Reading the Fractures of Genesis: Noah’s Flood

In 1996, Old Testament scholar David Carr published the results of his study into the history of Genesis’s formation. This book, Reading the Fractures of Genesis, made significant progress in unravelling the mystery of how the Pentateuch was composed. I’d like to write a few articles about Carr’s model and how it applies to various passages, beginning with the flood.

It has been recognized by scholars since the 18th century that Genesis and the Pentateuch were composed from multiple sources. The dominant model that most are familiar with today is Wellhausen’s Documentary Hypothesis (DH) from the late 19th century, which proposed four independent sources (J, E, D, and P) whose texts had been combined to form the Torah. Today, the original DH has been largely abandoned in favour of newer models, although many of its fundamental principles are still widely accepted.

Carr’s book gets back to basics and focuses solely on Genesis for developing his source theory. He draws attention to the presence of “fractures” or fault lines in the text of Genesis that cause problems for interpretation. According to another scholar — G.H. Hartman, whom Carr quotes — these fractures indicate “the tension that results between producing one authoritative account and respecting traditions characterized by a certain heterogeneity” (p. 11). In other words, Genesis attempts to create a unified story out of traditions that aren’t always in agreement, as demonstrated by clues throughout the text.

Genesis and Gilgamesh

Carr looks to the Epic of Gilgamesh as an example of the way ancient texts were worked and reworked over time, since it came from a related culture and has obvious commonalities with Genesis. And unlike a modern historian, who creates a brand new narrative after evaluating his sources, the authors of religious texts like Genesis were more likely to reuse traditional narratives that had been handed down to them, even though they could heavily modify or even “completely recontextualize” them (p. 21). The author was usually an anonymous writer, adding to or combining earlier texts and blurring the line between author and editor or redactor.

The Challenges of Reconstructing Transmission History

The Synoptic Problem and the creation of the Diatessaron — an early harmony of the four canonical Gospels — provide a useful case study for Pentateuchal studies. Carr reproduces a brief text from the Diatessaron to show how it combines phrases from multiple source Gospels even within the same sentence. It is clear from his example that if we possessed only the Diatessaron but lacked the individual Gospels, extracting the original source texts would be impossible. Nevertheless, the Diatessaron contains numerous clues that it was composed out of earlier documents. Carr shows that the same indicators are present in Genesis, and they can tell us something about the sources used in its composition.

The indicators Carr focuses particularly on are breaks in narrative continuity, contradictions, and doublets — similar passages occurring twice in the same text. Furthermore, shifts in terminology, emphasis, and ideology that correlate with these indicators can strengthen the case for different authors. An example from the Diatessaron would be the switch in terminology from “kingdom of God”—used in material taken from Luke—to “kingdom of Heaven”—used in material taken from Matthew—that often correlates to doublets, such as Jesus’ comments about riches. 

Carr’s example from the Diatessaron:

HOW DIFFICULT IT IS FOR THEM THAT HAVE RICHES TO ENTER INTO THE KINGDOM OF GOD! Verily I say unto you, it is difficult for a rich man to enter into the kingdom of heaven. And again I say unto you, IT IS EASIER FOR A CAMEL TO PRESS THROUGH THE EYE OF A NEEDLE, THAN FOR A RICH MAN TO ENTER INTO THE KINGDOM OF GOD. …VERILY I SAY UNTO YOU, THERE IS NO MAN THAT LEAVETH houses, OR BRETHREN, OR SISTERS, OR FATHER, OR MOTHER, or wife, OR CHILDREN, or kindred, OR LANDS, for the kingdom of God’s sake, OR FOR MY SAKE, AND FOR MY GOSPEL’S SAKE, and that doth not receive twice as many IN THIS TIME AND IN THE WORLD TO COME inherit ETERNAL LIFE.

Luke: boldface

Carr’s Composition Model for Genesis

In modern scholarship, there is near-universal recognition of a Priestly layer (the “P” source) that is distinct from other voices in the Pentateuch. Scholars tend to differ on whether P was an independent source document that was incorporated into the Pentateuch, or whether a Priestly redactor rewrote the Pentateuch, leaving his fingerprints throughout. Models that mix these approaches exist as well (pp. 43-47). In his book, Carr revisits the evidence in Genesis for a Priestly source and develops his own model: that there was a dependent Priestly source that was written to replace an earlier text (the non-Priestly or “non-P” text). Then, in a “remarkable intertextual move”, another redactor interwove P with the non-P text it was designed to replace. The first passage Carr uses to demonstrate this is the Genesis flood story, which is one of the most “fractured” parts of Genesis.

