By: Michael Press
In 1942, Warner Bros. Pictures released the film George Washington Slept Here. Based on a play by the famous team of Moss Hart and George S. Kaufman, its title alludes to the signs on buildings proudly proclaiming that a famous historical figure (Washington, Napoleon) stayed the night. The movie stars Jack Benny and Ann Sheridan as a married couple: Ann Sheridan buys an old colonial house – without her husband’s knowledge – because it is believed that George Washington once spent the night in the house. Jack Benny is angry on hearing of his wife’s purchase – and this anger is only magnified when evidence is discovered that it was not George Washington who stayed at the house but rather Benedict Arnold, the archetypal traitor of the Revolutionary War. Just when things seem darkest, however, the family dog comes to the rescue by digging up an old letter proving that George Washington had slept at the house after all.
What does the absurdity of this plot have to do with the search for the historical David? I would suggest that there are more similarities than we might like to admit. For one thing, the archaeological search for David, and for the United Monarchy under him and his son Solomon, seems to assume that tangible proof – like the letter the dog dug up – exists, just waiting to be found. It appears to underlie claims of the discovery of buildings like David’s palaces in Jerusalem and beyond. And it is certainly an assumption encouraged by the way that biblical archaeological scholarship is presented to non-scholarly audiences. Not surprisingly, then, this assumption is found widely among the public. I’ve encountered it among students, who believed that archaeology might uncover a “deed” of one group or another proving that they have an ancient claim to the land of Israel. In this essay, inspired by Zachary Thomas’s response to an earlier piece of mine, I would like to explore assumptions like this, highlighting problems concerning data and methods – not just in the search for the historical David but in biblical archaeology more generally.
In any historical analysis, we should begin by surveying our sources. The written evidence for David consists primarily of biblical texts: Samuel and Chronicles. However, using them for historical reconstruction is complicated by the difficulties of dating them, and of fully understanding their authorship, context, and purpose. It is generally agreed that Samuel was written to a significant extent at least in the Iron Age/First Temple period, while Chronicles was written during the Second Temple period. Beyond this, we have little but educated guesses. Not only may these texts have been written hundreds of years after David lived, but they also certainly underwent centuries of editing and transmission before the manuscripts we have today were produced.
The earliest textual evidence we have for David, then, is the Tel Dan Stele, from the 9th century BCE. Note that this stele does not talk about David himself, but simply appears to refer to the kingdom of Judah as Bet David, the “House of David.” It was standard in inscriptions of the time, especially among Neo-Assyrian records, to refer to kingdoms by the names of founding figures: thus Judah is seen in the 9th century as having been founded by David. This does not mean that David must be a historical figure, but it is certainly the most reasonable conclusion. (Nevertheless, we must remember that the stele itself does not prove David’s existence; this conclusion is an additional interpretive step.) It is also important to underline the fact that the stele dates to roughly a century in a half after David would have lived. This is roughly the same as the amount of time between the Revolutionary War and George Washington Slept Here. Here’s a thought experiment: What if the earliest evidence we had for Washington’s existence was a text dated to the time of World War II? How confident would we feel in Washington’s existence, let alone in our detailed knowledge of his career?
This leaves the other class of evidence, material remains. Archaeological evidence for David himself has always been understood to be sparse. The archaeology of the United Monarchy and the 10th century has therefore been focused on David’s son, Solomon, who is depicted in the biblical book of 1 Kings as a major builder. The foundations of “Solomonic” archaeology were developed between the 1920s and the 1950s. It was at this time that William Foxwell Albright and other archaeologists standardized the chronological terminology of the archaeology of Palestine. Albright’s excavations at Tell Beit Mirsim, Tell el-Ful and Beitin (Bethel), along with major expeditions to sites like Megiddo (by the University of Chicago) and Beth-Shean (by the University of Pennsylvania) during the British Mandate – and, later, the Hebrew University excavations at Tel Hazor in the 1950s – established a basic chronology for Iron Age Israel (the time of the United and Divided Monarchies) based on pottery, as well as the occupational history of important sites.
