The Map of Indo-European Migrations

A map of Indo-European migration and the expansion of Indo-European languages.
The map is based on the hypothesis of symmetric wave expansion from central foci
with further migration along most suitable geographical routes.

(Prepared in 2008-2009, minor corrections 03-05/2012)



Indo-European migrations and the spread of Indo-European languages




Theoretical principles

Notes on the principle of wave migration

This is a more or less standard view of the Indo-European migration showing the spread of Indo-European language groups from an unknown PIE Urheimat area, which was situated somewhere near the Black or Caspian Sea before 2000 BC within the area bordered by the red dotted line. The center-of-gravity point of Indo-European language groups corresponding to the area of maximum language diversity is difficult to determine precisely, but it seems to be located in the Ponto-Caspian steppes, probably within the Don-Volga-Caucasus triangle.

The map is based on the wave migration hypothesis, which states that all human migrations tend to occur symmetrically choosing most suitable routes and avoiding inaccessible territories that impose geo-climatic constraints on the expansion. This allows to make rather detailed predictions of routes to India and Europe, since the number of physically allowable or economically viable passages in Eurasia tends to be rather limited. Most migrations must have proceeded along river valleys and arable land, and therefore can finally be reconstructed, especially if paleoclimatic and paleogeographic changes are taken into consideration.

In addition, this approach allows to pinpoint an idealized Urheimat for each ethnic group, which is represented by an imaginary spot in space-time, centered on the focus of the correspondent circular migration wave.

On the mechanism of migration

One likely error that people often make is assuming that the Indo-European migration was part of a powerful military aggression. This idea was popularized by Gimbutas, and was probably due to her psychological impressions of WWII. However, the slow pace of the migration reveals that this was unlikely for most of the time; rather we should assume that the migration was occurring in the form of gradual branching of homesteads of Indo-European farmers and cattle breeders. This process was generally dependent on the population growth rates during the late Neolithic/Eneolithic when the agricultural and technological advances increased birth rate, decreased mortality, and positively affected population dynamics.

Consider a simple example when a family of cattle raisers showed some positive demographic growth and had 2.5 grown-up children, so now they needed some extra pasture territory for their cattle and had to expand into a nearby empty field; as a result we have gradual, relatively peaceful expansion over time. It is true, however, that this field could have belonged to adstrate hunter/gatherers, but the contact with them did not have to be necessarily of a military type. It was likely to be occurring through intermarriages, acquiring the language of the more technologically advanced IE culture, migratory displacement, etc. Instead of imposing one's culture by force, it can be more effective to rather culturally seduce into it by offering new survival opportunities, such as more food and more efficient methods of transportation.

Apparently, this relatively peaceful spread displaced all the indigeneous languages by genetically integrating their speakers into the new-economy community and making the old-world languages fade into oblivion.

At times, there could also have been some fast, aggressive military expansions, skirmishes or erratic movements associated with social unrest, but generally this process seems to be rather steady, slow and peaceful, taking decades, ultimately thousands of years, and, for most of the time, probably even occurring at a constant rate (which can supposedly be calculated under certain conditions, hence, in theory, providing the possibility for dating PIE using purely geographical, non-glottochronological assumptions).


Continuous migration vs. fixed homeland

We should be cautious with the traditional term "homeland", or "'Urheimat", which now sounds more metaphorical than factual. In a real demographic process, migrations hardly ever stop at all and can hardly produce a once-and-for-all fixed, permanent region in space-time, associated with a historical homeland. Essentially, instead of asking, "Where do these people belong?", we should rather say, "What was the spatial and linguistic status of a particular ethnic group at a specific period of time?".

This continuous migration approach finally explains the difficulty of establishing an exact point in space where the Indo-Europeans supposedly came from, because possibly, in reality, no such geographic location ever existed. Accordingly, what we should do is attempt to determine population distribution for every period of time, which can finally be represented as an animation of changing migration maps that display a migration history of a particular ethnic group during the historical period in question.

Indeed, there can be convincing arguments for placing the PIE homeland virtually anywhere within the red dotted line, and there are sometimes as many PIE Urheimat theories as debate participants. On the other hand, an attempt to determine migration histories and consider the whole matter from the continuos-migration perspective should settle a number of tricky questions in the Indo-European studies and remove apparent contradiction among different theories, some of which may in fact represent nothing but different temporal and spatial stages of the same migration process.

However, we will still use the suitable traditional term "homeland" in many occasions, keeping in mind that this is just an idealized abstract concept describing a temporary historical stage of a continual process.

