Testing the Table of Nations

Summary

This article tests Genesis 10 against three bodies of evidence no ancient reader had:  the language families, ancient genomes, and a map.  It is a child of The Table of Nations.  Its siblings argue the identifications it tests, when the list was written and how it has been read.

Language:  the three lines are not language families, and every side now says so.  Only Japheth matches a family:  five of its seven peoples with a known language spoke Indo-European.  Ham's line holds twelve Semitic-speaking peoples against five under Shem.  Shem's holds Elam, whose language is related to nothing, and Lud, which if it is Lydia spoke an Anatolian Indo-European language.  The oldest Semitic texts, Old Akkadian and Eblaite from about 2400 BC,¹ come from Babylonia, the Shinar the Table gives to Nimrod and Ham, and north Syria.  The chapter lists tongue as one criterion of four and never claims one line, one tongue, so this is no evidence against it.  It does rule out any use of the Table as a map of where the language families began.  It also defeats three young-earth claims:  Henry Morris's "Hamitic" Babylonian language, the idea that Shem's line kept the first speech, and Arthur Custance's answer that the language evidence comes too late to decide anything.  On the Masoretic dates the oldest Semitic and Elamite texts are older than the dispersion itself.

Ancient DNA:  the ethnography holds where it can be sampled.  Canaan's peoples were one genetic population.²  The Philistines arrived at the start of the Iron Age with about 43 per cent of their ancestry from Bronze Age Crete,³ as Amos 9:7 brings them from Caphtor.

But the line assignments carry no genetic signal.  Canaanites under Ham and Israelites under Shem share one ancestry profile.  Minoan Crete under Ham and Mycenaean Greece under Japheth were "genetically similar".⁴  The Punic Carthaginians spoke a Canaanite language with almost no Levantine ancestry.⁵  These results fit a political and geographic reading of the lines, and Robert Carter's overlapping lines absorb them at the cost of making the assignments untestable.

Three results bear on time, because they count generations.  Canaan's admixture fits a post-Babel origin on any chronology.  A Bronze Age Armenian admixture falls about 2580 BC on the paper's own 28-year generation;⁶ with 25-year generations only the young edge of its range reaches the Masoretic dispersion, and it fits the Septuagint if the dispersion is put early in that text's range.  And the first farmers of the Levant and the Zagros were as different from each other as Europeans are from Chinese today.⁷  On Ussher's dates that difference had to arise and then be mixed away within about twenty generations, and drift can build it in ten generations only in groups of about 30 to 60 people.  No young-earth source proposes such groups.

Geography:  the map is drawn from the Levant.  Japheth is "north" only as seen from Syria and Palestine.⁸  Ham and Shem are not regions, and they overlap in Mesopotamia and South Arabia.  The Table names Crete, Ionia and Kush but not Dilmun, Magan or Meluhha, whose ships Sargon of Akkad moored at his own quays.⁹  That fits a compiler in Canaan or Egypt, Mosaic or later, and not a record drawn up in Shinar.

Verdict.  The Table's ethnography holds where it can be checked, and its map is Levantine.  Its three lines are neither language families nor genetic clusters, which costs nothing on the political and geographic reading of the chapter.  Read as biological descent from three brothers about 4,300 years ago, it meets two genetic results that strain the Masoretic dates, and the Septuagint answers the Armenian one if its dispersion is put early, about 2767 BC.

Part 1: Testing it against language

Genesis 10 closes each of its three branches with the same four criteria, "every one after his tongue, after their families, in their nations" (10:5, KJV), with "in their lands" added at 10:20 and 10:31.  Language is one of the four.  So it is fair to ask whether Japheth, Ham and Shem match language families.  They do not, and every side now says so, the young-earth writers who address it included.  Japheth is the one line that matches a family:  five of its seven peoples with a known language spoke Indo-European languages.  Ham's line holds more Semitic-speaking peoples than Shem's, twelve against five, and the oldest Semitic texts of all come from Babylonia, the Shinar that the Table gives to Nimrod and Ham.  Shem's line holds Elam, whose language is related to nothing, and Lud, which if it is Lydia spoke an Anatolian Indo-European language.  This does not count against Genesis 10, because the chapter never claims that a line is a language:  it lists tongue as one criterion among four, and a people can change its tongue.  It does rule out any use of the Table as a map of where the language families began, which is what Henry Morris and Anthony Osgood proposed.

