POSTSCRIPT TO "TEMPORAL FUGUES" (JAMS XIII)

Since this article was published in 1996, there have been some new publications relevant to Rohl's thesis and the themes explored in my review.

LEO DEPUYDT has published two articles relevant to the theory of Sothic cycles underlying conventional chronology.

1. In the first, he surveyed the evidence available from texts dual-dated in two different calendars over the period from 473 BC to 237 AD - nearly half a Sothic cycle. He conclusively demonstrated that the Egyptian civil calendar was unreformed over the entire period, despite at least one serious attempt to do so. This greatly strengthens the case for believing that the civil calendar was never reformed in earlier times, when there were no competing calendars of greater accuracy available, and so supports the theory of Sothic dating.

2. In the second, he reviewed the Ebers papyrus, which contains the possible Sothic synchronism for year 9 of Amenhotep I. This interpretation of the papyrus has been much questioned recently. Depuydt showed that it is most reasonably understood as a simple ready reckoner for correlating lunar months to the civil calendar at this time. Such an interpretation is consistent with the theory that the papyrus does in fact record a Sothic sighting.

KENNETH KITCHEN has published a revised edition of TIP, including a new preface surveying work on the Third Intermediate Period since 1986. Among other things, he draws attention to the effects of Jansen-Winkeln's identification of the Tanite tomb Hedjkheperre Takeloth to Takeloth I rather than Takeloth II. This allows us to reassign several inscriptions from Takeloth II to Takeloth I. I suggested in my review that certain difficulties in Rohl's chronology could be resolved by exploiting the invisibility of Takeloth I. Since Takeloth I is now becoming visible, this possibility no longer exists.

Most importantly for our purposes, this preface includes a brief and acerbic review of the theories discussed in my review, which succinctly makes many of the same points. Kitchen does accept the existence of a second king Hedjkheperre Shoshenq, but concedes nothing else. In particular, he draws attention to a graffito published by Spiegelberg in 1921 recording the descent of the flood in year 1, 3 Akhet, day 3 of Merenptah, and points out that this date cannot possibly lie in the flood months under Rohl's chronology.

Rohl has had this graffito reexamined in situ, and corrects the reading of the date to year 2, 2 Akhet, day 3. If the flood recession occurred on a fixed date every year, this difference of a month would have the effect of adjusting the correlation of the civil calendar to the Julian calendar by 120 years.

Additionally, Rohl has raised questions as to the actual event recorded by the graffito. He notes that its most recent analysis by Janssen interprets the verb hAi as used in this and similar graffiti to mean "to return" in the sense of returning to inundation, and argues that this is directly contrary to the established meaning of the verb as "to go down". The difference is the difference between the rising of the flood and its descent from the point of maximum, two points which are about a month apart - leading to an additional 120 years in the calendrical correlation. Thus, Rohl argues that the graffito shows in fact that the conventional chronology is 240 years in error.

Rohl's forceful assertions to the contrary notwithstanding, Kitchen and Janssen do not in fact share the same interpretation of the verb hAi - Kitchen uses the same interpretation as Rohl does. However, one may note that Rohl has given no consideration to Janssen's arguments and precedents for using hAi in the more general sense of "return"; nor has Rohl considered how the workmen high in the hills behind Deir el-Bahri are supposed to have known on a particular day that that day was the height of the flood. This is an event which can only be established in retrospect.

The major point that Rohl's discussion overlooks is that the effect of the changes he is proposing is to force the nominal start of the flood date backwards in time against the civil calendar. Compare the three interpretations of the graffito:

· Janssen: Start of inundation on 3 Akhet 3

· Kitchen: Start of descent on 3 Akhet 3, hence start of inundation at the start of 2 Akhet.

· Rohl: Start of descent on 2 Akhet 3 hence start of inundation at the start of 1 Akhet.

In principle, each of these steps forces the civil calendar closer into alignment with the solar calendar, where 1 Akhet 1 is the date of the rising of Sirius, marking the start of inundation. In effect, they are pushing the nominal date for Merenptah backwards in time, closer to the start of the Sothic cycle c1320 BC. In other words, if Rohl is completely correct, he has indeed moved the nominal date of Merenptah by up to 240 years, but in the wrong direction for his own theories!