This continual migration may stop at times, becoming static, because something has impeded its movement, for instance, in a situation when a migration wave is being blocked by geographic obstacles, or it begins to split up into smaller currents after running into a mountain range, etc. But in most other cases we observe a rather dynamic, changing picture.

From this viewpoint, as mentioned above, an idealized homeland should simply coincide with the focus of a symmetric migration wave that, at further expansion, can produce other secondary waves according to the Huygens-Fresnel principle in optics thus creating secondary language branches. The details of this migration model should be disccussed elsewhere.


The broad homeland hypothesis

An alternative plausible suggestion to the classical Gimbutas' theory could be that the Proto-Indo-European language was spread on the broad territory that included all of the Ponto-Caspian steppe and even some areas in Kazakhstan which were uniformly covered by people speaking nearly one single language even during the late stages of the expansion. This is corroborated by the evidence from Kazakhstan, where the Kazakh language, originally spoken by horse-breeding nomads, does not show much dialectical differentiation and seems to be surprisingly uniform for the large territory it covers.

This hypothesis also finds support in the possible general ethnogeographic principle, according to which people tend to occupy as much territory as their environmental adaptation would allow (see ethnogeography), as well as the aforementioned principle of symmetric expansion predicting that a uniform geoclimatic zone would lead to the formation of stationary migration waves that would bounce off the habitat boundaries finally covering all available territory with similar landscape and climate.

In fact, the climate is the only obstacle for the eastward expansion from the Pontic steppes since the continental isotherms in Eurasia tend to run from north to south along the meridians causing more severe winters, less precipitation, and more drastic temperature seasonal variance as one moves east, further into the continent — a fact that possibly explains why the Indo-European distribution seems to be skewed along the northwest-to-southeast axis.

In other respects, there seem to be few natural obstacles that impede migration to the east, e.g., frozen rivers could be crossed in winter. Consequently, almost nothing would have precluded the Proto-Indo-Europeans from occupying all of the Ponto-Caspian grassland territory and move further into Kazakhstan.

As a result, the diversification of the Proto-Indo-European dialects could have occurred in a natural way as their speakers reached the steppe boundaries and adapted to new types of geographic environment.


The reasons for the dispersal could also be technological

It's often assumed that the Indo-European expansion could have resulted from some sort of accidental catastrophe, such as the flooding of the Black Sea. Even though this hypothesis cannot be excluded or could correspond to early stages of the PIE expansion, a more plausible explanation should view the Indo-European spread from the evolutionary perspective deducing its causes from the ability of the Proto-Indo-European peoples to adapt to their environment. In humans, this adaptation can only be achieved through more effective technology, superior tools and more advanced level of civilization. In fact, there are several phenomena that can be regarded as candidates for unique Indo-European technological inventions: (1) the invention of agriculture by 8000 BC (as suggested by Renfrew, 1987); (2) horse domestication by 4000-2500 BC; (3) the invention of the wheel and heavy carts (before Sumerians, c. 3000 BC?); (4) the invention of light, spoked wheel and fast horse-drawn chariot (first attested in Sintashta, c. 2100-1700 BC).

The kind of the earliest heavy cart was probably very similar in design to the two-wheel carts still in use in some parts of Mongolia, offering a unique insight into the early history of wheeled vehicles, as can be seen in the image below. It is also known as Bulgaric (Chuvash) urapa and Turkic araba, arba, so evidently its use extended across Eurasia, but its unique, localized usage by Indo-Europeans is unlikely.

Two-wheeled heavy carts in western Mongolia

On the other hand, the use of horse-drawn chariot and light carts seems to be particularly typical of the early Indo-European cultures and rather ubiquitous on their territory. The chariot was the only effective ancient mean of transport, the indispensable ship of the steppe allowing them to reach as far as India and the Atlantic. In the battlefield, it was so efficient that its use in Ukraine extended well into the 20th century when the tachankas with machine guns were used during the Russian Civil War (1917–1920s) and apparently even later.

Przhevalski's horses in the Askania-Nova reserve, Ukraine

Feral horses on Manych-Gudilo Island, Kalmykia

A tachanka
monument in Kakhovka, Ukraine

Therefore, it is plausible to infer that it was the wheel, the horse-drawn carts and the chariot that turned the Proto-Indo-Europeans into what we know them to be: the ancient conquerors of Eurasia and the world.