Even the conservative case for the chapter's antiquity grants the point.  D. J. Wiseman in 1955:  "it cannot be denied that these 'colophons' correctly state that each list contains elements of geography, linguistics and physical affinities."¹⁰  But the lists "cannot be simply linguistic groupings;  e.g. Elamite (v. 22) so far as it can be traced is a non-Semitic language.  Moreover all attempts to trace existing languages back to these three parent groups have failed".¹⁰

The data.  The table lists every named people whose language is attested, with the family it belongs to in Glottolog 5.3, the standard catalogue of the world's language families,¹¹ and in The Cambridge Encyclopedia of the World's Ancient Languages (2004).¹  Only the identifications graded secure or probable in Genesis 10 Name by Name are counted.

The count by line is this article's own, from the table:

What the count shows.

1. Japheth matches a family. Five of seven are Indo-European: the Greeks, the Medes, Tabal, the Scythians and Togarmah's region. The two exceptions are both on Cyprus. Speiser, who thought the Japhethite list geographic, still saw it: "it so happened that most of the groups involved were of Indo-European stock, so that the list becomes ethnogeographic and ethnolinguistic in effect, as is actually indicated in vs. 5."¹² That is the strongest single linguistic point for the chapter. The branch that the text says was arranged "after his tongue" is mostly one family.
2. Ham holds more Semitic speakers than Shem does. Canaan and its coastal cities spoke Phoenician, which "is a member of the Semitic language family, specifically the Northwest Semitic branch of Central Semitic. Within Northwest Semitic it is a Canaanite language, the closest relatives of which are Hebrew, Moabite, Ammonite, and Edomite."¹ So the language of Canaan's sons, under Ham, is the nearest relative of the language of Eber's, under Shem. Akkadian, the language of Babel, Akkad, Nineveh and Calah, stands at the first branching of Semitic: "The earliest partition within the Semitic family separated West Semitic from Akkadian and Eblaite (see Ch. 8), which together are termed East Semitic."¹ Akkadian is attested from the "mid-third millennium", with Old Akkadian and Eblaite texts from "c. 2400 BC".¹ The oldest Semitic writing comes from the Akkadian-speaking cities of Babylonia, the land the Table gives to Nimrod and Ham, and from Ebla in north Syria, which the Table does not name.
3. Shem holds two peoples who did not speak Semitic. Elam: "The Elamite language is not related to Semitic, Sumerian, Hurrian, or Indo-European. The present grouping under Shem is due to political and geographic considerations."¹² Its texts begin "from about 2300 BC".¹ And Lud, if it is Lydia: "Lydian shares several characterizing innovations with Hittite and related languages and definitely belongs to the Anatolian subgroup of Indo-European as narrowly defined".¹
4. "Hamitic" is not a family. Ham's line spans Semitic, Egyptian, Libyan, Indo-European Hittite and Luwian, the isolates Sumerian and Meroitic, and Minoan. The old linguistic name "Hamito-Semitic" was dropped for that reason. Huehnergard, in the encyclopaedia's chapter on Afro-Asiatic: such names "have generally been abandoned because they imply a subgroup of 'Hamitic' languages (i.e., of all languages in the family apart from the Semitic languages) that is not indicated by any isoglosses."¹ Sumerian, the language of Erech, belongs to no family at all: "Sumerian is an isolate, like Ainu, Etruscan, Basque, or Burushaski."¹
5. The pattern does not depend on the doubtful rows. Drop every identification graded probable and Japheth is 3 of 4 Indo-European, Ham 12 of 18 Semitic. Drop Lud as contested and Shem is 5 of 6 Semitic, with Elam still outside.

Each side's explanation, in its own words.

