Site Section Links
KRONOS Vol X, No. 2
THE VENUS TABLETS: A FRESH APPROACH?
LYNN E. ROSE AND RAYMOND C. VAUGHAN
Copyright (c) 1984 by Lynn E. Rose and Raymond C. Vaughan
In an article entitled "The Venus Tablets: A Fresh Approach", Journal for the History of Astronomy, XIII (1982), pages 23-49, John D. Weir reviews previous studies of the so-called "Venus Tablets of Ammisaduqa" and offers "a fresh approach" that favors the long chronology (-1701 to -1680 for the reign of Ammisaduqa). Fifty years ago, an even longer long chronology was defended by Langdon, Fotheringham, and Schoch (hereafter LFS), who favored -1920 to -1899. Ever since Kugler attributed the Venus tablets to the reign of Ammisaduqa, the aim of investigators has been to find a retrocalculated sequence of disappearances and appearances of Venus that will match the sequence of disappearances and appearances that is reported on the Venus tablets. This use of the Venus tablets as a means of dating the reign of Ammisaduqa is generally seen as the only exact basis for the second millennium chronology of the entire Middle East. For once Ammisaduqa is dated, and the First Babylonian Dynasty with him, the chronology of that entire region of the world is supposedly placed on a firm footing. Even those who favor this approach, however, have often been unable to decide among the long chronology (-1701 to -1680), the short chronology (-1581 to-1560), and the two middle chronologies (-1645 to -1624 and -1637 to -1616).
In what follows, we shall argue that Weir's approach to these issues is neither fresh nor right. Not the least of our criticisms of Weir's article is that, in his review of the literature, he seems totally unaware of our own work in this area, even though he cites and discusses one paper that does at least mention our work. Weir could have saved himself from many errors if he had but consulted some of our writings on the Venus tablets.
Weir's "fresh approach" involves use of Peter J. Huber's retrocalculations of the Julian dates of the disappearances and appearances of Venus at Babylon in the second millennium. These retrocalculations of Huber's are supposed to be somewhat more accurate than older retrocalculations in that they take better account of tiny secular changes in the length of the day and the length of the month. Weir and Huber also make use of attested thirty-day lunar months from the reign of Ammisaduqa, which, along with attested intercalary months from the reign of Ammisaduqa, serve them as a guide in translating the retrocalculated Julian dates into Babylonian lunar dates. The retrocalculated Babylonian dates are then compared to an edited version of the Venus observations. (Actually, Huber does not tell us the retrocalculated Babylonian dates that he is using, but expresses his results in terms of implied values of the arcus visionis. )
Weir argues that, while discrepancies remain, the least objectionable match is with the long chronology. (In 1972 Weir had favored the middle chronology of -1645 to -1624.) Perhaps it is no coincidence that Huber himself has also come out in favor of the long chronology. When Huber was criticizing Velikovsky in 1974 (Huber's paper was published in 1977 in Scientists Confront Velikovsky), he argued that the data supported the short chronology (- 1581 to - 1560 for Ammisaduqa's reign). In 1982, in a monograph not mentioned by Weir - presumably because Weir's own paper was written well before Huber's monograph was published (though both appeared in 1982) - Huber switched to the long chronology. What effect, if any, this switch has on Huber's arguments against Velikovsky is something about which Huber has remained silent. In any case, no new Venus tablets, and no new information about the text, led to that switch. It is simply that Huber significantly reduced the number of observations from the tablets that he used in his studies (35 in 1974-1977, and only 30 in 1982). He did correct a few of his own earlier errors, such as in Year 5a, but it seems to have been the trimming of the reports down to 30 that led to this fortuitous and illusory focus on the long chronology. (Huber claims to have used 31 reports, but he is counting one that he rewrote by nine months; it is hardly plausible to classify that item along with "observations" or "data", as he does. His claim in his 1982 monograph (pages 23, 43) that he had earlier used 38 seems to be a scribal error on his part. He should have said 36 - or 35, since that same change of nine months was involved.)
For all practical purposes, the work of Weir and the work of Huber are so similar as to be virtually identical. It is difficult to say who influenced whom, but it may be noteworthy that Weir acknowledges the help of Huber, while Huber does not acknowledge having received any significant help from Weir. In this paper we shall be dealing both with Weir and with Huber.