The migration of main the Indo-European branches and the location of their foci

Certain historical conventions, plausible hypotheses, and even some educated guesses have been used to elucidate the position of individual homelands of Indo-European peoples. It goes without saying that the present-day level of Indo-European studies does not allow to reconstruct the position of all the Urheimats with 100% accuracy.

The Indo-Iranian migration

Also see a more detailed map of Indo-Iranian migration.
The center of Indo-Iranian expansion was probably located near the western Pamir Mountains and along the Oxus (Amu Darya) River. [Note that many historical records as well as the presence of dry river beds (the Uzboy(s)) in the area indicate that the Oxus flowed to the Caspian Sea during the early antiquity!]. As of now, it is suggested herein that the wave of the Indo-Iranians somehow reached the Pamirs, where it split into (1) the Proto-Iranian ethnic group that stayed along the Oxus and the Uzboy, and (2) the Proto-Dardo-Nuristano-Indo-Aryan group that, after inhabiting the western Pamirs for at least a thousand years, finally reached India by slow agricultural expansion.

There seems to be only one viable passage to India going over the northern Hindu Kush to Nuristan, which is consistent with (1) the present-day geographical distribution of the Nuristani and Dardic languages, (2) the distribution of archaeological sites in northern India. Even though another route around the Hindu Kush and along the southern regions of present-day Afghanistan is also feasible, it left fewer or no traces in the form of the spread of ancient or modern Indo-Aryan languages, and therefore was probably not taken. Consequently, the idea of the Nuristan Bridge to India seems to be very plausible. The late Indo-Iranians reached the upper Oxus in the Pamir mountains (where it is known as the Panj River), which resulted in the early separation of the Wakhi ethnic group which exhibits many linguistic peculiarities and archaisms. From there,
they were probably advancing very slowly due to the presence of significant geographical obstacles, and finally reached Nuristan located on the other side of the Hindu Kush that can essentially be seen as India. They most likely used the Yarkhun river valley (near the Chitral district) to descend from the Hindu Kush first towards the Kabul river, and finally to the Indus valley.

This set of hypotheses is corroborated by the fact that the Pamir-Nuristan region is the area of maximum Indo-Iranian language diversity.

The Bactria-Margiana Archaeological Complex (BMAC) located along the Oxus seems to coincide with the early East Iranian or Proto-Iranian (rather than Indo-Iranian) habitat, since the newly obtained glottochronolgical dates (see the lexicostatistical comparison of Indo-European languages) indicate an early separation of the Iranian and Dardo-Nuristano-Indo-Aryan groups by about 3000 BC, although this conclusion is preliminary and quite controversial so far.

Note the extreme diversity of the East Iranian languages, which is often underestimated in other studies. In fact, herein, we no longer maintain the idea of a clear-cut division of Iranian languages into East and West subgroups. Rather we suppose that there exists an intricate web of Iranian branches which spread along the Oxus from the Panj river and the Vakhan corridor in the Pamirs to the southern coast of the Caspian Sea somewhere between 1500 BC to 500 BC. It's also important to understand that the Oxus is the main water artery in the region and could metaphorically bear the name of the "Great Iranian River", serving as the most important and sometimes the only route for the migration across Central Asia for thousands of years.

Ultimately, it's not at all clear how the early Proto-Indo-Iranians could have circumnavigated the Caspian Sea. Generally, there are three geographic alternatives; the early Proto-Indo-Iranians could have traveled:

(1) across the Ustyurt Plateau, the Karakum desert and the steppes of northern Kazakhstan, which all are arid, inhospitable and presently uninhabited territories that are difficult to penetrate. More archaeological and especially paleoclimatic research is needed to clarify this point. Also, note that the Caspian Sea level has transgressed and regressed many times, and was probably about 5 m (15 ft) lower circa 3000 BC [see Kasymov, The Caspian Sea, 1987], so at the time, part of the shallow northern Caspian could be dry, the northern coast line was most likely located further to the south, and the Proto-Indo-Aryans could cut some corners when traveling along this route.

(2) around the southern coast of the Caspian Sea (Gilan and Mazendaran districts of present-day Iran), the area which has highly favorable agricultural conditions. This hypothesis is contradicted by the absence of the R1a1 haplogroup in this region, which indicates that it probably had been occupied by other highly effective agriculturists at the time of the Indo-Iranian arrival.

