persian progress. The information given—and doubtless more information about Arabia and Arabs as well—was held in the magnificent memory of Herodotus, until the waterless stretch that faced Cambyses led, on the guide words given above, as far as Orotalt and Alilat, and no further.

The body of material which grows out of the watering problem comes to medium length. Analogous sequences can be found, not different in kind, which range in bulk from the single-sentence gloss parenthetically put or hung on a gar-clause, which is generated direct from an idea or single word such as a spoken aside in the forward progress of live speech, to really extensive stretches of complex narrative which have not been fixed on the sustaining frame of a definite journal. Examples of the former are altogether too numerous to be worth listing; an example, at random, would be 5. 119. 2: "Those [Carians] who escaped were presently trapped at Labraunda in the sanctuary of Zeus Stratios, which is a large and hallowed grove of plane trees. The Carians are the only people we know of who make sacrifices to Zeus Stratios. When they found themselves trapped in this place, they made plans …" The long form might be shown if we enter the text at 5. 38. 2, where Aristagoras arrives at Sparta; and we are thus transported from the main line of the narrative, which has reached Miletus and the Ionian Revolt, and then pick our way forward, following the guide words and phrases until we come back to the Revolt at 5. 97. 3. We shall by then have been told how Cleomenes, not Dorieus, was then King of Sparta, and of the adventures of disappointed Dorieus (39-48); of the interview of Aristagoras and Cleomenes, with the use of that "map of the world displayed on a board which he carried around with him" (49-51); of the royal road (52-54); of Aristagoras in Athens (55. 1); and thereto the story of how the Athenians, who when last extensively mentioned had been in bondage to the tyrants or scattered abroad to escape them (1. 65. 1), had by now got rid of their tyranny, escaped the designs of Cleomenes, beaten the Boeotians and Chalcidians and been involved in hostilities with Aegina; further how a movement to restore Hippias was foiled by Socleēs of Corinth with the story of Cypselus and Periander (55-59); and finally, of the effort of the Athenians to stop Persia from supporting Hippias, their failure and consequent state of mind, as a result of which they let Aristagoras persuade them to send help to the Ionians (96-97).

After the Persian progress toward ultimate collision with Greece has struggled along a far from open course through Scythia, Libya, and Thrace, Herodotus has almost formally announced a new phase of the warmaking (5. 28. 1) and moved forward in a businesslike manner until Aristagoras in Greece embarks him on the diversions indicated above. One might say that throughout he keeps a firm hold on the leitmotiv, as indicated at 49-51, 55. 1, and perhaps 96. 1; but I cannot think that the excursus was meant to be so long or so complex. Rather, the first of it grows out of a gar-clause to elaborate "not through being more of a man but through birth" (39. 1); and after returning at least close to the matter of the lonian Revolt, Herodotus enters on the Athenian account by "Athens, now free of tyrants, thus" (55. 1). Further, there are excursuses within excursuses, as for instance when the reforms of Cleisthenes in Athens suggest the reforms of the elder Cleisthenes in Sicyon (gar-clause transition) and take us as far from Persia, Ionia, and 500 B.C. as the legendary base of the cult of Melanippus in Sicyon; or as when Socle s' speech against restoring Hippias brings us to the circle of Corinthian conspirators who pass the laughing baby tyrant from hand to hand and none has the heart to smash him on the ground.

Revision or addition under the circumstances in which Herodotus wrote could hardly have achieved this Chinese-puzzle pattern of minor parts within parts. It is rather what we should expect from a composition in the progressive style in which—subject to a guiding plan and a main line of narrative—the details are to some extent permitted to generate themselves.

Parallels in early composition. It is now generally conceded that there can be literature before there is writing. Grant that; then grant also that, even after writing becomes available to the composers of literature, there will still be a period in which the composer writes his work, or causes it to be written, but is not yet familiar or easy enough in the process to be able to think in terms of writing. The oral method of composition will carry over for some time; and for some time, too, the composition will be communicated by recitation or by reading aloud, not by silent reading to oneself with eye and mind alone.

