Out-of-India - From Theory to Truth: Part 2

[ukumar]

Sir, you have good argument for M17 but missing one crucial point for Z93. Z93, its parent (M417,z85,z645,z651) and cousin (z283,z280) are found in steppe ancient DNA. Whereas there is no report of parent (M417,z85,z645,z651) and cousin (z283,z280) in India so far. As such, Z93 probably came from outside. But arrival date is outside AIT window and its argument against AIT.

That is a useful data point. Taking images from 2 different Underhill papers we can map out M17 first and then M282 and M93.
With M17 being oldest in India in the first image (inset), the spread is over India, East and parts of west Europe and Siberia

The second image from another Underhill paper puts origin of M417 in Iran - and M17 the father of M 417 occurs in Iran as well. M417 bearing people could have back migrated to India spawning Z93 on the way an spread to Russia and Siberia. Meanwhile Z282 spawned from M417 has moved northwest to Europe.

Let me make an educated guess here. I need to look at Parsi genes. Could Parsis be responsible for spread towards Europe - with their "sa"-"ha" confusion getting to Greece?


M17
http://www.nature.com/ejhg/journal/v18/ ... 9194a.html


Z282/Z93
http://www.nature.com/ejhg/journal/v23/ ... 1450a.html

Yes, that is my guess too. I think r1a1a diversified in Iran and went to both Europe and India. You may want to read http://new-indology.blogspot.com/ if you haven't already. I think he is more closer to truth.

[ukumar]

These "Iranians" may have been part of early migrations that contributed to ANI because they went and mixed with ASI/Onge Andaman genes that are 24,000 years old in India (and 48,000 years old in Andamans)

But if you take Underhill dates - one paper puts the earlier M17 as over 12,000 years older - and the later M417 as 5800 and teh still later Z93 is undated. That would indicate migration of M17 to Iran, later mutation to form M417 and back migration to India of Z93. But Z93 is dated elsewhere as 15000 years old - which is older than its father M417 of Iran. The number of Iran samples of M417 is very small (4 or 6 can't recall) so that may require updating in due course

I think This contradiction is because you are comparing dates arrived from two different methodologies. Z93 can not exist before M417 mutation first and must be younger than M417. z93 date above is over estimated.

[hanumadu]

Did you not read lucottes paper that has been referenced consistently? The oldest z93 marker is found in PunjaI which either meant that it came from India or we have migration of earliest marker. m417 branched into r1a1a1a and r1a1a1b2 in pontine steppe. Highly possible we could have had bulk flow of the earliest z93 marker which then went out of India again which would explain why we don't have m417 and r1a1a1a but have the oldest z93 marker while Europeans have the yyoungest. This would also explain why the have the yoingest m17 while pakhoons have the oldest.

Yes, I have read it. It's based on modern DNA and its STR based TMRCA dates are known to be over estimated by 3 times. STR variance based analysis is outdated and superseded by better methodology ( whole genome analysis, ancient DNA etc)

Sorry, I don't understand your gene flow direction explanation. In absence of immidiate ancestors of Z93 in India, how can you avoid Arrival in to India?

Yes, that is my guess too. I think r1a1a diversified in Iran and went to both Europe and India. You may want to read http://new-indology.blogspot.com/ if you haven't already. I think he is more closer to truth.

This is pretty much what RoyG said.

[hanumadu]

He was gung ho and invested too much into this without deep study. Or he knew he was lying but carried on nevertheless. He is cracking and going by his tweets he is looking for support from others like Kalavai Venkat. But KV has his own doubts now.

And you are to blame for his predicament.

In the comments section of Prem's article I have made a detailed rebuttal of archaeology and linguistics and AIT and have stated how it could mean OIT also

Link to Prem's article
https://www.myind.net/there-genetic-evi ... heory-myth

Its good that your knowledge on the subject and others' too is being seen by a wider audience outside BRF without having to wait for your book to complete.

Thanks to Prem Kumar too for writing that article.

[ukumar]

Sorry, I don't understand your gene flow direction explanation. In absence of immidiate ancestors of Z93 in India, how can you avoid Arrival in to India?

Yes, that is my guess too. I think r1a1a diversified in Iran and went to both Europe and India. You may want to read http://new-indology.blogspot.com/ if you haven't already. I think he is more closer to truth.

This is pretty much what RoyG said.