• Mainstream critical: language was not the criterion where the lines cross families. Speiser on Canaan: "the subsuming of Canaan under Hamites appears at first to be strange. The solution is to be sought apparently in geographic and political connections rather than in ethnolinguistic relations."¹² On the Philistines, "the classification here given is evidently geographic".¹² In general: "the modern principle of classification is strictly linguistic, whereas the Bible employs several criteria concurrently, which cannot lead to uniform results."¹² Westermann goes furthest: "The genealogical pattern is only the form of presentation; it is not meant to indicate descent."¹³
• Conservative scholarly: repeated names mark layered populations. Kitchen: "some regions and communities were 'stratified' linguistically, culturally, and perhaps racially— new groups had overlain old, and another language had replaced a previously used one."⁸ That predicts two layers wherever a name stands under two lines. For Mesopotamia it is right: Akkadian overlay Sumerian. For Sheba and Havilah, which also stand under both Ham and Shem, no such layering is known. Allen Ross concedes the hardest case, Heth and the Amorites under Canaan, "a connection that cannot be supported by archaeology or history".¹⁴
• Creationist, Custance: "The three families are not predicated on the basis of language, either ... But when we come to the descendants of Ham we run into difficulties for it appears that in historic times the Canaanites, Philistines, and many Cushites spoke Semitic languages, while the Hittites (also Hamites, from Heth) may have spoken an Indo-European language. The trouble with linguistic evidence in this instance is that it really appears too late in history to be decisive."¹⁵ And the confusion at Babel "served chiefly as an affliction for the children of Ham, whose languages have proliferated bewilderingly from very early times to the present day ... The changes which took place in the Semitic family of languages were remarkably small."¹⁵
• Young-earth, Morris: Assyria was "a mixture of both Semitic racial stock and Hamitic (Babylonian) culture, language, and religion",¹⁶ and Shem's own speech escaped the confusion: "It is probable that this was a Semitic language (perhaps even Hebrew) ... it seems unlikely that Shem participated in the Babel rebellion; so it is probable that his own language was not affected by the resulting confusion of tongues."¹⁶
• Young-earth, Carter: the three lines overlap "across most of their range";¹⁷ no mapping of lines to languages is offered.
• Mainstream genetics: Iosif Lazaridis and his colleagues, in a 2022 study of 727 ancient genomes from Anatolia to Iran, put the general point: "languages can be replaced with little or no genetic change, and populations can migrate and mix with little or no linguistic change."⁶ Part 2 gives the clearest case, the Punic Carthaginians.

Assessment.  Since every side now agrees that the lines are not language families, the language test cannot be used as evidence against the Table.  The text never claims one line, one tongue.  It can be used against three specific young-earth claims.

First, Morris's "Hamitic (Babylonian) ... language".  The language of Babylon and Assyria was Akkadian, which is East Semitic.  There is no Hamitic language for Babylon to have spoken.

Second, the idea, in Morris and implied in Custance, that Shem's line kept the original Semitic speech while Ham's was confused.  That model predicts the oldest Semitic texts in Shem's territory and the greatest Semitic diversity there.  The oldest Semitic texts are from Babylonia, Nimrod's Shinar, and more Semitic-speaking peoples stand under Ham than under Shem.

Third, Custance's answer that the language evidence "appears too late in history to be decisive".  On the Masoretic text the dispersion at Babel falls 101 to 340 years after the Flood, about 2247 to 2008 BC on Ussher's Flood date (the Tower article sets out the window).  Old Akkadian and Eblaite texts of about 2400 BC and Elamite texts of about 2300 BC are older than that whole window, so on the Masoretic chronology the evidence is not late at all.  It is written before the dispersion it is supposed to postdate.  On the Septuagint, the dispersion falls 531 to 870 years after a Flood of about 3298 BC, so its earliest point is about 2767 BC, and the same texts are about four to five centuries after it.  That leaves the "too late" answer some room, on the Septuagint only.  This is the same split between the two chronologies that runs through the language-family articles.

Osgood's count belongs here too.  He expected "a total of up to 46 initial cultures and languages" after Babel.¹⁸  Glottolog lists 238 language families and 183 isolates, 421 units with no demonstrated relation between them.  That is not by itself evidence against Babel, because the comparative method loses its signal after 6,000 to 10,000 years on any chronology, a point Did Language Begin Once? argues in full.  But 421 is nine times Osgood's 46, and the Table's own seventy is not a count of language families either.

Part 2: Testing it against ancient DNA

Since 2016 geneticists have sequenced the bones of several peoples Genesis 10 names:  Canaanites from Sidon and the southern Levant, Philistines from Ashkelon, Egyptians of the Old Kingdom and later, Minoans of Crete and Mycenaeans of Greece, Bronze Age Anatolians and Armenians, Punic Carthaginians, Arabians of Bahrain and medieval Nubians.  The results fall into three groups.  First, the ethnography holds where it can be sampled:  Canaan's peoples were one genetic population, and the Philistines arrived from the Aegean at the start of the Iron Age, as Amos 9:7 says.  Second, the line assignments have no genetic signal:  Canaanites under Ham and Israelites under Shem share one ancestry profile, Minoan Crete under Ham and Mycenaean Greece under Japheth are "genetically similar", and the Punic Carthaginians spoke a Canaanite language with almost no Levantine ancestry.  Third, a few results bear on time, because they count generations rather than reading radiocarbon.  One fits a post-Babel Canaan on any chronology.  One, from Bronze Age Armenia, reaches the Masoretic date only at the edge of its error range and fits the Septuagint if the dispersion is put early in that text's range.  And the deep genetic differences among the first farmers of the Near East, which had to arise and then be mixed away before the Bronze Age, are hard to fit into the Masoretic centuries at all.  The deep genetics of human origins, the Y-chromosome tree and the out-of-Africa gradient belong to Y-Chromosome Adam and Noah and Out of Africa or Out of Babel?, and the radiocarbon dates of the skeletons to Egypt and the Flood date.