Illustration of Noah's Ark from the Nuremberg Chronicles, 1493
Illustration of Noah’s Ark from the Nuremberg Chronicles, 1493

Doublets in the Flood Story

In particular, the redundant and sometimes contradictory doublets in Genesis 6 through 9 imply the interweaving of two sources. Consider, for example, the two separate but similar reports of Noah and his family entering the ark with the animals:

Gen 7.7-9 (non-P): And Noah with his sons and his wife and his sons’ wives went into the ark to escape the waters of the flood. Of clean animals, and of animals that are not clean, and of birds, and of everything that creeps on the ground, two and two, male and female, went into the ark with Noah, as God had commanded Noah.

Gen 7.13-16a (P): On the very same day Noah with his sons, Shem and Ham and Japheth, and Noah’s wife and the three wives of his sons entered the ark, they and every wild animal of every kind, and all domestic animals of every kind, and every creeping thing that creeps on the earth, and every bird of every kind—every bird, every winged creature.  They went into the ark with Noah, two and two of all flesh in which there was the breath of life. And those that entered, male and female of all flesh, went in as God had commanded him.

The possibility of two sources grows much stronger when we compare the doublets with their associated vocabulary and themes. Carr separates the entire passage into two parallel strands—the non-P strand and the P strand. He then identifies 15 words and themes that are found almost solely in the non-P strand, and another 22 found mainly in the P strand. Many of these differences reflect the distinct viewpoint of the two sources; for example, non-P refers to God as “Yahweh”, while P refers to God as Elohim, because the revelation of the name “Yahweh” to Moses occurs much later in the P narrative (in the book of Exodus). Similarly, non-P distinguishes between clean and unclean animals and has Noah sacrifice one of every clean animal upon leaving the ark, while P has no sacrifice and makes no reference to clean or unclean animals, since the laws governing sacrifices and cleanliness originate with Moses much later in his narrative.

P also contains themes and vocabulary found only in other P passages, notably the creation story in Gen. 1. Similarly, non-P contains themes and vocabulary found only in non-P passages, such as the Eden story in Gen. 2–3.

In short, the flood story is rife with seams that indicate the mixing of similar but distinct narratives. I found it instructional to divide up the story myself, following Carr’s suggested source division, and to indicate some of the differences Carr identifies. You can download a PDF from the following link:

The Genesis Flood Story Separated into Sources

17th-century Flemish painting
17th-century Flemish painting (source)

Establishing Dependence between Non-P and P

Having convincingly summarized the evidence for the flood story as a combination of two source narratives, Carr assesses the reasons for thinking (1) that one of the sources was dependent on the other, and (2) that P was based on non-P instead of the other way around.

One clear reason for dependence is simply how similar the two narratives are. They are a close match not only episode to episode (as Carr puts it), but sub-episode to sub-episode and even comment to comment. This would be extremely unlikely if the two narratives were independent. Moreover, although two flood stories have close parallels in older Near Eastern flood myths, P and non-P are more similar to each other than to any other story.

Carr also finds the use of certain rare words, such as tebah (“ark”), in each strand a sign of dependence.

The Direction of Dependence

Carr finds several reasons to consider P to be dependent on non-P rather than the reverse. One of the clearest indicators is the places in which P is “more expansive” than the parallel texts in non-P. P adds significant detail to the building of the ark, the flood, and other details in the story, as well as producing a framework that connects to God’s promises and covenant throughout Genesis.

Another reason is that P imitates non-P in creating a detailed chronology for the flood that is unnecessary in P, but necessary for non-P. In particular, the duration of the rain is responsible for the severity of the flood in non-P, but rain isn’t even a factor in P’s flood, which is caused by the opening of portals in the firmament and below the earth.

Noah's Ark by Aurelio Luini, c. 1556
Noah’s Ark by Aurelio Luini, c. 1556

Excursus on the Chronology of the Flood Story

I feel that the chronology of the flood deserves more attention than Carr gives it. The best analysis available can be found in a 1980 paper by Niels Peter Lemche. It perfectly explains several problems with the chronology and sheds further light on the redaction process.

At first appearance, the story contains two sets of dates: the relative dates of the non-P source and the absolute dates of the P source. Reading the story as a whole yields a chronology that is impossible to make sense of.  Separating the story into its two source strands helps a great deal, but difficulties remain. Non-P seems to be missing part of its chronology — specifically, the section about the building of the ark is gone in favour of P’s account, and the reference to “another seven days” in 8.10 suggests a missing period of 7 days between the raven and the first dove. The P chronology intends to describe a year-long flood. Assuming a solar calendar with 30-day months, it begins in Noah’s 600th year, has 150 days of flooding and 150 days of receding floodwaters, with mountaintops becoming visible on 1.10.600 (1st day, 10th month, 600th year), precisely 300 days after the presumed start of the flood on 1.1.600, and the earth becoming dry on 1.1.601, precisely one year after the start of the flood. To summarize the P dates:

7.6: Noah was 600 years old when the flood of waters came on the earth.