(3) along the Yaxartes (Syr-Darya) River from the Aral Sea directly to the Fergana Valley and then further to the Pamirs, which is corroborated by the genetic distribution of the R1a1 haplogroup and the maximum diversity of the Indo-Iranian language concentrated near the Pamir mountains. [Note that there are reasons to believe the Aral Sea has formed only by the beginning of the new era, considering it is not mentioned in the old Greek and Persian historical records, except as an allusion to marshland located in that area]


The Balto-Slavic migration

Herein, Proto-Slavic has been integrated into the Balto-Slavic group, which is consistent with the considerable lexicostatistical and morphological proximity of Proto-Slavic, Old Prussian, and modern Baltic languages. The hypothesis of their genetic unity can now be regarded as highly convincing and has been demonstrated by many authors and many different means (for instance, see the Balto-Slavic section at this site). In a nutshell, it should also be mentioned that there is an about 50% cognate match between Lithuanian/Latvian and Slavic in Swadesh-100 (see here); a 45% match in Swadesh-200 in the study by Girdenis-Mazhulis (1994), and 47% in the study by Novotna & Blazhek, which demonstrates the close relationship of Baltic and Slavic groups (cf. only 35% for English-Russian in Swadesh-100). This is rathet solid evidence for the close proximity of the Slavic and Baltic language groups, previously argued to be differnt by unscrupulous authors that apparently did not do any lexical research.

The area of maximum Balto-Slavic language diversity can be traced to the forested regions of the upper Dniepr and Pripyat rivers in northern Ukraine and southern Belarus. Generally, it should be noted that the Dniepr river could have played a significant role in the northward movement of the Proto-Balto-Slavic peoples from the Pontic steppes. This area also has high levels of the R1a1 haplogroup concentration, generally associated with the Satem or Indo-European speakers.

According to the Vistula Veneti hypothesis, the Proto-Slavic homeland has been placed to the north of the Carpathian mountains (southeast Poland).

The Vendic subgroup (preserved in the form of Polabian until 18th century) can be shown to be a rather separate unity because of the drastic differences between Polabian and West Slavic languages, which can be explained either by its early separation from Proto-Slavic or late intense interaction with Middle German (further research required). The rest of the Slavic languages are sort of a historical Johnny-come-lately to the Indo-European scene, and many of them probably still preserve much of the dialectical continuum phenomena (esp. in the western area), therefore their internal classification should better be viewed and explained from the areal point of view, rather than the traditional East-West-South tripartite clear-cut division (see A Continuum of Slavic Languages). This traditional internal classification of Slavic has already been argued by several authors, including Isidore Dyen and others.


The Proto-Celtic migration

The Celtic languages are shown as several distinct groups because of the remarkable and obvious differences between Gaulish, Brythonic, and Goidelic branches. Basically, it is not correct to consider them all as "just Celtic", since these are in fact very lexicostatistically different (sub)groups with an early temporal separation. Also, note that we know very little of Lusitanian, Celtiberian and other obscure early representatives of the Celtic cluster, so their exact position within the Celtic branch is still questionable. (A possible inclusion of Proto-Albanian as either part of or close to "Celtic" has also been considered separately as a plausible hypothesis, see here).

In any case, the late Proto-Celtic homeland should be positioned to the north of the Alps, somewhere in southern Germany, probably along the upper Danube. The geographical reasons for this are evident: the Alps constitute a significant barrier for migration, whereas the Danube, on the other hand, is a very suitable route to central Europe from the area of the Black Sea.

Moreover, we should shortly consider the frequently mentioned R1b haplogroup and its relatedness to the early Indo-Europeans. Genetic studies demonstrate that the areas of Western Europe finally occupied by Celtic speakers have high statistical concentration of R1b haplogroup. The R1b is particularly frequent (1) in Basques (88%); (2) in apparently post-Pictic (herein, regarded as arguably pre-Celtic) population of northern England (80-90%); (3) in the Hausa population of Sudan (40%) and some other peoples of sub-Saharan Africa; (4) in Armenians (30%) and Ossetians (40%); (5) in Bashkirs (Turkic) in the Southern Ural Mountains (85%) (see wiki for details). These maximum concentration areas should correspond either to a haplogroup homeland or subsequently formed genetic refugia that remain after the rest of the territory was diluted by the flood of a differnt type of DNA brought by new-comers. The latter conjecture seems to be the case, judging from the multiple nature of these concentration islands characterized by their location in poorly accessible regions (such as mountains), and the scarcity of R1b in Asia (e.g. 0.5% in India). Therefore, the origins of this, so to say, "Sahara-Picto-Vascono-Uralo-Caucasian" haplogroup, which most likely originated before 15,000 BC, cannot be Indo-European, it should be regarded as linguistically unrelated to PIE and going back to the Mesolithic. As a result, the R1b pre-Celtic clans constituted a different linguo-genetic substrate that, after the arrival of Indo-Europeans, finally produced what we now know as "Celtic languages", which were the result of an "imperfect" acquisition of a certain mainstream Centum proto-language by the local pre-Celtic population.