How did archaeologists identify what belonged to the 10th century BCE, the time of the United Monarchy? To answer this, it is important to understand how archaeologists date things in general. Dating artifacts and strata is a difficult process. Techniques such as radiocarbon or dendrochronology can be used, but they require specific conditions and (at least for radiocarbon) are not precise enough for historical reconstruction. In most cases, archaeologists must rely on ceramics – pottery – for dating. Because pottery styles rise and fall in popularity, we can determine whether one type of pottery is earlier than another, but this only provides us with a relative chronology. In order to provide a relative chronology with absolute dates, we need to be able to tie it in to known chronological sequences. For the Near East, these are usually provided by Egypt and Mesopotamia, for which we have long king lists with pretty good absolute dates.
In ancient Israel, there are good chronological anchors for the 12th century, the end of Egyptian control over the region – for which there are many objects found inscribed with pharaohs’ names; and there are good anchors in the late 8th century, when Assyrian campaigns left their mark in Assyrian annals, site destructions in Palestine, and occasional royal inscriptions found among them. However, we have a gap in between of four centuries – and the time of the United Monarchy, the 10th century, falls right in the middle. When sites like Megiddo and Hazor were being excavated, there was no clear way to date things precisely to the 10th century. Instead, excavators observed the earliest phases of major monumental architecture after the end of the Bronze Age (that is, after the Egyptian withdrawal in the 12th century), and concluded that this must be the remains of the kingdom of David and Solomon.
The danger of this method of dating – not by correlating relative ceramic chronologies with established king lists but by historical assumption – can be seen by comparison with another example: the archaeology of the Muslim conquest of Palestine. For many decades, scholars believed that the transition to the Islamic period in Palestine in the 7th century CE was marked by widespread destruction, and pointed to the archaeological record as providing proof. But that proof is very problematic. Jodi Magness has shown that, when archaeological contexts from this period were first being excavated in the Jerusalem area in the early to mid-20th century, the ceramic chronology was not precise – instead, excavators generally assumed that destructions or abandonments around this time must be attributed to the Muslim conquest. In fact, these destruction and abandonment layers were not necessarily synchronous at all, and the assumption about the Muslim conquest led them to be misdated by a century or more! However, because these excavations helped to define the pottery chronology, the errors were repeated in subsequent work; and, until the last 20 years or so, very few people thought to question the consensus, because it fit with what everyone “knew” about the Muslim conquest.
Here, we see conclusions flawed by what classical archaeologist Anthony Snodgrass has called a positivist fallacy. Briefly, this is the idea that what is observable (in the archaeological record) is also significant. This fallacy actually involves a set of assumptions that we find frequently in archaeology of the historical periods, in the attempt to combine text and artifact. In the case of destructions caused by the Muslim conquest, we see two related assumptions – one, that the event (widespread destruction with the Muslim conquest) was real; two, that it must be visible in the archaeological record. Thus, any destruction, or abandonment, from anywhere near the early 7th century was automatically linked to this historical event – even though destructions and abandonments might easily have other causes, and even though historical sources indicate that the Muslim takeover was generally peaceful. To look at it another way, it is assumed that excavation at a seventh-century CE site must uncover evidence of the Muslim conquest — just as in George Washington Slept Here, where digging at the house must uncover evidence about Washington himself. Or just as in Jerusalem, where a large building excavated in Jerusalem and (possibly) dating to the 10th century BCE must be mentioned in the Bible (witness the debate, referred to by Thomas, over whether structures in Jerusalem are David’s palace or the “fortress of Zion” of 2 Samuel 5:7.) Of course, we must remember that most evidence in the archaeological record has nothing to do with political history. It instead relates to how ordinary people worked, slept, and ate, lived and died.