To put it another way: suppose Rohl's chronological theories and his interpretations of the flood graffito were both correct. His chronology requires the date of Merenptah to be about 350 years later than is conventionally supposed. His interpretation of the grafitto requires the civil calendar to be well aligned with the solar calendar in the reign of Merenptah. But in the reign of Osorkon III a flood event is recorded on 3 Peret 22, some six months out of phase with the solar calendar. Now, if Merenptah's civil calendar had been left unreformed (and assuming the Osorkon III event to have occurred close to the mean date), the flood event in his reign should have happened, in Rohl's chronology, some 3 months earlier, i.e. late in 4 Akhet. James and Rohl have each noted that the theory of Sothic cycles would be invalidated if the civil calendar had ever been reformed, since any such reform would most likely not be known to us. Now, the only way to reconcile the Merenptah graffito and the Osorkon inscription within Rohl's chronology is to suppose that such a reform did in fact occur some time after the reign of Merenptah. But the reform which Rohl requires in order to fit the flood graffito into his theory is not only very substantial (a cumulative effect of about 3 months) but also one which pushes the civil calendar even further out of alignment with the solar year than it already is. Such a reform is not credible.

Since the dates of the flood can easily vary by several weeks from year to year, an additional question arises, which Rohl has not discussed at all: the precision with which flood data can be used to estimate absolute dates. Janssen's study was based on data collected in the 19th century by Willcox and Craig, but extensive records also exist from almost the entire Islamic period from the Cairo Nilometer. These records include 329 actual or reasonably inferred records of the date of "plenitude", i.e. the date that the flood reached a height sufficient to inundate the fields and guarantee a year's harvest. We may note in passing that this concept of a date of plenitude is exactly the concept which underlies Janssen's interpretation of the Merenptah graffito. The Gregorian equivalent of the date of plenitude fell between July 17 and October 24 over the period 1074 to 1888, with a mean date of August 17, corresponding to a date of August 13 at Thebes, and a standard deviation of 12 days. 97% of the dates lie within the period July 29 to September 11, a period of 45 days. Thus, a reasonable estimate of the precision with which a flood date recorded in the civil calendar can be used to estimate a solar year is 180 years.

In short, the flood graffiti are not sensitive enough to discriminate between different models of the conventional chronology. However, they are sensitive enough to discriminate between Rohl's chronology and the conventional chronology, which are around 350 years apart at the time of Merenptah. And they favour the conventional chronology.

Before leaving this topic, it is necessary to address one passing comment made in Rohl's response. This comment is now standard fare in "New Chronology" circles on the Internet, but it is a canard, and so requires rebuttal. Rohl states (p46):

It is interesting to note that Janssen dismisses another of Spiegelberg's original 1889 texts as a Nile flood record because:

His (Spiegelberg's) suggestion that graffito 883 is also a 'Nilinschrift' (Nile text), as he calls them, is probably not correct since, as we shall see below (note. 63), its date does not suit that interpretation.

Note 63 states:

So it appears that graffito 883, if W. Spiegelberg read the date correctly as II peret (of year 4), cannot by any means refer to the beginning of the inundation …

Circularity abounds here. Texts are rejected by Janssen because they do not conform to chronological expectations, then the remaining filtered texts are subsequently employed by Kitchen to argue for the accuracy of the conventional chronology.

In other words, Rohl is accusing Janssen of rejecting a valid "Nilinschrift" graffito because its date does not match his (Janssen's) theory. This is completely false. Examination of Spiegelberg's book shows that the text of Theban graffito 883 is entirely lost except for the date. It is evident that Janssen was legitimately responding to Spiegelberg's speculation that the missing text was a "Nilinschrift". It is also evident that Rohl, who has earlier attacked Janssen and Kitchen vigorously for not having personally examined graffito 862 in situ, has not even bothered to look at what Spiegelberg actually had to say about graffito 883.

PHILIPPE BRISSAUD, the chief excavator at Tanis, has published a very detailed rebuttal of Rohl's analysis of the Tanite royal tombs, which makes it clear that the archaeology of the site is far more complex than Rohl has presented it to be. This paper includes a study of various issues omitted in my review, such as Rohl's claims that the northern side of Psusennes I's tomb was squeezed by the presence of a pylon built by Osorkon II. He also refutes Rohl's implicit assumption that a cramped tomb structure necessarily implies the need to avoid pre-existing buildings, by pointing to a number of other tombs in the complex which are equally cramped but were not so constrained.