Consequently, it's reasonable to assume that the Celtic languages must have been rather innovative and, in theory, could have preserved certain features from the R1b linguistic substratum, which was probably essentially Proto-Vasconic (judging from its 90% concentration among the Basques), although presently, it would be quite difficult to establish any elements in Celtic going back to Proto-Basque.

Finally, after the period of Roman Empire, these Celtic speakers again changed their language by switching to modern Romance and Germanic languages of Western Europe, such as Spanish, French and English, and now constitute the majority population of Western Europe, with the Celtic languages marginalized to the northwest of the British Isles.

[As a digression, one might also wonder how this unrelated R1b group got to northern and central Africa; it had probably happened with the migration of Neolithic farmers (after 8000 BC) traveling along the Nile into Sudan, Chad, Cameroon, and the Sahara which had been more humid before 3000 BC].


The Proto-Germanic migration

It seems to be geographically plausible that the early Proto-Germanic speakers moved to the area of northern Germany from the Black Sea region traveling across southern Poland to the north of the Carpathian Mountains.

Judging from the high concentration of R1b markers (~50-60%) and the diluted presence of R1a1 (~10-15%) in Germany, we may deduce the existence of the same R1b substratum as in the case of the Celtic migration. Apparently, the Germanic languages were also innovative and may result from an acquisition of the Indo-European superstratum by the original population of Central Europe.


The Proto-Anatolian migration

The Proto-Anatolians (including Hittite as the main representative) could in fact be nearly autochthonous to Anatolia, having stayed there for quite an extended period of time. This seems to the most likely hypothesis at the present point; it is based on the early split of the proto-Anatolian languages from the main Indo-European stem.

On the other hand, if we assume that the Proto-Anatolians moved to Anatolia from somewhere else, then the eastern route from the Caucasus region would be slightly more plausible, being supported by such arguments as (1) the relatively early appearance of Hittites in Anatolia by 1800 BC (whereas the journey around the Black Sea would have landed them there at about the same time as the Greeks, that is, about the time of the Trojan War (c.1200-1300 BC), when the Hittite Kingdom was already in decline); (2) the cuneiform writing and the use of cylindrical seals borrowed from the Assyrians, which indicates cultural contact with the eastern cultures rather than with Egypt.



Dating Indo-European proto-languages

The split periods of Proto-Indo-European subbranches can be calculated on the basis of historical, archaeological, genetic, and, in some cases, glottochronological evidence.


The Proto-Germanic stage can roughly be dated between 1800 BC and 250 BC, which approximately matches the dating of local Nordic Bronze Age culture (1800-500 BC) [The Trundholm sun chariot, the sun wheel sign, the petroglyphs depicting ships, bronze working, amber trade, burial mounds, etc]. On the other hand, the glottochronological evidence also tends to place the internal Proto-Germanic split in the second half of the first millennium BC. For instance, Toporova (2000) (based on Maurer and Schwartz) provides 500 BC as the date for the start of internal splitting in Proto-Germanic, whereas Starostin & Burlak (2001) date it to circa 100 BC (however the latter Starostin's nonlogarithmic glottochronology seems to yield incorrect dates).


Proto-Celtic can plausibly be dated by the Early Bronze Urnfield culture (1300-750 BC) [wagon chariots including chariot burials, fortified settlements, bronze, iron, inhumations in urns, burial mounds, etc]. As a result, the Iron Age Hallstatt (c. 800-475 BC) and La Tene (c. 500-50 BC) cultures are typically associated with late Proto-Celtic or the early Celtic periods.


The dating for Proto-Albanian (Proto-Illyrian (?)) is very uncertain. Besides, there are some reasons to believe that Albanian might in fact be connected to the Celtic languages, albeit this is controversial and deserves a separate discussion (see the lexicostatistical comparison of Indo-European languages).


Proto-Italic can be dated by the Apennine culture which entered the Italian peninsula by c. 1350 BC. Of course, Latin, as we know it, appears only by 500 BC after the fall of the Etruscans. The Venetic language, as revealed by its inscriptions (including a few complete sentences), was also closely related to the Italic languages, and for this reason not shown separately on the map above.