Because this positivist fallacy has gone mostly unrecognized in biblical archaeology, the results obtained have not been critically reexamined. Consider the case of the six-chamber gates (the city gates of Hazor, Gezer, and Megiddo that Yigael Yadin identified as the sign of Solomon’s building activity in 1 Kings 9:15). Thomas strongly implies (as many would agree) that the evidence linking these gates to Solomon was persuasive until Israel Finkelstein proposed his low chronology. In fact, the connection was quite weak long before the Tel Aviv School came along. Gates of this type have been found far beyond Hazor, Megiddo, and Gezer – even at non-Israelite sites like Ashdod; and we are certain that some of these were constructed later than the 10th century. Thus it was clear long ago that these gates were not a hallmark of Israelite construction or of the 10th century, let alone the blueprint of Solomon’s architect. They were simply a popular gate type of the Iron Age II. In addition, as William Schniedewind has shown, the archaeologists discussing these gates have failed to approach the biblical verse on which the identification is based and its context critically. Instead, all of them have read the passage as if it is a straightforward, unedited eyewitness report.
Our sources for the United Monarchy, then, all have their difficulties. The evidence they provide is limited and fragmentary, or even unrelated to our problem. One obvious question raised by this debate, then, is: What is the threshold of evidence we need for accepting a reconstruction? I believe it needs to be much higher than scholars often place it. To demonstrate, let’s consider the case of Khirbet Umm Laqis. Umm Laqis (in modern Hebrew, Horbat Leqesh) is a site located about 15 miles east-northeast of Gaza, on the ancient road between Gaza and Hebron. In the 19th century, many explorers and scholars were interested in it and a nearby site, Khirbet Ajlan, because they identified them – due to the similarity of the names and their general location – with biblical Lachish and Eglon. This was what motivated the Palestine Exploration Fund to have Flinders Petrie dig at Umm Laqis in 1890, as the first stratigraphic excavation of a mounded site in Palestine. However, after three days of digging Petrie confirmed his initial impressions that the site was “late” (that is, going back only to the Roman period) and therefore not of interest. Petrie then moved to the nearest promising site that appeared to be of “biblical” (Bronze and Iron Age) date, Tell el-Hesi, three miles to the southeast. When excavations revealed extensive Bronze and Iron Age remains, Petrie was satisfied that Hesi was the site of ancient Lachish and this identification became widely accepted – with the explanation that, with Hesi’s abandonment, the name Lachish migrated to a new site in the Roman period.
This wide acceptance lasted until Albright identified Lachish instead with Tell ed-Duweir (about 10 miles to the east of Kh. Umm Laqis) in the 1920s, and British excavations at that site in the 1930s confirmed Albright’s proposal. Today, there are two opinions about Umm Laqis and Lachish: some suggest that the name had simply migrated a greater distance, from Tell ed-Duweir to Kh. Umm Laqis; others, that the British excavations finally disproved the link between Umm Laqis and Lachish. In fact, neither claim is correct: any link between the two names was conclusively severed by the great French scholar Charles Clermont-Ganneau in the 1890s. Clermont-Ganneau showed that real name of Kh. Umm Laqis was in fact Mulaqis – meaning the element Umm was not a separate word but an integral part of the name, and that the name was etymologically unrelated to Lachish. (Unfortunately, Clermont-Ganneau’s conclusions were largely ignored at the time.)
In this example, we are dealing with two names that are superficially very similar, with sites located very close to each other (about 10 miles), and in languages about which we know a great deal. And in fact it was that detailed knowledge – in particular, careful consideration of the linguistic evidence — that was necessary to provide an explanation of the names. So what of Charles Steitler’s suggestion that Tou (or Toi) of 2 Samuel and Taita from Luwian inscriptions – two names with much less of a superficial resemblance – are identical? His evidence consists of the variants Thaei (in a few Septuagint manuscripts) and Thainos (in Josephus) for Tou/Toi in 2 Samuel, which he then explains as having changed from a to o through the Canaanite shift (the change of Proto-Semitic long a to long o).