It cannot be overemphasized that neither Huber nor Weir (nor other authors such as Reiner and Pingree, van der Waerden, LFS, Kugler, Schiaparelli, and Sayce) ever attempted, even for a moment, to make astronomical sense of the data as they stand. For more than a century now it has been the almost-universal practice to work only with censored and astronomically edited "data". Which observations are deleted or edited will vary slightly from author to author, but no author who tries to match the observations to retrocalculations has ever worked with the full set of unedited data, that is, the data as recorded rather than as molded to astronomical expectations.
To be sure, the unedited data do present textual problems of many sorts: damaged tablets, difficult readings, conflicting readings, apparent scribal errors, and so on. But all of these are routine matters in cuneiform studies and have little or nothing to do with astronomy. We sharply distinguish between these textual problems and the entirely different concerns that have led to an astronomical censoring or editing of the data. Let us use Year 12 as an example. All six of the surviving tablets that deal with Year 12 support or are consistent with an invisibility lasting from Nisan (the first month of the Babylonian calendar) to Ulul (the sixth month). The several tablets show slight variations in the length of the invisibility, with either 5m 16d, 5m 17d, or 5m 18d, and there are also slight variations in the dates of disappearance, with either I 8 or I 9, and in the dates of appearance, with either VI 24 or VI 25. Some, like Huber, simply omit Year 12 from their calculations. Others, like LFS (pages 105-106), engage in radical rewriting of the data. They just change I 9 (5m 16d) VI 25 to II 29 (2m 6d) V 5; then, not by coincidence, they find that the "observation, thus restored, is excellent".
Our own efforts have been directed both to the textual problems and to the astronomical problems. Our purpose here is not to argue for our textual reconstruction; that has already been done in our previous papers, especially "Analysis" and "Artificial Insertion" (KRONOS II:2 and V:4). Rather, we shall simply list below, for purposes of the present discussion, our determination of the best readings of the Venus tablets. The year numbers in the left-hand column are a modern convention, which we follow just as a matter of convenience: each year number is assigned on the basis of the year in which the disappearance occurred, and two disappearances in the same year are distinguished as a and b. It should be noted that there is no textual support whatsoever for the placement of "Years 19 to 21b" immediately after "Years 1 to 17", with only the hypothetical "Year 18" intervening; the relative order of those groups of years and the supposition of a "Year 18" are simply modern conventions, too. In what follows, we give the year number, the direction and date of disappearance, the interval of invisibility, and the date and direction of appearance. "VI*" indicates a second or intercalary Ulul.
The preceding represents our best assessment of the testimony of the surviving tablets. Some readings are non-controversial, such as Year 7, where all five surviving tablets are consistent with one another. In other cases, such as Years 8b, 13b, 14, and 16b, our reconstructions are based on the overall weight of the surviving texts, but are more tentative. Nevertheless, we have argued in our previous papers that each of these readings is better supported than any alternative reading. We do of course assume from the general pattern of the observations that Venus was an inner planet, requiring strict alternation of superior and inferior conjunctions. Aside from that, however, no astronomical expectations have influenced our reconstruction, which is based solely on the relative weight of the supporting texts.
Such, then, is the observational record as it has emerged from our essentially textual analysis. How much of that material is actually used by those who would compare their astronomical retrocalculations to the observational record? Much less than is generally supposed. Many observations, no matter how well attested, are either discarded or radically rewritten - or else the fact that they contradict the retrocalculations is overlooked. Thus LFS radically rewrite Year 12, while Huber discards it, and Weir is unable to make any use of it. Whatever the approach, the recorded data are being carefully left out of consideration.
When the observations are in one way or another purged of whatever will conflict with retrocalculation, what can be learned from the relationship between retrocalculation and the observations that remain? We suggest that the only rational answer is, Nothing. If that "relationship" in one way works out a little better for the long chronology and in another way works out a little better for the short chronology, what do the data show about the date of Ammisaduqa? We again suggest that the only rational answer is, Nothing. The actual data play too small a role for any meaningful results to emerge.