7.24: And the waters swelled on the earth for 150 days.

8.2, 5: And God made a wind blow over the earth, and the waters subsided…. At the end of 150 days the waters had abated.

8.5: In the 10th month, on the first day of the month, the tops of the mountains appeared.

8.13:  In the 601st year, in the first month, on the first day of the month, the waters were dried up from the earth.

So far, so good. We have a coherent chronology, and every date makes sense. But then we get these problematic dates, which are also usually placed by scholars in the P strand:

7.11: In the 600th year of Noah’s life, in the 2nd month, on the 17th day of the month, all the fountains of the great deep burst forth, and the windows of the heavens were opened.

8.4: And in the 7th month, on the 17th day of the month, the ark came to rest on the mountains of Ararat.

8.14: In the 2nd month, on the 27th day of the month, the earth was dry.

So P gives two different dates for the start of the flood, two different dates for the mountaintops appearing, and two different dates for the drying of the earth. Lemche’s brilliant solution is that these latter dates were added by the redactor (call him “R”) who combined P and non-P to produce our text of Genesis. The intent, in adding them, was to produce a new chronology that utilized both P’s absolute dates and J’s relative dates. It went something like this:

  1. Original non-P began with the building of the ark over 40 days (now omitted) followed by 7 days of waiting inside the ark. 40 + 7 = 47 days. R added these 47 days to the start of P’s flood on 1.1.600 (7.6), resulting in a new start date: 17.2.600 (7.11).
  2. R combined P’s 150-day flood and 150-day abatement into a single 150-day period, which, added to 17.2.600, gives the date 17.7.600 for the appearance of the mountaintops.
  3. Since the flood lasted one year until the drying of the earth in P, R added one year to 17.2.600 and got 17.2.601 as the end of the flood. 11 more days were later added to account for the difference between the solar and lunar calendars, resulting in 27.2.601 as the text now reads.

For whatever reason, R chose to include most of the dates from non-P and P alongside his revised dates, resulting in the tangled timeline of the final text. Since these chronologies are difficult to describe in purely textual terms, I’ve created a diagram that some might find useful:


Interestingly, because this analysis of the flood chronology serves as evidence that non-P originally included an ark-building episode, we might consider that to be another doublet in the narrative, even though the final text ostensibly gives only P’s elaborate version.


The flood story provides an ideal text for identifying the compositional history of Genesis because of how obvious many of its editorial seams are. Carr’s model — that the Priestly text, written to replace an earlier version, was instead combined with that earlier version — explains very well the text in its present form.

Conservatively oriented readers may prefer to read the text synchronically — as though it were written all at once by a single author, as Jewish and Christian traditions have typically claimed. However, this way of reading the text simply is not capable of explaining the text’s many oddities and complexities. Models like Carr’s have greater explanatory power, allowing us to make greater sense of Genesis through analogy with other texts we know were based on multiple sources.

My earlier article on Exodus and the plagues of Egypt also discusses the composition of that passage in terms of Priestly and non-Priestly layers. Check it out if you haven’t already.


David M. Carr, Reading the Fractures of Genesis: Historical and Literary Approaches, 1996.

Niels Peter Lemche, “The Chronology in the Story of the Flood”, JSOT 1980/5.

14 thoughts on “Reading the Fractures of Genesis: Noah’s Flood

  1. I haven’t read Carr, so I’m interested in the source model he proposes. From what I gleaned here: J is essentially non-P, while both E and P are simply P, with later editions from R (and D omitted here because the source isn’t in the flood story). Is this a correct understanding of his proposal?

    Liked by 1 person

    • Hi, Mark. In the case of the flood story, Carr’s non-P corresponds to what other scholars have typically labelled as J. If I understand him correctly, Carr doesn’t think the J and E sources existed at all. Citing the current trend in scholarship, he regards non-P as a series of redactions that linked discrete traditions, not as a combination of discrete sources like J and E.

      In other words, the “original” Genesis was the non-P version, which was a patchwork of different traditions put together over time. Then P rewrote it from the ground up to serve a larger framework. And lastly, someone combined P and non-P, resulting in our current Genesis with all its doublets and editorial fractures.

      Carr doesn’t posit a D source, but he does discuss a “Deuteronomistic” redaction layer that adds elements here and there (but none in the flood story).