My bad. I didn't understand as much We are on same page then.

[peter]

These "Iranians" may have been part of early migrations that contributed to ANI because they went and mixed with ASI/Onge Andaman genes that are 24,000 years old in India (and 48,000 years old in Andamans)

But if you take Underhill dates - one paper puts the earlier M17 as over 12,000 years older - and the later M417 as 5800 and teh still later Z93 is undated. That would indicate migration of M17 to Iran, later mutation to form M417 and back migration to India of Z93. But Z93 is dated elsewhere as 15000 years old - which is older than its father M417 of Iran. The number of Iran samples of M417 is very small (4 or 6 can't recall) so that may require updating in due course

I think This contradiction is because you are comparing dates arrived from two different methodologies. Z93 can not exist before M417 mutation first and must be younger than M417. z93 date above is over estimated.

Is it possible for you to explain in a bit non techy language why the conclusions of two underhill papers are different?

[shiv]

These "Iranians" may have been part of early migrations that contributed to ANI because they went and mixed with ASI/Onge Andaman genes that are 24,000 years old in India (and 48,000 years old in Andamans)

But if you take Underhill dates - one paper puts the earlier M17 as over 12,000 years older - and the later M417 as 5800 and teh still later Z93 is undated. That would indicate migration of M17 to Iran, later mutation to form M417 and back migration to India of Z93. But Z93 is dated elsewhere as 15000 years old - which is older than its father M417 of Iran. The number of Iran samples of M417 is very small (4 or 6 can't recall) so that may require updating in due course

I think This contradiction is because you are comparing dates arrived from two different methodologies. Z93 can not exist before M417 mutation first and must be younger than M417. z93 date above is over estimated.

The other possibility is that the four (or six) Iranian samples that showed a clustering of M417 at 5800 yrs ybp is not representative of the entire truth because of sampling bias.

Sampling bias itself is well recognized and even mentioned in at least one genetics paper that I have where it was pointed out that Pakistanis (presumably sampled because they were living in the West) were over represented in "South Asia" genetic studies. This is still going on and in a paper I read yesterday - may have been the Lucotte study there were 300 Pakistanis and 400 Indians. Now how can that be representative of "South Asia" - they call it "Pakistano-Indian". This is laughable and specious. India has 6 times Pakistan's population and 4 times the area but gene researchers take whoever is convenient - often people of subcontinental origin living outside India - such as Gujaratis in the US as representative of "Indo-European speakers" of India

And again in the last few days - among the papers mentioned on this page they have used a population of Indians living in Malaysia as representative of "India/South Asian" ancestry. With this type of gandmasti going on in science we cannot really expect too much of the exact truth. We have to take results as an approximation of reality.

But I thought the sampling in some papers - particularly the Reich paper and one by Thangaraj, Tamang, Singh etc to be good.

[shiv]

Here is the Lucotte paper:
http://www.omicsonline.org/open-access/ ... 000150.pdf

Check the table

They have taken 280 Pakis from Islamabad and Kraachi, and 484 Indians from "Punjab, Gujarat, Andhra and Tamil Nadu"

But the paper says:

We have refound in our samples the clear distinction initially
established by Pamjav et al. [21] between Indian Z93 populations and
European Z280 populations: all our South Asian populations are Z93,
while almost all our European populations are Z280. Datations show
that the Z93 Pakistano-Indian group is the most ancient (about 15,5 K
years); in Europe, the Eastern populations are the most ancient (about
12,5 K years) and the Northern ones the most recent (about 6,9 Kyears).

[Prem Kumar]

Shiv: thanks for your insights & great comment on the MyInd piece

Muppala & others: thanks for sharing

ukumar: plz read the chronology I proposed in the last page. Even if we assume Lucotte's chronology is superseded by Underhill's 2014 paper, I don't see how it changes things. Moreover I agree with Shiv's view: it is a documented fact (one of the papers mention it) that the earlier genetics studies over-represented Paki populations which was subsequently found to be a grossly misleading because Paki populations are similar to European populations and don't show the R1a1 diversity that you find within India. I suspect this habit hasn't stopped

Btw, read one of the very well-researched papers by Swarkar Sharma that I referenced in the MyInd piece. This talks about R1a1* widespread diversity in India and attributes it to Brahmins (it also points out its presence in Indian tribes at a frequency that's greater than that of Central Asians or Europeans!)