The case for attributing the Venus tablets to the reign of Ammisaduqa has been severely eroded over the years, though this erosion is usually not appreciated or mentioned. Weir and Huber do at least note the erosion, insofar as they indicate that Years 9 to 17 and Years 19 to 21b fit rather poorly with retrocalculation, and insofar as they raise the possibility that various entries on the Venus tablets (Years 19 to 21b in particular) may accordingly not be from the reign of Ammisaduqa at all. But Weir and Huber seem not to appreciate the significance of this. One of the two principal traditional reasons for attributing these Venus tablets to the reign of Ammisaduqa was that the tablets were supposed to cover twenty-one consecutive years, the exact number of years that Ammisaduqa ruled. But if "Years 19 to 21b" are not from the same stretch of observations, then we no longer have a set of observations covering twenty-one consecutive years. We are left only with the Year 8b year-formula, "Year of the Golden Throne", which corresponds to a year-formula for Ammisaduqa's eighth year. This evidence is not negligible, but neither is it very compelling, especially if it has to bear the entire weight of second millennium chronology. After all, year formulas were sometimes used by more than one king, and Hommel and others have argued that the Year 8b year-formula is a later insertion anyway. If one agrees with Hommel and the others about the year-formula being a later insertion, and if one agrees with those who see Years 19 to 21b and perhaps even Years 9 to 17 as misplaced (Huber scraps all of Years 19 to 21b and either scraps or rewrites nearly half of Years 9 to 17; Weir decides to keep these entries, but finds them unhelpful), then there remains nothing of the original case for attributing the Venus tablets to the reign of Ammisaduqa. These difficulties are discussed in our own writings, and are also the principal subject of a paper by the late Livio C. Stecchini, "The Twenty-One Years of Venus" (KRONOS VII:3), which was published about the same time as Weir's paper.
It may be that the "data", as edited by Weir and Huber, do now fit the long chronology better than the short chronology or any of the middle chronologies. But "better" is highly relative. Is the fit good? No. Even rewritten "data" fit the retrocalculated events pertinent to the long chronology very poorly. And the actual data are in such conflict with those retrocalculations that almost anyone would be embarrassed to introduce a word like "fit". In one way or another, Weir and Huber have to end up rejecting more than half of the Venus observations: those observations simply do not conform to the requirements of retrocalculation.
Weir and Huber do not flaunt the failures of their approach, but those failures can be detected by the careful reader. An earlier instance of this had to do with Huber's 1977 paper. In the second installment of " 'Just Plainly Wrong': A Critique of Peter Huber" (KRONOS IV:2), it was noted that, for the short chronology that Huber then favored, sixty percent of his retrocalculations failed to fit the observations as recorded on the tablets. Consideration of the attested intercalary months raised that failure rate to seventy-six percent.
Using somewhat less strict criteria than were used in "Just Plainly Wrong", we shall now see how Weir and Huber have fared in their 1982 efforts. In Weir's case, this means noting how many observations are deleted or rewritten, and how many of the remaining observations involve discrepancies that by any reasonable standard (such as the very standards otherwise espoused by Weir and Huber) are unacceptable. More specifically, notice in Weir's Table 3A on page 34 that many recorded disappearances and appearances fall during the computed interval of invisibility, that many recorded disappearances precede the computed disappearances by more than the few days or so that could be explained by bad weather, and that many recorded appearances follow the computed appearances by more than the few days or so that could be explained by bad weather. Notice also that at least one of the discrepancies is acknowledged in the text but not in the table itself: Weir, like Huber, has simply altered the record for Year 9. (We recognize the astronomical implausibility of the long invisibility in Year 9, yet its beginning, duration, and end are all well documented. Our objection is to the self-serving nature of the choices made by Weir and Huber: which data to reject and how to rewrite them.)
In general, let us suppose that bad weather might advance the beginning of an invisibility by up to five days and that exceptionally good viewing conditions might delay it by one day. Let us also suppose that bad weather might delay the ending of an invisibility by up to five days and that exceptionally good viewing conditions might advance it by one day. This six-day interval that we are allowing Weir at either or both ends of an invisibility is actually rather generous: Reiner and Pingree, for example, allow themselves only about half that much leeway (see page 15 of their 1975 monograph). When an invisibility ends before its computed beginning, or begins after its computed ending, we are counting both the disappearance and the appearance as misses.