It is largely accepted that the Proto-Hellenic tribes entered the Balkan peninsula somewhere between 2000 and 1600 BC. For instance, Bartonek (1987) has 1900 BC for the start of the internal splitting of Proto-Hellenic.


Proto-Thracian & Proto-Phrygian
Proto-Thracian was a "satemized" language (apparently close to Balto-Slavic, not Armenian, as it's often claimed), first attested circa 500 BC, when it was mentioned by Herodotus. Phrygians must have migrated to Asia Minor at least by the end of the Hittite Kingdom (c. 1200 BC), which was probably connected to the migrations of the Sea Peoples. Phrygia began to flourish by 700 BC.


The Indo-European chariots could not be used on the forested landscape, so any archaeological evidence from this region is difficult to assign to any particular language group. Arguable glottochronological assessments seem to converge to dates c. 1000-1500 BC (1250 BC, Starostin (2004); and 1400 BC, Novotna&Blazhek (2007).

*Proto-Volgaic [the name is coined herein, hence the asterisk] must have been a hypothetical branch of PIE that traveled northward along the Volga river, and left archaeological traces in the form of the Poltavka culture (2700-2100 BC) [which was similar to the Yamna culture, but with an increased use of metal] and the later Abashevo /ah-BAH-sheh-voh/ culture (1600-1500 BC, spreading from present-day Kaluga Oblast to Bashkortostan) [The Abashevo culture includes inhumations in kurgans, copper-smelting, horses, chariots (Tambov Oblast), the sun and fire cults].

The existence of the *Proto-Volgaic branch is evident not only from archaeology, it is logically deducible from the present symmetrical expansion theory, since the Indo-European spread to the north from the Don River is mostly pre-determined by symmetry considerations, whereas the Volga River offers a very suitable route for the northeward migration at the point of the approach between the Don and the Volga. Presently, this "Volgaic" branch seems to be extinct. Arguably, it could be the pressure from Proto-Volgaic that prompted the Finno-Ugric peoples to expand in several directions, and finally made them advance as far as the Urals and the Baltic Sea coast.


*Proto-Arkaim-Sintashta (or *Proto-Andronovo) is another hypothetical branch of PIE that traveled behind the Ural Mountains creating the Sintashta (c. 1800 BC) and Arkaim /AR-ka-EEM/ (c. 1600 BC) settlements in to the southwest of Southern Ural [chariot burials, round fortified towns, inhumations in kurgans, etc]. Anthony & Vinogradov (1995) dated a chariot burial at Krivoye Lake to around 2000 BCE, which makes it the earliest spiked wheel ever found.

The contuation of this Sintashta-Petrovka culture further into West Siberia is known as the Andronovo culture /ahn-DRAW-naw-voh/ (2300-1600 BC). The Andronovo culture can hardly be seen as the Proto-Indo-Iranian stage:

The identification of Andronovo as Indo-Iranian has been challenged by scholars who point to the absence of the characteristic timber graves of the steppe south of the Oxus River. Sarianidi (as cited in Bryant 2001:207) states that "direct archaeological data from Bactria and Margiana show without any shade of doubt that Andronovo tribes penetrated to a minimum extent into Bactria and Margianian oases... Klejn (1974) and Brentjes (1981) find the Andronovo culture much too late for an Indo-Iranian identification since chariot-wielding Aryans appear in Mitanni by the 15th to 16th century BCE. [wiki]

Though Helimsky (1997) suggested that the Andronovo culture could in fact belong to the third or forth unknown bracnh of Indo-Iranians. In any case, the ethno-linguistic attribution of the Andronovo is very controversial.


Proto-Andronovo can also be arguably identified with the very early Proto-Tocharian branch. Proto-Tocharians were attested in the Turfan Depressin near the Taklamakan Desert in the form of Caucasoid Tarim mummies datable to c. 1000 BC.

"The cemetery at Yanbulaq contained 29 mummies which date from 1100–500 BCE, 21 of which are Caucasoid—the earliest Caucasoid mummies found in the Tarim basin. Notable mummies are the tall, red-haired "Charchan man" or the "Ur-David" (1000 BCE); his son (1000 BCE), a small 1-year-old baby with blond hair protruding from under a red and blue felt cap, and blue stones in place of the eyes... DNA sequence data shows that the mummies happened to have haplotype characteristic of western Eurasia in the area of south Russia...The textiles found with the mummies are of an early European textile and weave type and are similar to textiles found on the bodies of salt miners in Austria of around 1300 BCE... Han Kangxin (as cited in Mallory & Mair 2000:236–237), who examined the skulls of 302 mummies, found the closest relatives of the earlier Tarim Basin population in the populations of the Afanasevo culture situated immediately north of the Tarim Basin and the Andronovo culture that spanned Kazakhstan and reached southwards into West Central Asia and the Altai." [wiki].