  2. Fascinating post, Paul.

    I’ve read an interesting apologetic argument that purports to explain the “fractures” in the flood narrative. They cite Gordon (I think) Wenham, and say that the structure of the narrative is chiastic. Because of this, the apologists say, we know that it was written by a single hand and that this fact makes sense of all the apparently conflicting details. The conflicts were on purpose, to demonstrate some deeper meaning… or something like that.

    Do most scholars agree that there is a chiastic structure here? How does this play out in mainstream critical analysis? What do you think of this argument?

    Always look forward to your thoughtful posts!

    — Michael


    • Thanks for the comment, Michael. There is a two-part paper that directly addresses your question:
      J. A. Emerton, “An Examination of Some Attempts to Defend the Unity of the Flood Narrative in Genesis: Part I”, VT, 37/4 (Oct 1987); and “Part II”, VT, 38/1 (Jan 1988).

      Do most scholars agree that there is a chiastic structure here?

      In short, no. The chiasmus argument comes from a single author (F. I. Andersen, The Sentence in Biblical Hebrew, 1974) in a book that is not actually about source analysis. Emerton looks at the chiasmus claim in “Part II” of his paper. Basically, Andersen’s argument is simply that if the two parts of a chiasmus combine P and J material, it’s more likely the work of a single author. Emerton looks at Andersen’s claim of a chiastic structure behind the flood story and finds that it is very selective in ignoring material that doesn’t fit, and it still does nothing to explain why, for example, the people and animals board the ark twice, or how the ark comes to rest in the mountains long before the mountaintops appear. Emerton concludes that “[Andersen] has not succeeded in weakening the theory that the flood story has been composed of two sources.”


    • I don’t think there’s a consensus on the number of authors, because scholarship seems to be moving away from the old idea of discrete sources (i.e. JDEP) to an idea of fragments combined in various stages by redactors, essentially erasing the difference between “author” and “redactor”. So for a model like Carr’s, you have disparate patriarchal stories combined by redactors at one or more stages into a “non-P” document, which was combined with the P document, with further revisions at different stages.

      Reinhard Kratz in his essay “The Pentateuch in Current Research: Consensus and Debate” (2011) suggests that the only consensus is for three distinct literary strata in the Pentateuch: Deuteronomy, the Priestly Writing (P) plus the Holiness Code (H), and the Covenant Code.

      The amount of material on these topics is mind-boggling, and I can’t offer much of an opinion yet.


      • At one time some scholars believed that these stories were written during the reign of King Josiah as national folklore to unify the northern and southern kingdoms (what remained of the northern kingdom). Is that view still in vogue? And, what percentage of scholars would you say believe that the Exodus story is folklore and not historical?


        • Based on what I’ve read, I would say few if any scholars now think any part of the Pentateuch was written before the exile. Some might see Deuteronomy as a pre-exilic work, but even that seems unlikely, since it anticipates the exile. Much of the non-Priestly material is based on earlier folklore (e.g. the Samarian Jacob tradition and the Hebronite Abram tradition), but the few references to such characters elsewhere in the Old Testament show no awareness of the Pentateuchal version.

          As for the Exodus story, I would say the number of scholars who think it is historical (outside of conservative religious institutions) is close to zero. There are a few hold-outs here and there.


  3. For whatever reason, R chose to include most of the dates from non-P and P alongside his revised dates, resulting in the tangled timeline of the final text.

    There could have been more than one redactor. The first combined P and non-P establishing the harmonised chronology. A subsequent scribe/copyist with access to P then added the original dates back in, not understanding the nature of the work. Sorta like later scribes of the NT tended to harmonise Mark to Matthew and Luke.


  4. are we to understand the one version of the flood account assumes that the jewish holy/unholy clean/unclean rules were known even to noah? if this is true, then rule against consumption of fine would go back to time of noah according to one version of the story?


    • That might be overthinking it. If sacrifices were simply standard practice among ANE cultures, and natural dietary taboos were already widespread, the author might not have even considered how Noah “knew” which animals were clean and unclean. After all, Noah’s sacrifice closely parallels the sacrifice that Utnapishtim offers at the end of the Gilgamesh flood story, and J (non-P) is closely following the Gilgamesh narrative plot point by plot point.

      Furthermore, some basic understanding of sacrifice rituals is already assumed in the Cain and Abel story (though seriously lacking in crucial details).


  5. Hi, I really appreciate your research into the flood story. This is truly fascinating and enlightening. I’d like to bring to your attention an article by Prof. Meir Bar Ilan who, in my opinion, does the best job in accounting for the flood chronology (with minimal speculative reconstructions and omissions). The article is entitled “The Calendar in the Flood Narrative” and can be accessed here: and here Thanks again for your wonderful blog.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s