We invite interested readers to examine the "Computed Babylonian dates" in Weir's Table 3A; if they are compared to the allegedly "Recorded Babylonian dates" in that same Table 3A, there is a failure rate of forty-eight percent. So that readers may easily check our results for themselves, we shall use the following concise format for summarizing the hits and misses from the disappearance in Year l to the appearance in Year 17 (top line) and from the disappearance in Year 19 to the appearance in Year 21b (bottom line):
hh mh mh mm mm mh hh hh mm hm hh hh hm mm mm hm mh hh hh hm hh mm mh mh mm
Note that forty-eight percent of these are misses. That may not be quite as bad as the failure rate for the other, shorter chronologies, but it is nevertheless a very poor "fit". If the "Computed Babylonian dates" are compared to our own textual summary and reconstruction given earlier in this paper, we find a failure rate of sixty percent:
hh mh mh mm mm hh hh hh mm mm mh hh hm mm mm mm mh hm hh hm mm mm mm mh mm
Huber's situation is very similar to Weir's, although Huber's results are stated in terms of implied values of the arcus visionis. Let us count anything between 4.5 degrees and 8.0 degrees as acceptable. This is about the same as the range that we used with Weir. If the "data" in Huber'sTable 4.2 on pages 16-17 are used, his failure rate is forty-six percent:
hh hh mm hh mm hh hh hh mm hh hh hh hm mm mm hm mh hh mh hm mm mm hm mh mm
If our own data are used, then his failure rate rises to fifty-two percent:
hh hh mm hh mm hh hh hh mm mm mh hh hm mm mm mm mh hh mh hh mm mm hm mh mm
Weir follows Huber in rejecting "14U". That is, both of them throw out the attested intercalary Ulul in the fourteenth year of Ammisaduqa, because it spoils their desired "fit". In his 1974-1977 work, Huber even introduced a 13U, that is, a completely unattested intercalary Ulul for the thirteenth year, because it seemed necessary in order to preserve any semblance of the sort of "fit" that he sought. He at one time tried to ignore the attested 13A, that is, the intercalary Adar in the thirteenth year. Now, by rewriting Year 13b, he has managed to keep the 13A intercalation. Even with all these maneuvers, however, the fit remains poor.
As we have seen, the "fits" achieved by Weir and Huber do not amount to much anyway. And if they had used the attested Ulul in Year 14, their plight would have worsened considerably. The failure rate of Weir's Table 3A would rise to seventy percent:
hh mh mh mm mm hh hh hh mm mm mh hh hm mm mm mm hm mm mm mm mm mm mm mm mm
Huber's failure rate in Table 4.2 would rise to sixty-six percent:
hh hh mm hh mm hh hh hh mm mm mh hh hm mm mm mm hm mm mm mm mm mm mm mm mm
Our comments have been based on what Weir and Huber chose to publish in their 1982 writings. We have noticed, however, some discrepancies and obscurities that we cannot clear up until Weir and Huber reveal more about their pre-publication exchanges. In the meantime, let the reader beware of the Weir-and/or-Huber treatment of Years 4, 12, and 13b. The main problem is that while Weir's "Computed Babylonian dates" in Table 3A are supposed to be derived from Huber's, they are sometimes quite different from the dates that would be implied by Huber's Table 4.2.
Weir and Huber do some frantic scrambling in an effort to dispose of the intercalary month 14U. Schnabel reported finding such an intercalation; he also reported finding a 5U (see LFS, page 61). In neither case do LFS identify Schnabel's source. Those who trust Schnabel should accept both. Those who distrust Schnabel should reject both. But Weir and Huber will never reject 5U, no matter how weak its documentation. It is true that in two respects the 5U might be slightly more acceptable than the 14U. First, Schnabel told Langdon that he found two sources attesting 5U, and only one source attesting 14U. Second, as Weir points out on page 48, the eighth year of Samsuiluna (an 8U) had a year-formula not too dissimilar to that of year fourteen of Ammisaduqa, which suggests that Schnabel might have misidentified a Samsuiluna source as from Ammisaduqa. We grant that these two points are worthy of consideration. But both points were introduced quite late in the game. Weir accepted the 14U in 1972 (see page 37 of his book), and Huber accepted the 14U in 1974 (see page 10 of his "April 1974" version). It was only after they became aware of the attested 13A that they decided to attack the attested 14U. Until these matters are better resolved, we think that both 13A and 14U should be accepted as attested. If Weir and Huber are so uneasy about 14U, they should be asked why they never have any doubts about 5U. The answer seems to be that they must have a 5U and must not have a 14U (at least now that an attested 1 3A is known).