What the young-earth reading predicts, in its own words.  Robert Carter of Creation Ministries International states the genetic claim most carefully.  The Table "traces the male lines only.  Thus, it is also a record of Y chromosome descent."¹⁷  But the three lines overlap "across most of their range",¹⁷ and in a 2023 article on the Philistines he spells out how little that leaves:  "Yet, this is true for the Y chromosome only.  After the Flood, the family of Noah stayed together ... Thus, when we say, "the Philistines came from Ham", what we mean is that the men in leadership carried his Y chromosome."¹⁹  So the young-earth claim that ancient DNA can test is not that each Table people is a separate genetic cluster;  Carter disowns that.  It is that every people descends from the three brothers, and that all the population structure among them arose after the Flood and the dispersion.  On the Masoretic text the Flood is 2348 BC on Ussher's date and the dispersion falls between about 2247 and 2008 BC;  on the Septuagint the Flood is about 3298 BC and the dispersion between about 2767 and 2428 BC.

What ancient DNA can and cannot measure.  The papers below report three kinds of result.  They model a genome's ancestry as proportions from older sampled populations;  proportions alone fit a recent and a deep history equally well.  Some date a mixture in generations, from the way the mixed segments of DNA shorten with each generation;  that is a count of generations, independent of radiocarbon.  And they report the radiocarbon date of each skeleton, which is a separate question.  None measures when two source populations split, and none can read a person's language or the name of his people.

The ethnography holds.

• Canaan. Lily Agranat-Tamir and colleagues sequenced 73 people of the Bronze and early Iron Age southern Levant, from about 2500 to 1000 BC, and found that "different "Canaanite" groups genetically resemble each other more than other populations."² In their discussion: "This suggests that the archaeological and historical category of "Canaanites" correlates with shared ancestry".² Marc Haber's team sequenced five people from Bronze Age Sidon, about 1700 BC, and found that "present-day Lebanese derive most of their ancestry from a Canaanite-related population, which therefore implies substantial genetic continuity in the Levant since at least the Bronze Age."²⁰ Sidon, "the firstborn" of Canaan in 10:15, is a continuing people.
• The Philistines. Michal Feldman's team sequenced ten people from Ashkelon and found that "the early Iron Age population was genetically distinct due to a European-related admixture."³ The best model gives the newcomers' contribution as "around 43% of ancestry from the Greek Bronze Age "Crete_Odigitria_BA" (43.1 ± 19.2%)".³ That is Amos 9:7's "Philistines from Caphtor" in genetic form, and it arrives at the Bronze to Iron Age transition, when the Egyptian records first name the Peleset. The sample is four individuals, and the 43% carries an error of 19 points either way.
• Egypt and Kush. Egyptians from Abusir el-Meleq, 1388 BC to AD 426, "shared more ancestry with Near Easterners than present-day Egyptians".²¹ Medieval Nubians from Kulubnarti had "~43% Nilotic-related ancestry ... with the remaining ancestry consistent with being introduced through Egypt and ultimately deriving from an ancestry pool like that found in the Bronze and Iron Age Levant."²² Mizraim and Cush are brothers in 10:6, and their peoples share a component that reached Nubia through Egypt. Both samples are late: nothing here tests the Table's date.

These are real points for the Bible's history.  Young-earth writers have noticed them.  Carter writes of the Philistine result that "The migration of a large contingent of people from Crete to the coastal Levant fits the genetic and archaeological picture perfectly."¹⁹  Shaun Doyle of CMI answered the 2017 press stories that the Sidon genomes disproved the conquest narrative, accepting the result:  "There appears to be little reason to doubt the findings."²³  His point was the conquest, not the genetics:  "The Bible says they definitely survived, and persisted in the land, exactly as the DNA evidence indicates!"²³  Jeffrey Tomkins of the Institute for Creation Research went further:  "Once again science confirms the Scriptures as being historically accurate".²⁴  These papers do confirm that the peoples were where the Bible puts them.  They do not test whether the peoples descend from Noah's sons on a young-earth timescale, and none of the three responses discusses the admixture dates the papers report.  Doyle and Tomkins repeat the age of the Sidon skeletons, "around 3,750 years" and "about 4,000 years", without comment.²⁴

The line assignments have no genetic signal.