http://www.nature.com/jhg/journal/v54/n ... 0082a.html

[shiv]

Moreover I agree with Shiv's view: it is a documented fact (one of the papers mention it) that the earlier genetics studies over-represented Paki populations which was subsequently found to be a grossly misleading because Paki populations are similar to European populations and don't show the R1a1 diversity that you find within India. I suspect this habit hasn't stopped

And in case you are asked by the Manasataramginis or Kalavai Venkat's of the world to provide a peer reviewed journal ref for that and don't feel like saying "shiv said.." here is one:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3234374/
Am J Hum Genet. 2011 Dec 9;89(6):731-44. doi: 10.1016/j.ajhg.2011.11.010.
Shared and unique components of human population structure and genome-wide signals of positive selection in South Asia. (Metspalu et al)

In contrast to Pakistani populations, populations of Indian origin have been underrepresented in previous genomic scans of positive selection and population structure. Here we report data for more than 600,000 SNP markers genotyped in 142 samples from 30 ethnic groups in India. Combining our results with other available genome-wide data, we show that Indian populations are characterized by two major ancestry components

Modeling of the observed haplotype diversities suggests that both Indian ancestry components are older than the purported Indo-Aryan invasion 3,500 YBP. Consistent with the results of pairwise genetic distances among world regions, Indians share more ancestry signals with West than with East Eurasians. However, compared to Pakistani populations, a higher proportion of their genes show regionally specific signals of high haplotype homozygosity.

Genome-wide scans on the Human genome diversity panel (HGDP) data involving 51 global populations have revealed that South Asia, represented by Pakistani populations, shares most signals of recent positive selection with populations from Europe, the Near East, and North Africa

Importantly, the Pakistani (Indus Valley) populations differ substantially from most of the Indian populations and show comparably low genetic differentiation (within the FST range of 0.008–0.020) from European, Near Eastern, Caucasian, and Indian populations

[shiv]

Reading and quoting genetics papers is one thing but I now realize that one needs to look at the background of the authors in detail and what they are trying to prove. many authors are archaeology and history depts trying to infuse new life into a subject that might be dead by co authoring with geneticists who are at the sharp cutting edge of modern research.

It is also important to look at the size of population studied and exactly where they came from because, like I said papers have been published using Gujaratis or Pakis in the US and South Indians in Malaysia. These details become unimportant to the person who is looking for exciting new data but they are important to reviewing the research.

[A_Gupta]

Shiv, in your comment on Prem Kumar's article you wrote:

All the archaeological findings of chariots, horses, wheels, corded ware, pottery etc have ZERO evidence of any language. You cannot look and horse bones or pottery or human bones and guess the language spoken at the time of those archaeological findings.

I do not disagree with you. But there is this way of looking at things - from the Rg Veda, you can try to deduce a material culture and natural surroundings, and then ask whether it matches the archaeological record from whatever era. So you might not be able to have any evidence of language; but you may have evidence of whether a text was contemporary to particular archaeological findings; and hence deduce something about the language.

[shiv]

Shiv, in your comment on Prem Kumar's article you wrote:

All the archaeological findings of chariots, horses, wheels, corded ware, pottery etc have ZERO evidence of any language. You cannot look and horse bones or pottery or human bones and guess the language spoken at the time of those archaeological findings.

I do not disagree with you. But there is this way of looking at things - from the Rg Veda, you can try to deduce a material culture and natural surroundings, and then ask whether it matches the archaeological record from whatever era. So you might not be able to have any evidence of language; but you may have evidence of whether a text was contemporary to particular archaeological findings; and hence deduce something about the language.

There is a serious garbage in garbage out problem with this theory, In the case of IE languages - the assumption is made that an Indo-European language was associated with those graves/bones etc.

How can that assumption be made? It could be any language. Clearly this is a case of "petitio principii" where the premise is the same as the conclusion. IE languages are widespread over x, y and z geographical areas today , therefore the language in one of those areas 4000 years ago was, even in the absence of any clues an IE language. And the date of the language is the date of the grave. Iran now has Persian which has a lot of Arabic in it. Whoda thunk that it was originally IE? South America has IE languages now. Why not assume that the ancient language associated with South America graves was IE? In each case these assumptions are not made because there is other evidence to show reality. In the case of BMAC/Andronovo/Sintashta graves the absence of evidence of any language is used as an excuse to cook up the story that it was IE. That is utter bullshit.