However, the earliest Tocharian texts are attested only in 570 AD.

The Old Hittite Kingdom dates to 1750–1500 BC. Around 2000 BC, the region centered in Hattusa, which would later become the core of the Hittite Kingdom, had been inhabited by people with a distinct culture speaking a non-Indo-European language (Hattic). The early Hittites borrowed heavily from Assyrian traders — in particular, the cuneiform writing and the use of cylindrical seals, which might indicate that Proto-Hittite speakers moved into the area of Hattusa from the Caucasus Mountains not later than 2300 BC.


Late Proto-Armenian can be roughly dated to c. 1000 BC. The Yerevan City (then-known as Erebuni) was founded in 782 BC by an Urartian king Argishti I, and by 600 BC Armenia first emerges in historical records. The earliest testimony of the Armenian language dates to 410 BC (the Bible translation by Mesrob Mashtots).


It seems that horses were known to the Indo-Aryans, although the scarce evidence for their presence in the form of horse bones in the Bactria-Margiana Archaeological Complex (2200-1700 BC) is not sufficient (Mallory) to identify the BMAC with the Proto-Indo-Iranian Urheimat. [However, some sources in Russian cite Yermolova (1986) for the presence of horse bones at Kelleli (2000 BC), as well as seal stamps depicting horses, etc.]. Consequently, the dating of Proto-Indo-Iranian can hardly be deduced from historical or archaeological records. The current arguable glottochronological assessment suggests an early splitting of Proto-Indo-Iranian — not later than 3000 BC. The reason for this quite early dating is the remarkable lexical difference between modern Iranian, Nuristani and Dardo-Indo-Aryan languages (see lexicostatistical comparison of Indo-European languages).

The Mitanni tribes entered Anatolia at least as early as 1600 BC.
Mitanni can probably be seen as a branch that separated from the early Proto-Iranian or Proto-Indo-Iranian stem somewhere before 2000 BC, which later turned westward traveling along the southern coast of the Caspian Sea. As a result, Mitanni, though poorly attested, seem to preserve the word intial s- in "satta" (seven) and some other archaic features characteristic of the Indic languages, which are absent in Iranain languages. But (!), the claim of their direct descent from Indo-Aryans [such as by Mayrhofer (1986-2000)] seems to be highly implausible! [This point should be demonstrated separately in detail. In a nutshell, the Mitanni names can be explained from the Avestan language even better than from Sanskrit; the Mitanni gods can hardly be seen as Aryan, since the Iranians and Indo-Aryans had largely the same pantheon; some Mitanni lexical items, such as -mazda in Priyamazda are in fact recognizably Avestan; the number *eka (one) attested in Mitanni is not specifically Indic, but can be found in many Iranian languages. As a result, there's no reason to any assume specific long-range connections to India, and on the basis of the Occam's razor principle, the Indo-Iranian identification of Mitanni seems much more plausible].

Late Proto-West-Iranian can be dated to about 1000 BC.

Median is an extremely poorly attested language traditionally considered to be North West Iranian, whereas the Median Kingdom first appears in historical records in 836 BC, which is the earliest historical date for a West Iranian language. Moreover, the lexicostatistical analysis shows that Avestan and Old Persian were quite mutually readable, which is consistent with the quick propagation of the Zoroastrian faith in the early Persian society. Therefore, Avestan could also be tentatively placed into the West Iranian group. Note that the traditional historical dating of Zoroaster's activity is c. 614 BC (258 years before Alexander), whereas Old Persian appears to have been first established in written form by 519 BC.

Proto-East-Iranian & Proto-Saka
There are difficulties even in demonstrating that East Iranain and Saka languages once constituted a single proto-state, separate from the main Proto-Iranian stem. Consequently, we don't have any clear historical dates for their hypothetical unity. Generally speaking, Proto-Saka(n) does not exhibit the lenition typical of East Iranian languages, and therefore should probably be seen as a third independent Iranian branch.