• Canaan (Ham) and Israel (Shem). Agranat-Tamir's team found that "both Arabic-speaking and Jewish populations are compatible with having more than 50% Middle Eastern-related ancestry" of the Bronze Age Levantine kind.² The ancestry the papers measure does not separate the people the Bible calls Canaanites from the people it traces through Eber. The Bible itself records heavy mixing between them (Genesis 38:2; Judges 3:5-6), so this is not a refutation. It means that in the one region where both lines are sampled, the line assignment carries no genetic content.
• Crete (Ham) and Greece (Japheth). The Table puts the Caphtorim, Crete, under Mizraim, and Javan, the Greeks, under Japheth. Iosif Lazaridis and colleagues: "We show that Minoans and Mycenaeans were genetically similar, having at least three quarters of their ancestry from the first Neolithic farmers of western Anatolia and the Aegean, and most of the remainder from ancient populations like those of the Caucasus and Iran."⁴ The Mycenaeans add a small "northern" share. The language line runs the Table's way, Mycenaean Greek against undeciphered Minoan. The genes do not.
• The Philistines after two centuries. Feldman: "within no more than two centuries, this genetic footprint introduced during the early Iron Age is no longer detectable and seems to be diluted by a local Levantine-related gene pool."³ The name "Philistine" outlived the ancestry that came with it.
• Carthage. Harald Ringbauer and colleagues sequenced 196 people from fourteen Phoenician and Punic sites of the sixth to second centuries BC: "Levantine Phoenicians made little genetic contribution to Punic settlements in the central and western Mediterranean between the 6th and 2nd centuries BCE, despite abundant archaeological evidence of cultural, historical, linguistic, and religious links."⁵ Their own question: "how and when Canaanite-Phoenician culture and language were adopted by people without any detectable Levantine ancestry."⁵ This is the clearest case on record of a people whose name, language and religion are Canaanite while its ancestry is not.

Each of these fits the reading of the Table as a classification by political, cultural and geographic ties, which Westermann, Speiser, Wenham and Ross all hold (The Table of Nations).  Each is also absorbed by Carter's overlapping lines.  The cost of absorbing them is the one How the Table of Nations Has Been Read named:  if a Philistine is a Hamite because his leaders carried Ham's Y chromosome, and no Y lineage has been assigned to Ham in a primary young-earth source read here, then no genome can count for or against the line assignments.  The young-earth reading of the lines is then not a genetic hypothesis, and it should not be claimed as genetically confirmed.

Three results that bear on time.

1. Canaan's admixture, counted in generations.  Haber's team dated the mixture that made the Canaanite ancestry from present-day Lebanese genomes:  "around 181 ± 54 generations ago, or 5,000 ± 1,500 ya assuming a generation time of 28 years".²⁰  At 25 years a generation that is about 4,500 years ago, with an interval that contains every young-earth dispersion date.²⁵  This result does not refute a post-Babel origin for the Canaanites, and it should not be reported as if it did.  Its own statistical support is weak:  the authors "took a lenient minimum Z-score = 2 to be suggestive of admixture".²⁰  What it does require of a young-earth model is that the two source populations, the first farmers of the Levant and a people related to Chalcolithic Iran, were already distinct when they mixed.

2. Armenia, counted in generations, with a radiocarbon anchor nobody disputes.  Lazaridis's 2022 study of 727 ancient people from Anatolia to Iran dates a mixture in Bronze to Iron Age Armenia:  individuals "admixed 52.2±8.0 generations (1460±224 years) prior to their average date of 1119BCE, or ~2579BCE (mid-3rd millennium BCE), assuming a generation length of 28 years".⁶  The anchor, 1119 BC, is late-second-millennium radiocarbon, a range the young-earth literature does not seriously dispute;  Egypt and the Flood date finds the historical and radiocarbon clocks within 178 years of each other there.  The generation length is checked in the same paper:  a regression of radiocarbon dates on generation counts across Balkan populations returns "the generation length at 28±4 years, virtually identical to its independent empirical estimation of 28 years."⁶  At 28 years the mixture falls about 2805 to 2357 BC;  at 25 years about 2625 to 2225 BC;  only with 20-year generations does the centre fall inside the Masoretic window.²⁶  With the paper's own 28-year generation the whole range is older than the Masoretic dispersion;  with 25 years only its young edge reaches it.  On the Masoretic dates, too, the two parent populations, the Early Bronze Age people of Armenia and people related to the steppe herders, would have had to form and become distinct after Babel and before they mixed.  On the Septuagint, with the dispersion at its earliest, about 2767 BC, the central date falls about two centuries, some seven generations, after Babel and fits;  with the latest Septuagint dispersion, about 2428 BC, the central date is before Babel and only the young part of the range fits.  Of the results in this section, this is the one that separates the two young-earth chronologies.