This is like digging near a graveyard and saying that a grave has been found even without any bones, headstone or coffin because there are other graves in the vicinity

In terms of science this is complete nonsense. But it is par for the course for linguistics.

[A_Gupta]

There is a serious garbage in garbage out problem with this theory, In the case of IE languages - the assumption is made that an Indo-European language was associated with those graves/bones etc.

I'm not sure you understood. Let me try again. I mean if a text shows a landscape with birch trees and snow, a large difference between the length of summer days and winter days, etc., then that can be tried to be matched up with a geographical area. If a text shows certain material culture, say spears, but not swords, then that can be matched with archaeological findings. I'm not saying that is the case; I'm saying that the argument that language and archaeology without written artifacts found are totally unlinked is not true.

[UlanBatori]

Small problem with recent neuj that grave remains have been found in site well inside Rajasthan - much bigger town than Mohenjodaro or Harappa.

Delicate pooch: So Pakistan was much greater in those times? When did the Vedic custom of cremation come to be adopted? Was it, ahem, imported along with horses' behinds from Germany/Poland?

Isn't going by grave remains, sort of biasing the statistics? If 1% buried their dead to wait for the Great Ascent, and 99% simply cremated them and sent any remains downriver, aren't we going to have a rather biased idea of the population?

Now about this writing bijnej. Reading The HINDU sitting in rural Kerala, I had the notion that Madras was a city in England: the writing was entirely in a British sort of context. Hard to convey unless one is from rural India but I think you might grasp what I mean.

Even today, Indians use terms like "Did Yeoman Service". Wtf is a "Yeoman"? Lot of those in India?

[ukumar]

Is it possible for you to explain in a bit non techy language why the conclusions of two underhill papers are different?

I'll try. I am writing this from memory so pardon me for any errors.

First thing to understand is nature of the research in population genetics. Genes by themselves are real thing. But they don't provide direct ancestry and chronological information. Over the years researchers have developed different analytics method to infer ancestry information from genes. There has been gradual improvement in accuracy and details derived from genes. Unfortunately reachers are slow to incorporate new learning and we would see published papers long after the method is discredited. So to reconcile findings from different papers, one need to consider limitations of used method.

SNP and STR:

Initial phase started with ability to sample certain mutation locations (SNP) within Y DNA and mtDNA. Study of the mutation was very effective in providing uni-parental ancestry. This is when different haplogroup (like R1a, R1b, R1a1a etc) were identified and showed that all modern humans traces back to single group of parental group in Africa. Over the years more and more sub groups were identified providing additional clarity. Please refer to https://en.wikipedia.org/wiki/Haplogrou ... 1a1a1b1.29 to see how details emerged within R1a clade over the years.

Reachers used mutation variance (STR) to infer coalescing dates for happlogroups. Dates derived from such analysis are now shown to overestimate by ~3 times. Coalescing dates also depended on number of nodes in branch. As more nodes are identified, dates changed. One thing to keep in mind is that variance is function of both effective population size and time depth.

Very early papers for Indian genetics were published using this method and before Z93 and Z283 were discovered. For all practical purpose they are now outdated and shouldn't be used by layman.

Autosomal DNA:

As ability to sample DNA improved, reachers started to analyze segment of DNA for all chromosomes. David Reich's lab came up with methodology to identify admixed component from DNA. They applied it first for India to identify ANI and ASI components. Later they established that Europeans are also admixed population (WHG, EHG, EEF, CHS etc).

Reich's lab then came up with method to identify admixture dates and published http://genetics.med.harvard.edu/reich/R ... a_Date.pdf which we are all familiar with. One limitation of this method is that it only dates recent admixture. So if ANI admixed ANI+ASI population after long time, it would only date recent admixture.

Some papers derive relative chronology of component in different population using a variance. It also requires assumption about effective population size and mutation rates. One need to consider mutation rates used to compare dates from different papers.

Whole Genome:

It is now possible to sample and analyze whole genome. Whole genome analysis is the most accurate method today for modern DNA analysis. Recently published paper http://www.nature.com/ng/journal/vaop/n ... .3559.html from 1000 genome would be most up to date. This paper shows that expansion of Z93 occurred in last 5500 years.