On the central feature of Rohl's argument - the abutting walls between Psusennes I's tomb and Osorkon II's - he notes, as I did in my review, that the evidence presented by Rohl is not of itself sufficient to allow one to decide which tomb was built first. But he also presents detailed data, which I had been unaware of, and which Rohl had ignored, in support of the Montet/Lézine hypothesis that queen Kapes took over and substantially rebuilt a large portion of a pre-existing tomb for her son Osorkon II. Finally, he notes that the remains of tomb VII of the Tanis complex, first published in 1987 but ignored by Rohl, partly underlie tomb I, the tomb of Osorkon II. Tomb VII was therefore built and destroyed before Osorkon II's tomb was built. The remains of tomb VII incorporate a limestone lintel taken from an even earlier building which clearly names Akheperre Psusennes I. This single fact is sufficient to demolish Rohl's entire reconstruction.

In response, Rohl has disputed many points of Brissaud's rebuttal of his arguments. We will focus here only on the critical Psusennes lintel. The essence of Rohl's response is to note a comment by the expedition's epigrapher FRÉDÉRIQUE VON KÄNEL that the lintel was only partly cleared in 1984 and 1985 because the lower part of the block remained buried under the casement wall of tomb II. Since tomb II is the tomb adjacent to that of Osorkon II, and the date of construction of this casement wall is not precisely known, Rohl triumphantly concludes that the lintel in question does not disprove his chronology, and accuses Brissaud of misrepresenting the data. He concludes with an alternate reconstruction, based on the presence of ushabtis of Osorkon II in tomb VII, which makes this tomb an antechamber for tomb I.

In fact Brissaud and von Känel are both correct. Figure 8 in Brissaud's original publication shows the interface between tombs I and II, and also clearly shows the remains of tomb VII at the point in question as partly underlying both. The remains of tomb VII are actually separated vertically from the wall of tomb II by a layer of fill, but they directly underlie tomb I. This figure is simply ignored in Rohl's response.

After a hiatus of several years, in which ISIS even considered dissolution, publication of JACF has resumed, with JACF 8 in 1999. This issue contains the rebuttals by Rohl of the Kitchen and Brissaud criticisms noted above. It also contains several other articles supporting and opposing the New Chronology. Of particular relevance here, CARL JANSEN-WINKELN has published a brief critique of Rohl's use of long genealogies. He gives reasons for supposing that two of the key genealogies used by Rohl - the genealogy of Khnemibre in Wadi Hammamat and the genealogy of the High Priests of Memphis - cannot be regarded as reliable, noting in particular, in the case of the former, the highly duplicative nature of the genealogy, and its physical isolation, as I noted in my review; and, in the case of the latter, the short gap between Ramses II and Amenemnisu which I also discussed. He further notes conflicts with contemporary evidence for the reigns of Amenhotep III, Seti I and Ramses II. He draws attention to the genealogy of Basa, whose importance I noted above, and on much the same grounds. Finally, he notes that the Neseramun genealogy (which I did not discuss) is critical to establishing 22nd dynasty chronology and summarises how.

A brief response by BOB PORTER argues that the standard reconstruction of the Neseramun genealogy also contains problems, without suggesting how to resolve them in the New Chronology. He asserts, without demonstrating the fact, that other key genealogies used by Kitchen to support the standard chronology (such as the Nakhtefmut genealogy) are in fact consistent with both systems. The issues raised by the Basa genealogy are ignored. He accepts Jansen-Winkeln's point, that generations earlier than about 3-4 generations before the author of a long genealogies are likely to be suspect, unless supported by contemporary evidence, and turns it around to cast doubt on the Pasenhor genealogy, ignoring not only Rohl's own analysis of this genealogy (discussed in my review) but also the fact that the early generations of Pasenhor give the genealogy of kings of Egypt of the dynasty still ruling in Pasenhor's day. This genealogy must have been very well-known: Porter's critique is analogous to dismissing a modern descent from Edward III because it includes generations from that king back to Henry II!