3. The first farmers were as different as Europeans and Chinese.  Lazaridis's 2016 study of 44 ancient Near Easterners found that the first farmers of the southern Levant and of the Zagros "were strongly genetically differentiated, and each descended from local hunter-gatherers.  By the time of the Bronze Age, these two populations and Anatolian-related farmers had mixed with each other and with the hunter-gatherers of Europe to drastically reduce genetic differentiation."⁷  The size of the difference:  "the populations at the four corners of the quadrangle had differentiation of F ST =0.08-0.15, comparable to the value of 0.09-0.13 seen between present-day West Eurasians and East Asians (Han)".⁷  Every Bronze Age people of Genesis 10 that has been sampled, the Canaanites, the Sidonians, the Egyptians, the Anatolians and the Aegeans, is a mixture of those streams.  On a young-earth reading all of that differentiation arose after the Flood and was then mixed away before the Bronze Age skeletons.  The order, differentiated first and mixed later, is seen in dated skeletons;  it is not a model output.  On Ussher's dates the time from the dispersion to Haber's Sidon of about 1700 BC is about 550 years, roughly 20 generations, for both steps.  Drift can produce a difference that large in ten generations only if each group stayed at an effective size of about 30 to 60 people, and a group that small would show inbreeding comparable to the children of first cousins or closer.²⁷  No young-earth source proposes groups that small, and no young-earth response to this paper was found.  The prediction can be checked against the runs of homozygosity in the earliest farmer genomes;  this article has not done that check.

One genome inside the Egyptian chronology dispute.  Adeline Morez Jacobs and colleagues sequenced a man buried at Nuwayrat, "Radiocarbon dated to 2855-2570 cal. bce, he lived a few centuries after Egyptian unification",²⁸ whose ancestry is best modelled as "77.6 ± 3.8% ancestry represented by genomes from the Middle Neolithic Moroccan site of Skhirat-Rouazi ... and the remainder most closely related to genomes from 9000 to 8000 bce Neolithic Mesopotamia (22.4 ± 3.8%)".²⁸  An Egyptian, under Mizraim, with a fifth of his ancestry from the region of Shinar fits a partly Mesopotamian origin for Egypt's founders.  It fits a Neolithic farming spread just as well, and the authors note that "the timing of the admixture event cannot be estimated directly".²⁸  His radiocarbon date is before Ussher's Flood;  on the Septuagint he lived within a century or two of Babel and was already a two-way mixture.  Whether that radiocarbon date stands is the question Egypt and the Flood date settles, and it is one genome.

What has not been sampled.  No ancient genomes were found for Bronze or Iron Age Mesopotamia, Shinar and Asshur, for South Arabia, Sheba, Hadramawt and the Joktanites, for Kerma or Napatan Kush, or for the Medes and Persians.  The only Arabian genomes, from Bahrain in about 300 BC to AD 600, are "a mixture of sources from ancient Anatolia, Levant, and Iran/Caucasus",²⁹ and they are too late to test anything.  The regions where the Table is densest in Shem's line are the regions with no ancient DNA.

Part 3: Testing it against geography

Every map is drawn from somewhere, and the Table's is drawn from the Levant.  Japheth's peoples form an arc from Ionia round through Anatolia to Media, and they are "north" only as seen from Syria and Palestine;  seen from Babylon, Media is east and Ionia west.  Ham and Shem are not regions at all:  both run from the Mediterranean to Mesopotamia and South Arabia, and they overlap in both.  The Table's far reach runs west into the Mediterranean, to Crete, Ionia and perhaps Spain, and south up the Nile to Kush.  It names nothing in the Persian Gulf, though the kings of third-millennium Babylonia moored ships from Dilmun, Magan and Meluhha at their own quays.  That fits a compiler in Canaan or Egypt, Mosaic or later.  It does not fit a record drawn up in Shinar near the dispersion.  It costs only the strongest form of the one-off reading, and it is evidence for no particular date.