Ancient DNA:
With ability to sample DNA from ancient remains, one can now apply all above methods on ancient DNA. Ancient DNA have added benefit of independent dating and location. IMHO, finding from ancient DNA trumps any from modern DNA.

I hope this helps.

[UlanBatori]

Same question. What good is DNA analysis for a population that used cremation? I am surprised that they haven't concluded that most Indians who lived between ~1700 and 1947 were of Neanderthal descent like most British.

[Karan M]

https://www.myind.net/there-genetic-evi ... heory-myth

I can't see the Hakim's replies anywhere.

A suggestion (IMHO) to Hakim.

Keep a blog with all your replies listed. All that effort goes waste if some webpage goofup wipes out the serious effort undertaken.

[ukumar]

ukumar: plz read the chronology I proposed in the last page. Even if we assume Lucotte's chronology is superseded by Underhill's 2014 paper, I don't see how it changes things. Moreover I agree with Shiv's view: it is a documented fact (one of the papers mention it) that the earlier genetics studies over-represented Paki populations which was subsequently found to be a grossly misleading because Paki populations are similar to European populations and don't show the R1a1 diversity that you find within India. I suspect this habit hasn't stopped

Btw, read one of the very well-researched papers by Swarkar Sharma that I referenced in the MyInd piece. This talks about R1a1* widespread diversity in India and attributes it to Brahmins (it also points out its presence in Indian tribes at a frequency that's greater than that of Central Asians or Europeans!)

http://www.nature.com/jhg/journal/v54/n ... 0082a.html

Prem Kumar,

Thanks. I read through the paper and have few comments for you to consider. Paper is from 2009 and doesn't mention/test down stream r1a1a branches Z93/Z283. Y DNA result in Table 1 shows it clearly. So it is likely that all R1a1* they found was in fact Z93 (and European was Z283) as results since 2009 has shown. They even mention the need for higher resolution to validate their conclusion: "However, it is important to discover novel Y-chromosomal binary marker(s) for a higher resolution of R1a1* and confirm the present conclusions"

Sampling bias does exist but it doesn't impact age of mutations. There is always a chance that further sampling would discover non Z93 R1a1a in India but it hasn't materialized in last few years so I am not hopeful.

[ukumar]

I guess he is referring to the Underhill paper from 2014 which we've been discussing above. If Underhill is stating that M417 coalescence is 5800 YBP, it means that Z93 mutation happened at that timeframe. But I don't see how that supports AIT.

1) Underhill's 2014 paper states that R1a* (M420) originated in Iran circa 25000 years ago

2) This paper doesn't contradict his 2010 paper, because the 2010 paper said that R1a1a* diversity is highest in India. Remember that R1a1a* (M17) is a grand-son of R1a* (M420)

3) The 2014 paper also does not contradict that Z93 originated in India. In fact his paper shows a distinct split between European males (almost all of them fall under Z282) vs South Asian males (almost all of them fall under Z93)

So, the way I read it, the following is the chronology:

1) R1a* originated in Iran 25K years ago (2014 Underhill)
2) We don't know much about its direct descendant R1a1*
3) Its grandson R1a1a* (M17) has maximum diversity in India (2010 Underhill)
4) Great grandson R1a1a1* (M417/Page7) split into Z280 (Europe) and Z93 (South Asia) circa 5800 years ago (Underhill 2014)
5) Z280 are mainly confined to Europe - it & its children are missing in South Asia
6) Z93's main child (M780) has maximum diversity & frequency in India and decreases as we go away (into Iran, Central Asia etc)
7) Z93's has other children - for example, Z93* is present in South Siberia and its lack of diversity there is attributed to Founder Effect (maybe one of our Rishis went there and had lots of babies)
Ancient DNA from 3 German males circa 4600 years ago (Corded Ware culture), who geneticists think, carried R1a1a*

In summary, it looks like Iranians moved to India sometime after 25000 years ago. Then the R1a1a* (M17) mutation arose in India (don't know when). Then, sometime circa 5800 years ago, the Z280 (European) & Z93 (South Asian) mutations arose. The German graves are after this.

I don't see any AIT here. There are hints of OIT, for sure

Key question is when and where #4 occurred. 5800 bp proposed by Underhill seems most likely. You would have to reconcile ancient DNA finding to firm up your proposal.

[ukumar]

I think This contradiction is because you are comparing dates arrived from two different methodologies. Z93 can not exist before M417 mutation first and must be younger than M417. z93 date above is over estimated.