The glottochronological and lexicostatistical assessments (Starostin (2004), and herein) tend to view the splitting of Proto-East Iranian, Proto-Avestan, Proto-Saka, and Proto-West-Iranian as a very complex process mostly occurring somewhere between 1200 and 500 BC.

There's hardly any specific evidence for dating this branch. Its geographical position on the "Indian side" of the Hindu-Kush, and the current lexicostatitical comparison may indicate its proximity to Indo-Aryan languages.

Lexicostatistically, Dardic can be viewed as the most archaic core of Indo-Aryan languages with the most phonological and lexical differentiation. One of the Dardic subgroups seems to have given rise to the Indo-Aryan Prakrits, but the exact details of this process are unknown and unlcear. Essentially, the question of dating Proto-Dardic is connected to the emergence of Indo-Aryan languages in India, which can roughly be dated to 1700 BC.

Parpola (1999) identifies "Proto-Dardic" with "Proto-Rigvedic", suggesting that the Dards are the linguistic descendants of the bearers of proto-Rigvedic culture ca. 1700 BC, pointing to features in certain Dardic dialects that continue peculiarities of Rigvedic Sanskrit, such as the gerund in -tvi [wiki].

Proto-Indo-Aryan (with Dardic exluded)
Late Proto-Indo-Aryan can be dated to the period between 1900-1200 BC. This dating is consistent with the period of the slow decline of the Harappa (by 1600 BC), Mohenjo-Daro (by 1900 BC), and BMAC (c. 1700 BC) civilizations, which indicates that certain migration processes could have taken place in that area around that time, though it is not obvious whether there was an Indo-Aryan military invasion into the Indus Valley civilaztion, or its disappearance was due to climatic changes.

A regional cultural discontinuity occurred during the second millennium BCE and many Indus Valley cities were abandoned during this period, while many new settlements began to appear in Gujarat and East Punjab and other settlements such as in the western Bahawalpur region increased in size. Shaffer & Lichtenstein (in Erdosy 1995:139) stated that: "This shift by Harappan and, perhaps, other Indus Valley cultural mosaic groups, is the only archaeologically documented west-to-east movement of human populations in South Asia before the first half of the first millennium B.C.." This could have been caused by ecological factors, such as the drying up of the Ghaggar-Hakra River and increased aridity in Rajasthan and other places. The Indus River also began to flow east and floodings occurred [Flam (1981, 1991) and Mackay (1938, 1943)] [wiki].

The Cemetery H culture in Pumjab (1900-1300 BC) [including cremation in urns unlike the Indus civilization where bodies were buried in wooden coffins, the sun or star motifs, the rice as main crop, the use of mud bricks for building, etc] and the Gandhara grave culture in Pakistan (1800 BC- 600 BC) [including inhumations, horse burials, etc.] seem to mark the earliest archaeological presence of the Vedic Aryans on the Indian subcontinent.

Vedic Sanskrit is often (glottochronologically, or rather just by the rule of thumb (?)) assigned to about 1000-1200 BC. The oldest Indo-Aryan kingdoms are known to be well-established by about 700 BC.


Proto-Indo-European in general
The relatively late historical appearance of individual Proto-Indo-European branches — no earlier than 2000 BC in most cases — leads to the conclusion that PIE (except for Proto-Anatolian) could have still existed as a single unity, or a group of closely related dialects somewhere circa 3000-3500 BC. The currrent glottochronological analysis rather suggests that the Proto-Indo-European split occurred circa 4000 BC (without Proto-Anantolian) or circa 5500 BC (with Proto-Anatolian included).

The frequent question is how do these estimations conform to the Gray-Atkinson model (2003)? Look at their model in the following way. The Hittite and Tocharian data they added to the preexisting Dyen's datasets are most likely miscalculated for the simple reason that these data cannot fill in the 200-word lists they used, nor is it easy to find correct cognates for these languages, nor calculate glottochronological corrections for their greater age in a proper fashion. What is left of their model is our conclusion that the more or less statistically robust split must have occurred between 7300 BP and 6500 BP. That was the period of the differentiation of the main Indo-European branches (Greek, Indo-Iranian, Balto-Slavic, Italic, Germanic) that have many representatives, which reduces the possibility of statistical errors. That period is equivalent to 5300-4500 BC, which is just 900 years away from the estimations in the current articles, and that seems to be pretty close (especially considering both the incompleteness of the research in this publication and the outdatedness of their Dyen's datasets). The difference is, of course, due to misjudgments in the determination of cognates and calibration timing, but, in any case that can be seen as good first approximation agreement.



2008-2009 (c)