The claim, from each side.  Kitchen, the conservative:  "From a Levantine (and even Egyptian) point of view, the groups and places under Japhet are entirely northern, in location or origin — northeast, north, northwest — in Anatolia/northwestern Iran, and the Aegean/eastern Mediterranean (and beyond)."⁸  Ham's line "is by far the longest series.  The basic coverage is northeast Africa, Arabia/Levant, and Mesopotamia, with outliers to the north and east."⁸  Westermann, the critic, says the same of Japheth:  "Looking at them from Israel, they are the peoples from the far north, further north than Syria and Assyria, the majority from Asia Minor across to Armenia."¹³  He reads the chapter as tracing the lines "across the whole map of the world as known at that time."¹³  Speiser:  for Japheth "The principal criterion was geographic."¹²  Arthur Custance, the creationist, rejected a purely geographic origin for the scheme, since the dispersion it would describe "did not occur until sometime later", but granted that "the writer knew only that some of Ham's descendants had entered Africa, that a large part of Shem's descendants had settled in Arabia, and that Japheth was still not very far to the north".¹⁵

The map.  Distance and bearing from Jerusalem to each place whose identification Genesis 10 Name by Name grades secure or probable, computed for this article.³⁰

What the map shows.

1. Japheth is a northern arc centred on the Levant. Every located Japhethite place lies between west-north-west and east-north-east of Jerusalem. From Babylon, Madai and Ashkenaz would be east and Javan west. The direction Ezekiel gives Togarmah, "of the north quarters" (38:6), is a Levantine one.
2. Ham and Shem are not regions. Ham runs from Crete and Cyrenaica through Egypt, Nubia and Canaan to Babylonia and Yemen. Shem runs from Lydia through Syria and Assyria to Elam and Yemen. The two overlap in Mesopotamia, where Nimrod's cities and Asshur stand, and in South Arabia, where Sheba and Havilah each appear under both. That is the overlap Carter describes, and the layering Kitchen proposes (Part 1).
3. The radius is about 2,200 km, with one exception. Every located place lies within about 2,200 km of Jerusalem, except Tarshish if it is Tartessos in Spain. That is the objection J. Simons raised to Tartessos, which Westermann quotes: "A further difficulty is that the map of the world would be extended considerably by this one name."¹³ Esarhaddon's one cuneiform Tarshish, which his editor reads as Tarsus, keeps the list inside the radius (Genesis 10 Name by Name).

The test that tells viewpoints apart.  Bearings alone cannot place the compiler, since a list drawn up in Egypt or Canaan would look much the same.  What can is the set of places that lie inside the Table's radius and are missing from it.  Measured from Babylon, Dilmun on Bahrain is about 920 km away, Magan on the coast of Oman about 1,510 km, and Meluhha, if it is the Indus, about 2,350 km;  Persis and Urartu, both missing from the Table (When Was the Table of Nations Written?), are about 860 and 670 km.³⁰  The Table reaches farther than that in other directions:  Javan is about 1,650 km from Babylon, Lud 1,620, Caphtor 1,800 and Cush 2,040.  And Babylonia knew the Gulf by name from the third millennium.  Sargon of Akkad, about the 24th century BC:  "At the wharf of Agade he made moor ships from Meluhha, ships from Magan, (and) ships from Tilmun."⁹  The translator, A. Leo Oppenheim, places Magan and Meluhha in that period "on the eastern shores of Arabia";⁹ the modern identifications of Magan with Oman and Meluhha with the Indus are the standard ones but are not argued here.  On either reading they are Gulf names.  A list of the world drawn up in Shinar would be expected to know the Gulf.  The Table looks west and north from the Levant instead.

What else the map leaves out.  Nothing east of Media:  no India, Central Asia or China.  Nothing in Africa beyond Kush.  Nothing in Europe beyond the Aegean, unless Tarshish is Spain.  Inside the Levant, no Ammon, Moab, Edom, Midian or Amalek, whom Genesis derives later from Lot, Abraham and Esau (When Was the Table of Nations Written?).  And for the date, no Persians, Chaldeans, Urartu or Arabs (When Was the Table of Nations Written?).