While a few Greek manuscripts have these (or other) variants with a, the overwhelming majority have o (mostly Thoou). Given the large number of variants, each appearing in a small number of manuscripts, it is best to understand them as spelling errors (just like Thomas argues for yod/waw). Meanwhile, it is hard to take the Canaanite shift explanation seriously: why would a small group of Greek manuscripts preserve a pre-shifted vowel from two millennia earlier (when most do not)? And, more importantly, how could Toi have undergone the Canaanite shift when that shift is attested already in the Late Bronze Age and routinely dated by scholars to the mid-2nd millennium: how did Taita’s name undergo a change that had occurred 500 years before he lived?
The evidence put forward in this case is very weak, the arguments unacceptable. But even if they were acceptable, they would not prove the connection between Tou and Taita, only that the connection is plausible. And even if a connection were proven, it would only be a piece of circumstantial evidence: it would say nothing directly about the kingdom of David. In short, we have a weak argument about an implausible explanation of circumstantial evidence.
What we see, in the case of Toi/Tou as in others, is that reconstructions of the kingdom of David and Solomon rest uncomfortably on little evidence. That evidence cannot hope to prove the kingdom of David and Solomon by itself. What we are doing, then, is not debating the evidence, as meager as it is, but the assumptions we use to interpret it. This is true not only of the United Monarchy but of biblical archaeology more generally. We can see this illustrated more clearly with a different example, that of the Conquest Hypothesis. The Conquest Hypothesis was developed by Albright to explain the emergence of the people Israel in Palestine; it is based essentially on the biblical account of the conquest of the land in Joshua. Archaeologists today often tell the story of how the Conquest Hypothesis used to be the dominant one until it was disproven by archaeological evidence in the 1980s and 1990s. Again, we have the narrative of objective researchers bowing to the weight of the evidence. But this is not really true. It was already quite certain by the middle of the century that the Conquest Hypothesis was fatally flawed. Excavations at three of the sites most central to the conquest story in Judges – Ai (et-Tell), Jericho, and Gibeon (el-Jib) – had each revealed that the cities weren’t even inhabited at the end of the Late Bronze Age (the time of the alleged conquest), let alone destroyed. As for those few sites (like Hazor) that did have destructions around that time, these destructions weren’t synchronous and none could be attributed to a specific cause or group. And yet biblical archaeologists continued to insist for decades that the Conquest Hypothesis was still valid. Albright, G. Ernest Wright, and others explained away each problematic case with what look to us like rationalizations, but to them looked like reasoned arguments.
What led to the prolonged life of the Conquest Hypothesis was not the evidence – which was exceptionally weak – but rather positivist assumptions about the nature of evidence. And one assumption in particular was especially influential: that the Bible is in essence a historical document. Albright and his students were not biblical literalists, but they believed in God’s revealing himself in history: therefore the desire to have the Bible be essentially historical was very strong; and their interpretations were seriously affected by this belief.
We see the same problem with the archaeology of the United Monarchy: the meager evidence only coheres into a story – and specifically the story of the United Monarchy – if we start with the belief that the biblical account is essentially historical.
Some might read these words and raise the objection that cases like the Israelite Conquest refer to the earlier historical books, while later historical episodes are more reliable as history (and that David might be grouped with those). There is no denying that there is a clearer historical background to much of the books of Kings in particular. But here too the relationship between text and archaeology is far from simple. Let’s consider the example of the Hezekiah-Sennacherib narrative (2 Kings 18-19/Isaiah 36-37, and 2 Chronicles 32). This is often seen as a parade example of the meshing of text and artifact, with the Assyrian campaign to Judah attested in the evidence of the biblical accounts; Assyrian annals; the reliefs of the Assyrian king Sennacherib’s palace showing the siege and conquest of Lachish; and the archaeology of sites such as Lachish itself, with its destruction and Assyrian siege ramp.