Although the Homeric poems were undoubtedly Herodotus' chief model for composition on the scale and scope of the History, I can make little use of these for parallels in the progressive style of composition. The unity of composition for the Iliad and Odyssey, even the authority for any given text, is still in dispute; and even if we grant that our text (approximately) of the Iliad is the authentic work of one author, various considerations, particularly the traditional nature of Homer's material, disturb the analogy. We may, however, cite a few cases of the misplaced addition or explanation, mostly having to do with the larger progress of the campaign at Troy; for instance, the non-heroic armor and weapons of the Locrians (13. 712-22), the peculiar arrangement of the Achaean ships (14. 29-36), the presence of such specialists as steersmen and supplycorps personnel in the Achaean army (19. 42-45), come into the poem late, as if thought of late and then immediately set down. In general, Homer can best be used to interpret Herodotus by referring from specific passages in Herodotus back to Homer.

Archaic poetry contains several cases which seem to make it clear that the author in question composed and wrote down his piece straightforward without working back. Solon in his elegy addressed to the Muses prays for prosperity to be got honestly, since criminal action immediately draws down the punishment of the gods; it is immediately seen that this is not true; God may wait, even past the lifetime of the wrongdoer, and strike his innocent descendants or fellow citizens. The rejected proposition is left in the text, being in fact used to further the progress of the poem. Pindar can mention the horrible story of Pelops in the stewpot or of Heracles thrashing major divinities, reject or at least deprecate the allusion, but leave it standing in the text. This is contradiction and correction-in-stride. Unpremeditated excursus, another sign of oral thinking in composition, also appears. Theognis starts an elegy with the reflection that there is only one "virtue" which does people any good, and that is "having money." He is of course parodying Tyrtaeus 9 and could, like his original, progress to the end and repeat his point, clinging firmly to the familiar frame of the catalogue; but when he has progressed from Rhadamanthys to Sisyphus, he allows the clue of that name to lead him down with Sisyphus to Hades and Persephone, whence no mortal ever returned except Sisyphus, before he returns to his catalogue and finishes his poem. Elsewhere he simply drifts and winds up talking about something quite different from what he began with.

Such passages as the above do not perhaps prove that systematic rewriting and revision were unknown to Greek literature until a relatively late period. They are, however, what we should expect to find in the work of gifted writers who rejected, or did not know of, that kind of revision. If we remember the difficulties of practicing such revision on any sizable scale with the materials available at the time, perhaps we should not ask whether Herodotus and his predecessors wrote in a continuous forward sequence, but whether they could have written in any other way. If they could not, we should hardly be surprised to find the sorts of anomaly and irrelevance noted above.

The composition of Herodotus: The major scheme. But the complexity, the sheer size of the work of Herodotus, is prodigious compared with that of all predecessors except Homer. Beyond the contradictions and the sometimes wayward excursions we have noted and paralleled, we ought to find, in a forward-written first draft, faults in the major scheme, cases of material which he should have presented, or has implied he will present, and has not presented.