Let us explain why it is not enough to compare Weir's "Computed Babylonian dates" in his Table 3A to his own version of the "Recorded Babylonian dates" from that same Table 3A, and why it is not enough to compare Huber's "arcus visionis values" in his Table 4.2 to his own "data" from that same Table 4.2. The reason is that sometimes a computation is being compared to itself. That is, an actually recorded date has been changed or rewritten so that it will fit the computed date. In other cases, there may be two or more dates recorded, and Weir and Huber each (openly) pick the one that is closest to their retrocalculated date. In Year 11, for example, three sources support one date and only one source supports another date; but Weir and Huber both opt for the minority report, simply because it is one day closer to the retrocalculated date. It might be noted that it still misses, by some three days (the date of appearance even lies well within the computed invisibility, which is a situation that they admit that they find totally unacceptable), but they seem to think that this is better than missing by four days.
The point is, of course, that minority readings are chosen because of astronomical considerations. This is no way to do either textual reconstruction or statistics. That is why the astronomical retrocalculations should be compared to our version of the data, rather than to Weir's or Huber's "data"; our presentation above has no astronomical bias, and is a textual reconstruction based purely on the weight of the evidence.
Even the retrocalculated dates are doubtful, when they are expressed in the Babylonian calendar. For those dates have been altered by an arbitrary use or non-use of intercalary months, especially the non-use of 14U. (We could also raise objections about the arbitrary use of 19U and 20A in the computed dates, but perhaps we need not bother. Even with this tampering, the Year 19 to 21b sequence remains so bad that Huber discards it in its entirety; Weir keeps that sequence, but cannot do much with it, since only one date out of eight provides a good match.)
Thus, neither the recorded dates nor the computed dates are pristine: each set has been doctored to fit the other. Even so, the fit is poor. That in itself is surprising; with so much editing, one would expect a much better fit if the Venus tablets have anything at all to do with Ammisaduqa.
It is strange that Weir would have recourse to "a modification in the shape of the Venus orbit" (pages 43, 46) in order to explain some of the discrepancies that he encounters. For he remains committed to retrocalculation. If the orbit of Venus has undergone modification (Weir has in mind a change in eccentricity), how can we rely on retrocalculation of the present orbit in order to determine what the ancients would have seen at a time when Venus was on an orbit of a different shape? The underlying assumption of retrocalculation is that we understand the mechanisms by which the planets' orbits have evolved; hence, to propose an unexplained "modification in the shape of the Venus orbit" is to call into question the whole basis for retrocalculation. Weir's proposal is simply not consistent with his unquestioning acceptance of retrocalculation as a legitimate tool. (For all of our criticisms of Huber, it is to his credit that he understands that he cannot have it both ways here.)
What is not to Huber's credit is that for years he has been describing the Venus tablets as the worst set of data that he, as a professional statistician, has ever seen. What does this mean? What is so dismally bad about the data? There are indeed many minor discrepancies among the various surviving versions of the document. We have already illustrated such discrepancies in the case of Year 12. There also seem to be some scribal errors here and there; perhaps that is the very reason for those discrepancies in Year 12. And some of the surviving tablets are rather badly damaged. But all of these difficulties are commonplace in cuneiform studies, and in ancient philology generally. There is nothing of this sort about the Venus tablets that is at all unusual or out of the ordinary. And, indeed, these sorts of textual problems are not what Huber is talking about here. What he really means, but almost never says, is that the data are in conflict with retrocalculation. But that of course does not prove that the data are bad. It may be that the retrocalculations are misleading. Huber is assuming, with no justification whatsoever, that if the data are in conflict with retrocalculation then the data are wrong. That is his only reason for saying that this is the worst set of data that he has ever seen.
Retrocalculation is a very iffy business at best. Suppose that a spacecraft was launched into orbit this morning. Retrocalculation would show where that spacecraft was located on its orbit at this same time yesterday, when it was still on its launching pad!
We suggest that the Venus data may be quite good, aside from the minor textual problems that we have already discussed. Perhaps scholars have too long allowed retrocalculation of orbits to persuade them of how bad the data are. (After all, if there have been orbital changes within historical times - as Weir himself often suggests - then retrocalculation based on present orbits would be entirely illegitimate.) Perhaps Weir and Huber should try to see what the data may tell us about the orbits. That would indeed be a fresh approach.