These correspondences often led scholars, especially in the early to mid-twentieth century, to claim that the account in Kings was fully vindicated. But details in that account are clearly wrong: the amount of tribute listed is not the same as that given by Sennacherib; Tirhakah was not yet king of Egypt at the time of the invasion; and the miraculous deliverance of Jerusalem through the killing of 185,000 Assyrian soldiers cannot possibly be accepted. In addition, there are various chronological problems with 2 Kings 18-20. And in fact when we look at the episode from a literary perspective and in its broader biblical context, we begin to see how artfully it is constructed, with a series of speeches arranged in a chiastic structure, speeches that reflect in every major way themes and vocabulary from the book of Isaiah and the previous chapter and a half of Kings. So, for instance, instead of wondering whether we have the exact words of the Rabshakeh’s speech, as many scholars have, we see that the Rabshakeh is in fact a construct of the biblical author – whatever the reality of whether such an official gave a speech at Jerusalem during Sennacherib’s campaign.
Biblical archaeology has tended to see the historical books of the Bible as being split between myth and history, with a dividing line separating the two chronologically; once we cross that threshold, we enter true “history”. The history of the field is the history of that line sliding back and forth: the Albright school wanted the line at the Patriarchs; Israel Finkelstein and Tel Aviv school between the 9th and 7th centuries; the biblical “maximalists” at the 10th century (or slightly earlier). What these scholars all seem to agree on, though, is that once we arrive on the near side of the line, we are safely within history. But if the Hezekiah-Sennacherib narrative, the supposed parade example of historical confirmation, is in fact literary artifice, then how are we to judge things like the reign of David, for which we have so much less – in fact, almost non-existent – evidence? Many biblical narratives from the historical books may well be related to a historical reality, but that relationship is complex and unpredictable. We cannot simply assume it is a binary, yes or no one. We cannot simply accept biblical narratives as unedited primary sources.
The continuing search for the historical David appears, like the conclusions of the Albright School, to be driven – problematically – by heavy investment in the historicity of the Bible. And here lies the paradox: heavy investment in the Bible is what motivates work in the field in the first place, and what drives public interest. Without it, our field simply would not exist. (Why else would anyone care about a tiny corner of southwestern Asia?). We must be careful in crafting responsible scholarly approaches that acknowledge this investment, with all its positives and negatives. In any attempt to historicize the United Monarchy, we must also historicize our own scholarship by acknowledging this context. After all, mythic history has a tremendous appeal, to scholars and non-scholars alike. It is why people still care, over 200 years later, where George Washington slept – and why people still care, 3000 years later, where David might have slept.
Many of these ideas were presented in lectures for the courses “Bible and Beyond: Reading Early Jewish Literature” and “The Archaeology of Israel and Modern Religious Identity” at Indiana University in Spring 2015. Many thanks to the Borns Jewish Studies Program, Eva Mroczek, and the students in these classes for help in developing them.
All content provided on this blog is for informational purposes only. The American Schools of Oriental Research (ASOR) makes no representations as to the accuracy or completeness of any information on this blog or found by following any link on this blog. ASOR will not be liable for any errors or omissions in this information. ASOR will not be liable for any losses, injuries, or damages from the display or use of this information. The opinions expressed by Bloggers and those providing comments are theirs alone, and do not reflect the opinions of ASOR or any employee thereof.
 This does not mean that Samuel is always more reliable than Chronicles. The author of Chronicles relied in part on Samuel as a source; but the earliest texts we have of the book of Samuel were written centuries after the composition of Chronicles. The author of Chronicles (and therefore versions of Chronicles in different languages) is essentially a witness to a form of Samuel just as later manuscript versions of Samuel are; all went through the same processes of transmission and translation.
Further, the Hebrew text of Samuel, the Masoretic Text (MT), is notoriously corrupt. In many cases, the readings provided by Chronicles are in fact superior to those of the Hebrew of Samuel – not just through copying errors but through various editorial changes in Samuel. As an example, consider Ishbosheth, the name of a son of King Saul in Samuel. In Chronicles, the same name appears as Eshbaal. It was widely believed that the form of the name in Chronicles was more original, and that editors of Samuel changed it because of the embarrassment of having the name of another god (Baal) in a divine name. And we know that the name Ishbaal goes back to the Iron Age because of the recent discovery of a jar with this name at Khirbet Qeiyafa: Yosef Garfinkel, Mitka R. Golub, Haggai Misgav and Saar Ganor, “The ʾIšbaʿal Inscription from Khirbet Qeiyafa,” Bulletin of the American Schools of Oriental Research 373 (2015), pp. 217-233.