Squeeze-out. We do find such cases. Sometimes the material in question ultimately finds its way into the text; at other times it is lost for good. After the capture of Thasos, Darius "sent various heralds directed to various parts of Greece, with orders to claim earth and water for the King" (6. 48. 2). Other heralds went to his own seaboard cities to order a shipbuilding program. "When the heralds arrived in Greece many of the mainlanders gave them what the Persian demanded, and all the islanders to whom they came with their demands did so. Now with the rest of the islanders giving earth and water to Darius, the Aeginetans also did so; and when they did, the Athenians immediately, etc." (6. 49. 1-2). The Athenian action leads into a long excursus on affairs in the Greek homeland involving an account of the Spartan kingship and the Spartan kings, with a part of the career of Cleomenes and his death, and returns to the main narrative at 6. 94. The heralds marked the point of entry from the Persian Journal to the excursus; we do not hear of their fate at Athens and in Sparta until 7. 133, but there they do make their delayed appearance. But the men-de clause of the text involves "many of the mainlanders" and "all the islanders to whom they came with their demands." We are not to know what mainlanders, for the men-clause is left, the de-clause (islanders) developed. But even these are developed by the figure of all s-te-kai, so that the Aeginetans exclude "the rest of the islanders" and of these, too, we hear no more. This type of squeeze-out may be noted elsewhere, as in the exclusion of "the other big cities of Assyria"; the loss of the subsequent fate of the Paeonians around Pangaeum and Lake Prasias "who were not conquered at first by Megabazus"; the failure to narrate the Persian annexation of Macedonia, for which 5. 21 is inadequate; the failure to record the fortunes of the Carians between their great victory at Pedasus and their ultimate submission; the disappearance of the 4000 Athenian cleruchs from Chalcis; the failure to note the message for help sent to the Plataeans; the disappearance from the narrative of the Persian land force which marched on the Peloponnese; of the "many other renowned Persians and Medes" besides Ariabignes lost at Salamis; and of the Corinthian group at Plataea; Exclusion may be deliberate, as at 7. 99. 1: "I shall not give the names of the other taxiarchs, since I am not forced to do so, but I shall speak of Artemisia …" Mostly, examination of the text will show that the squeezing out is a result of the style, where in a clause of the men-de type the second member is carried forward and the first thus lost in transit.

Unfulfilled promises. These differ from the foregoing cases mainly in that Herodotus has assured us that he means to deal with the material excluded. The cases, well-known and much discussed, include the promise to deal with other kings of Babylon "where I shall be dealing with the Assyrians" (1. 184; polloi alloi and men-de construction; exclusion by interest in queens, and Nitocris and her wall?); the promise to give an account of the capture of Nineveh; the promise to give the reason why Athenades killed Epialtes "to be told in a later part of my story"). If these unfulfilments of promise rise, like the organic squeeze-out, out of the circumstances of writing, we should only be surprised not to find more of them. But Herodotus, though a writer, was able to control enormous masses of material in his head. He had been exposed to only the first phases of the sickness of Thoth, and the art of writing had not destroyed his memory or even impaired it much.

The problem and ideas of order. He might, indeed, have done even better if he had been able to work out more completely in advance a framework which would help to fix the order of his History. If the view I have been detailing is right and the first sentence in the History is rightly placed and represents the true beginning of the composition, Herodotus was faced with an unprecedented problem: to write the history of the Persian attack on Greece and the Greek resistance, with full notes on the peoples, barbarian and Greek, involved. The main line must be an orderly chronicle of the Persian progress, the only part of the complex which could be viewed as a continuous narrative. But barbarian attacks on Greece began with Lydia. From Croesus he found his way back to Gyges, as we have seen, and from Gyges forward again to Croesus through a chronicle of the Lydian kings dominated by the idea of recording their dealings with Greek cities. Then Croesus faces Cyrus and goes down, and from Cyrus Herodotus works, again, back to the beginning of Median power, thence forward via Median chronicle to Cyrus. From Cyrus, the main line is secured by the succession of Persian kings. Follow them, and you will finally come to the great invasion.

This was the main frame. To it adheres, in the first line, the succession of notices concerning peoples little known (or, Herodotus would say, inaccurately reported) to Greek audiences; and also a succession of notices concerning Greek states which can be joined directly to the main line of the History or joined to other notices which join to other notices—this procedure can be continued indefinitely. For these two types of major excursus no orderly over-all frame was available. The first would demand a written periodos gēs; the second, a history of Greece. Supposing Herodotus could have written either or both, he would have had three works on his hands instead of one. Instead, the inclusion of both types of excursus depends on finding a point of departure in the body of the Persian progress. But there is a difference once inclusion has been started. The barbarian notices can be organized into subdivisions treating of geography, ethnography and, if practicable, history. For the Greek, there is no such organization possible, and we depend far more on the comparatively casual guideword and generated series.