The Neseramun genealogy also figures briefly in a paper by PETER JAMES, NIKOS KOKKINOS and I. J. THORPE summarising progress in the argument of Centuries of Darkness as of 1995. This gives a slightly more detailed discussion on the chronology of the 21st dynasty than is presented in their original book, in which they argue that the genealogical and prosopographical evidence allows it to have as little as two generations of independent existence. In particular, a chart is presented showing the Neseramun genealogy in a form which allows the authors to argue that Siamun of the 21st dynasty must be a contemporary of the 22nd dynasty king Osorkon I. However, the supporting analysis developed by James and Robert Morkot is as yet unpublished, and in particular the reasoning developed by Kitchen has not yet been publicly addressed. In Kitchen's view, the bulk of the Neseramun genealogy must be raised by a generation from the position shown by James et al., which would allow the conventional chronology to proceed.

None of these articles addressed the primary genealogical issue raised in my review: that the New Chronology, if correct, forces parallelism into all the major benefices in the land which we can trace in detail.

One of the key problems for the New Chronology, briefly alluded to early in my review, and not addressed in any publication at that time, is the extreme compression required for the late 19th and 20th dynasties. Rohl has suggested that Ramses IX and Ramses XI may have reigned in parallel, and AD THIJS is now publishing a series of articles amplifying and supporting this proposal. The case is very interesting and suggestive, but at best opens up a gap of only two decades in the current chronology. A far more serious revision is required before this problem can be regarded as solved for the New Chronologists.

Rohl's proposals on the matter are now available on his website, www.nunki.net. His theory requires a highly complex succession with as many as three kings reigning in parallel, apparently either as coregents or as kings of a divided Egypt. I attempt to summarise these proposals in the following table.

Ramses I 961-959

Seti I 959-940

Ramses II 940-873 Merenptah 885-865

Amenmesse 873-868

Setnakht 868-866 Seti II 868-862

Ramses III 866-835 Siptah 862-856

Tausret 856-855 Ramses IV 856-850

Ramses V 850-847

Ramses XI 847-817 Ramses VI 847-841

Ramses VII 841-836

Ramses VIII 836-835

Ramses IX 835-816 (wHm mswt 829-823) Ramses X 821-813

It would be pointless and unfair to try to critique this reconstruction at this time, since to my knowledge no analysis or argument has as yet been offered to support it. I will merely note that there is source data which goes directly against it, e.g. the conventional succession from Ramses III to Ramses IV is explicitly and unambiguously attested in the Deir el-Medina work journal pTurin 1949+1946.

A draft paper by DAVID LAPPIN, currently available on Rohl's web-site and planned for printed publication in JACF 9, addresses an important topic that I did not mention in my original review. Conventional dating of the Middle Kingdom is based on a predicted Sothic rising given in pBerlin 10012, dated to 4 Peret 16 in year 7 of an unnamed king, universally accepted to be Senusert III. In addition, pBerlin 10056 apparently gives a series of 12 consecutive lunar observations in years 30 and 31 of another unnamed king, generally agreed to be Amenemhat III, and a number of other lunar observations are known from these papyri, found at El-Lahun in the Fayyum. pBerlin 10056 certainly contains at least one error since one lunar month is recorded as being 31 days long. Past attempts to correlate pBerlin 10012 and the lunar data in pBerlin 10056 and elsewhere to obtain an absolute date for these kings, based on Sothic dates derived from pBerlin 10012, have led to results requiring a significant number of errors in pBerlin 10056.

Lappin has sought to address the problem in reverse, by conducting an exhaustive search over the lunar months in the years 1899-1549, using all possible interpretations of the ambiguities in the data, to find a best match for the observations of pBerlin 10056. His end result is to find a best match at 1649, which is very close to the date of 1653 predicted by Rohl's chronology. Lappin concludes that he has independently verified Rohl's proposed chronology, and demonstrated that significant calendrical reform must subsequently have taken place.

This experiment is potentially a significant contribution to the debate. However, in its current state Lappin's work cannot be said to have conclusively demonstrated the result he claims. For a start, his paper does not currently address the methodological issue of the baseline probability of finding a match by chance for any random pattern of data in an exhaustive search over a sufficiently large chronological range. Since we know pBerlin 10056 contains at least one error, it may well contain more, and not all observational error patterns can be detected just by examination of the data, since in most cases there will be a self-correcting dual error the next month. Fortunately, the existence of a detectable error - the 31 day month - assures us that we are looking at at least one run of at least three 30 day months, which does significantly reduce the search space.