Thomas is certainly wrong, then, to suggest that, in cases of differences between Samuel and Chronicles – such as Toi and Tou – we should automatically favor the form in Samuel. (This is besides the fact that the versions of Samuel themselves are split between Toi and Tou.)
 Jodi Magness, Jerusalem Ceramic Chronology: circa 200-800 CE (JSOT/ASOR Monograph Series No. 9; Sheffield: Sheffield Academic Press, 1993).
 Snodgrass has discussed this problem in several places. See, for example, An Archaeology of Greece: The Present State and Future Scope of a Discipline (Sather Classical Lectures 53; Berkeley: University of California Press, 1987), chapter 2.
 William M. Schniedewind, “Excavating the Text of I Kings 9: In Search of the Gates of Solomon,” in Historical Biblical Archaeology and the Future: The New Pragmatism, ed. Thomas E. Levy (London: Equinox, 2010), 241-249.
 Charles Clermont-Ganneau, Archaeological Researches in Palestine during the Years 1873-1874, Vol. 2 (London: Palestine Exploration Fund, 1896), p. 438. See already the comments of Edward Robinson, Biblical Researches in Palestine, Volume 2 (Boston: Crocker & Brewster, 1841), pp. 388-389.
 Forms of the name in different Greek manuscripts include Thouou, Thoou, Thooith, Thous, Thou, Thaei, Thainon, Thae, Thameis . . . as well as forms of Eliab! The great majority of manuscripts have Thoou or Thou. See Alan England Brooke, Norman McLean, and Henry St. John Thackeray, eds., The Old Testament in Greek. Vol. II. Pt. I. I and II Samuel (London: Cambridge University Press, 1927), p. 131.
 As in standard reference works: Israel Finkelstein, “The Great Transformation: The ‘Conquest’ of the Highlands Frontiers and the Rise of the Territorial States,” in The Archaeology of Society in the Holy Land, ed. Thomas E. Levy (New York: Facts on File, 1995), p. 363; Lawrence E. Stager, “Forging an Identity: The Emergence of Ancient Israel,” in The Oxford History of the Biblical World, ed. Michael D. Coogan (Oxford and New York: Oxford University Press, 1998), pp. 94-97 and Table 3.1. Both examples, from the mid- to late 1990s, emphasize that the archaeological turn against the Conquest Hypothesis was recent.
 See the convenient discussion by John J. Bimson, “The Origins of Israel in Canaan: An Examination of Recent Theories,” Themelios 15.1 (Oct 1989): 4-15.
 The theological views of the Albright School are discussed critically by Burke O. Long, Planting and Reaping Albright: Politics, Ideology, and Interpreting the Bible (University Park, PA: Pennsylvania State University Press, 1997).
 See especially Ehud Ben Zvi, “Who Wrote the Speech of Rabshakeh and When?” Journal of Biblical Literature 109 (1990): 79-92, and Paul S. Evans, The Invasion of Sennacherib in the Book of Kings: A Source-Critical and Rhetorical Study of 2 Kings 18-19 (VT Supplements 125; Leiden and Boston: Brill, 2009); also my forthcoming article, “‘Where Are the Gods of Hamath?’ (2 Kgs 18.34//Isa. 36.19): The Use of Foreign Deities in the Rabshakeh’s Speech,” Journal for the Study of the Old Testament.
 Thus, at least at times, Finkelstein appears to take this historicizing, positivist approach to the Bible just as much as scholars he criticizes. This point is also made by Schniedewind, “Excavating the Text of I Kings 9,” p. 244; for similar comments, see William G. Dever’s review of Finkelstein’s The Forgotten Kingdom, “Learning More About Israel? Or Israel?” Biblical Archaeology Review, July/August 2014.
 Marc Zvi Brettler makes a similar evaluation in The Creation of History in Ancient Israel (London and New York: Routledge, 1995), pp. 141-142.