Thus it happens that no real "Greek Journal" is worked out until the time of the great invasion, when the Greeks acted more as a single group than ever before. The method is rather one of "keeping up with the Greeks" as they make contact with the Persian progress. Occasionally we find a summarizing sentence, which seems to show consciousness that the whole History could be more thoroughly organized under a series of headings. "Thus Ionia was enslaved for the second time" and "thus the Ionians were enslaved for the third time, for they had been enslaved first by the Lydians and twice, now, by the Persians"; such statements suggest a schematic control of material. But it is not followed out, merely noted down as it occurred in the process of writing; for the first such note comes at 1. 92. 1 (following the fall of Croesus) not at 1. 27. 1 (conquest by Lydians) and reads: "Thus it was with the reign of Croesus and the first overthrow of Ionia." There are also rudimentary signs of recognition that a systematic chronology, and even synchronization of events or biographies (1. 12. 1; 1. 23), would be helpful. Two years are logged for the interval between the death of Atys and the preparations to attack Cyrus (1. 46. 1); an attempt to keep firm hold on the chronology is evident in the sequence following the fall of Miletus, but it is considerably disturbed by the cutback to the former career of Miltiades; an annalistic approach is indicated at 9. 121: "and during that year nothing further happened," but by now the work is all but concluded. Had the work proceeded from this point, an annalistic scheme might (in theory) have been followed, from thence forward; but an annalistic scheme would have been alien to the methods of Herodotus. There is a feeling of time, but no scheme; some time notices are afterthoughts with gar-clause or as parenthesis.

The progress of the History. In view of all the above considerations, and in view of the physical difficulties of composition and the still greater difficulty of revising composition, I think it best to regard the History as a document set down as it was composed and composed in substantially the same order in which we now have it. This does not mean that there could have been no deletions or marginalia whatever, it does mean that to posit such functions of revision is neither a valid nor a practicable approach to the study of the History. If the foregoing is right, we may draw some conclusions about the time and place of writing.

Born probably between 490 and 480 B.C., Herodotus set out from Athens to Thurii, probably about 443 B.C. It is not certain that he ever returned to Greece proper, though he may have done so. Between these two (tentative!) dates, he apparently participated in the expulsion of the tyrant Lygdamis from Halicarnassus, but subsequently left his own city for good; spent some time in Samos, traveled extensively, with visits to Egypt, Libya, the Black Sea and the coast of Scythia, and many parts of the Greekspeaking world including Macedonia and the coast of Thrace.

It seems most likely that he began to write the History at some time during the period of 450-444, and that he did so in Greece, probably in Athens (he would have set out from there for Thurii), at a time when his great period of traveling was behind him. Certainly he had been at Delphi before he wrote 1. 14. 2-3, which shows absolute familiarity with the offerings there. How long the idea had been in his mind before he began to write is something we can only guess at. The work was still in progress when he left, and had probably not progressed beyond or much beyond the first three books; for the early extant plays of Sophocles, Aiax and Antigone, show the influence of Herodotus strongly, but definite parallels can be established only with Books 1 and 3. He is also said to have given readings at Olympia, and to have read aloud to the Athenians, and been rewarded for it, in 445/4. The readings would then have been from the first part of his work; or else they were not readings, but lectures from memory.

There is no conclusive evidence that Herodotus ever returned from the west to Athens or any other place in mainland Greece. The latest event which is absolutely datable and referred to in his work is the seizure and execution of Aristeas and other ambassadors in 430 B.C. He died, therefore, at some time after this date; but if my view about composition is right, he lived long enough after it to complete 7. 138-9. 122, which was therefore composed after 430 and probably composed at Thurii or elsewhere in the Greek West.

When we use Herodotus as our main source for the reconstruction of Xerxes' invasion and the great battles of 480-479, it may therefore be helpful if we bear in mind (always granting the correctness of the foregoing) that his accounts were set down at least twelve or thirteen years after most of the material was collected and the sites visited, when and if they were visited at all; that they were probably set down from memory, without adequate notes, certainly without those topographical maps which modern scholarship requires; that they were subject to all the vicissitudes of omission and diversion involved in the process of composition indicated above; and finally, that they are the work of a historian who, as far as we know, had never seen a fullscale battle, who had more epic schooling than military imagination, and who was setting forth to write history before history was history. All this we must keep before our eyes as we study Herodotus the historian of the Persian Wars.