Secondly, when all the available lunar observations are taken into account, it turns out on Lappin's own account that 1773 is as good a match as 1649. Lappin justifies his choice of 1649 over 1773 by asserting that pBerlin 10056 is the best data we have - but there is no obvious reason to believe this, it just happens to be the longest continuous sequence of observations known.

A date of 1773 is only 16 years below the current lowest Sothic-based chronology. This chronology, in turn, is based, not only on the assumption that Sothic sightings were made at Elephantine, but also on a minimal reign-length Senusert III (19 years). But, there is now good evidence that Senusert III reigned for at least 39 years (further suggesting that the reign-length of 19 years given to Senusert II in the Turin canon, and assumed in recent years to be misplaced from Senusert III, can now be restored to him). Further, there is also considerable doubt as to whether any coregencies occurred in the 12th dynasty. If both these adjustments are accepted, the combined effect is to add 20 years to the distance between the observations of pBerlin 10056, in year 30 of Amenemhat III, and year 7 of Senusert III, the likely date of pBerlin 10012, making this distance not 42 years but 62. If pBerlin 10056 were correctly dated to 1773, this would bring the date of pBerlin 10012 back to 1835, which is within the range of dates predicted by assuming it contains Sothic data. Noting that the assignment of pBerlin 10012 to Senusert III is an indirect inference ultimately depending on a paleographical analysis, one could even argue (despite the unanimous opinion of scholars to the contrary) that this papyrus should in fact be assigned to Senusert II, which would raise the date to 1854 - comfortably within the Sothic range.

I am not arguing here that 1773 is the correct date for pBerlin 10056. The arguments presented by Wegner for a long coregency between Senusert III and Amenemhat III are very plausible, the southern location that would be required for a Sothic sighting in year 7 of Senusert III unlikely, and the possibility of assigning pBerlin 10012 to Senusert II even more remote. Further, such a date creates other problems - either the Second Intermediate Period must be shortened beyond anything that could be accepted as credible, or pEbers must be discarded as a Sothic sighting, lowering the date of the early New Kingdom by up to 50 years. Such issues would need to be resolved in order to accept this date. My point here is only that Lappin has not succeeded in disproving a Sothic-based chronology for the Middle Kingdom. But his choice of experiment is very valuable, and his results deserve to be subject to independent analysis.

STURT MANNING, a central figure in the debate over the date of the Thera eruption, has now published a major contribution on the implications of Thera, and scientific dating methods in general, for bronze age chronology. His central thesis is that Thera erupted in 1628, and that this can be accommodated with the ceramic evidence from Tell el-Daba, argued by Bietak and others to favour a date late in the next century, by a reanalysis of the ceramic chronology. As such, the bulk of the book is not directly relevant to the issues reviewed here. However, Manning contributes an excellent and extended appendix critically reviewing the bases of Egyptian chronology and analysing the points of uncertainty within it. This summary should be read by anyone seriously interested in understanding the conventional framework.

Manning also includes a brief appendix entitled "Why the standard chronologies are approximately correct, and why radical re-datings are therefore incorrect." This is aimed more at Peter James et al. than at Rohl, who Manning dismisses as unscientific on the grounds that he seeks to "create reality for the Old Testament", an agenda sometimes described as "secular fundamentalism." However, the argument applies equally to both schools. The fundamental datum used is the date of a sample of cargo or firewood taken from the Uluburun shipwreck, which also contained a somewhat worn gold scarab bearing the name of Nefertiti. This wood dates to c1300 BC, a date entirely in line with conventional chronology. He further notes that this date fits in with a network of dendrochronological and 14C dates being established across Europe and the Middle East. Against the objection frequently raised by New Chronologists, that any sample awkward for their theories is a piece of "old wood", which he clearly regards as generally being sophistry, Manning notes that this framework is also supported by short-lived samples such as seeds.

Finally, one should note that the tone of discussion between Rohl and his opponents, particularly Kitchen and Brissaud, has become highly acrimonious. The diatribe is often quite unpleasant to read, on both sides. This is very unfortunate. While I am quite certain that Rohl's views are wrong, the discussion is nevertheless highly worthwhile, both for the student of Egyptian chronology, in that exposes the issues to a degree not otherwise likely to happen, and for the subject itself, because the debate can only sharpen our understanding of the bases of chronology. Because this debate is rapidly becoming a dialogue of the deaf, it is becoming increasingly unlikely that these productive features will continue to be realised.