Assessment.  The Table's map is Levantine.  Its far reach is west into the Mediterranean, the direction of Phoenician and Israelite trade, and Westermann ties the Joktanite list to "South Arabian commercial shipping" under Solomon, following O. Procksch.¹³  For the young-earth side this is a real cost to one reading only.  If Carter, Lee and Sanford's "one-off documentation of the early post-Flood world"³¹ means a record drawn up at Babel or near it, in Shinar, then the Table should know the Gulf and should not call Media north.  Anthony Osgood's "photo start" of the nations "at the time of Babel"¹⁸ is that reading.  The map costs nothing to a reading in which Moses or a later Israelite compiled the list from a Levantine or Egyptian vantage, which is Wiseman's model and Kitchen's.  Wiseman put the compiler at the Egyptian court:  "the geographical information in Genesis 10 could have been available to the Egyptian court when Moses received his education there".¹⁰  For the skeptic, the Levantine viewpoint is what every Israelite date predicts, Mosaic, Solomonic or later, so it is no evidence against the antiquity of the core.  It tells against a Mesopotamian composition.  It also sits a little awkwardly with composition in the Babylonian exile, for the same reason, though an exiled writer working from older Israelite lists could keep their viewpoint.

Objections and responses

"Canaanites and Israelites were one people genetically, so Genesis 10 is wrong."  Genesis 10 does not claim that the lines are genetically distinct.  On the conservative reading the "sons" are partly political and geographic units, and Wenham and Ross say so (The Table of Nations).  The Bible itself records Israelite and Canaanite intermarriage (Genesis 38:2;  Judges 3:5-6).  The genetic result disproves a biological reading of the lines.  It leaves the chapter's ethnography untouched, and that ethnography passes the genetic tests where they exist:  Canaan's peoples were one population, and the Philistines came from the Aegean (Part 2).

"The language mismatches show that Babel's confusion fell unevenly on Ham."  That is Custance's answer (Part 1).  It is not testable as stated, since any distribution of languages could be called uneven.  The specific claims built on it fail:  the oldest Semitic texts are from Ham's Babylonia, not Shem's lands, and more Semitic-speaking peoples stand under Ham than under Shem.  And Custance's other answer, that the language evidence comes "too late in history to be decisive", fails on the Masoretic chronology, because Old Akkadian, Eblaite and Elamite texts are older than the Masoretic dispersion (Part 1).

Issues and unanswered questions

• The drift test is not done. Part 2 estimates that drift alone could build the difference between the first Levantine and Zagros farmers within ten generations only in groups of about 30 to 60, with inbreeding above that of first cousins' children. The runs of homozygosity in the earliest farmer genomes would test that directly. Ringbauer, Novembre and Steinrücken's 2021 study of homozygosity in ancient genomes is the place to look, and it has not been read.
• The young-earth responses to the genetics. Three were found: Carter 2023 on the Philistines, Doyle 2017 and Tomkins 2017 on the Sidon genomes. All welcome the results that fit the biblical history. None engages the admixture dates, the deep Neolithic differentiation, the Minoan-Mycenaean similarity or the Punic result. A wider search of the Journal of Creation and the Answers Research Journal may find more.
• Gaps in the ancient DNA. No ancient genomes were found for Bronze or Iron Age Mesopotamia, South Arabia, Kerma or Napatan Kush, or the Medes and Persians, which are the regions where the Table is densest in Shem's line.

Verdict

Genesis 10 passes the tests where it makes a claim the evidence can reach.  Canaan's peoples were one genetic population, the Philistines came from the Aegean, and the map is the one a compiler in the Levant would draw.  Its three lines are not language families and carry no genetic signal where two lines can be compared, but the chapter lists language as one criterion among four and never says that a line is a language or a gene pool.  The failure belongs to the readings that made those claims:  the use of the Table as a map of language origins, and any reading of the lines as Y-chromosome clades.

Read as a record of biological descent from three brothers about 4,300 years ago, the chapter meets two results that are hard to fit into the Masoretic centuries.  A Bronze Age Armenian admixture reaches the Masoretic dispersion only at the young edge of its range, and fits the Septuagint if its dispersion is put early.  The first farmers of the Levant and the Zagros differed as much as Europeans and Chinese and were mixed by the Bronze Age, which on Ussher's dates drift alone could do only in groups of a few dozen people.  No young-earth response to either was found.  Mesopotamia and South Arabia, where Shem's line is densest, have no ancient DNA yet.

Companion pages

• The literature: every source consulted for the Table of Nations articles, what it shows, and which way it cuts.
• Quotations: the key quotations behind the Table of Nations articles, each with its locator.
• The Table of Nations: what kind of document Genesis 10 is, and the verdict on every test.
• Genesis 10 Name by Name: every name, its identification, its earliest attestation outside the Bible, and its confidence grade.
• When Was the Table of Nations Written?: the horizon of first attestations, the peoples the list leaves out, each scholar's date, and the updating model.
• How the Table of Nations Has Been Read: Jubilees, Josephus, the medieval pedigrees, and the young-earth uses from Custance to Carter.

Sources