The other possibility is that the four (or six) Iranian samples that showed a clustering of M417 at 5800 yrs ybp is not representative of the entire truth because of sampling bias.

Sampling bias itself is well recognized and even mentioned in at least one genetics paper that I have where it was pointed out that Pakistanis (presumably sampled because they were living in the West) were over represented in "South Asia" genetic studies. This is still going on and in a paper I read yesterday - may have been the Lucotte study there were 300 Pakistanis and 400 Indians. Now how can that be representative of "South Asia" - they call it "Pakistano-Indian". This is laughable and specious. India has 6 times Pakistan's population and 4 times the area but gene researchers take whoever is convenient - often people of subcontinental origin living outside India - such as Gujaratis in the US as representative of "Indo-European speakers" of India

And again in the last few days - among the papers mentioned on this page they have used a population of Indians living in Malaysia as representative of "India/South Asian" ancestry. With this type of gandmasti going on in science we cannot really expect too much of the exact truth. We have to take results as an approximation of reality.

But I thought the sampling in some papers - particularly the Reich paper and one by Thangaraj, Tamang, Singh etc to be good.

Sampling criticism was mostly raised against 1000 genome project which relied on US Patels and Pakistani's for samples. Claim was that it was side effect of GOI policy at the time. Private genetic testing was in trend around 2009. I tested myself right around when Z93 was discovered and tested positive. So there are more Indian samples available in private domain but still Europeans dominates. Same thing is playing out in ancient DNA. Europe has excellent coverage and rest of the world is now catching up.

However, sampling bias would only impact discovery of mutation and not its age.

[shiv]

I'm saying that the argument that language and archaeology without written artifacts found are totally unlinked is not true.

Language and archaeology without written artefacts can certainly be linked to any geography of one's choice depending on the words. Sanskrit has no word for steppe, but that should not be an issue.

However the point that I am trying to make is the fact that artefacts from graves are being linked to a particular language or family of languages of common origin. That is actually nonsense. What is worse is what linguists have done

They have first taken Sanskrit and linked it to European geography. Having fixed the geography of Europe, they have then dug up artefacts in parts of Europe and have linked them back to Sanskrit/Indo-European. This is fudging of the crudest type.

Unfortunately generations of Indians have swallowed this fudging whole - it amazes me that people will jump up and say "Why Saraswati? Why not Haraxwati?" while they laugh derisively at the idea of "Tejo Mahalaya, not Taj Mahal". Witzels Haraxwati and Tejo Mahalaya should be equally credible

[shiv]

Sampling criticism was mostly raised against 1000 genome project which relied on US Patels and Pakistani's for samples. Claim was that it was side effect of GOI policy at the time. Private genetic testing was in trend around 2009. I tested myself right around when Z93 was discovered and tested positive. So there are more Indian samples available in private domain but still Europeans dominates. Same thing is playing out in ancient DNA. Europe has excellent coverage and rest of the world is now catching up.

However, sampling bias would only impact discovery of mutation and not its age.

I experienced the "GoI and genes" issue peripherally. There was a time about 15 years ago when companies in the US were paying Indian private hospital owner to give them samples of every tumour (plus some normal tissue) recovered at surgery. Representatives of the company would stand in the operating room with liquid N2 containers and pieces were placed within immediately, to be shipped to the US. That was a time when gene sequences were allegedly being patented, with the idea that the company that later uses the sequence would have all commercial rights on what they got. I used to work in a private hospital where the owner was making a few extra bucks by selling genetic material in this way. I was one of the objectors and recall taking this issue up at "higher levels". The law stopping such harvesting kicked in later

But the M417 sample in that Underhill paper is just 4 individuals or of 6 I think. That is a small number. The 5800 ybp date will need reconfirmation using larger sample sizes.

[A_Gupta]

Unfortunately generations of Indians have swallowed this fudging whole - it amazes me that people will jump up and say "Why Saraswati? Why not Haraxwati?" while they laugh derisively at the idea of "Tejo Mahalaya, not Taj Mahal". Witzels Haraxwati and Tejo Mahalaya should be equally credible

The linguists' claim is supposed to be somewhat more credible than Tejo Mahalaya, because the sound shifts from Saraswati to Harauvati (the old Persian cognate) are systematic, i.e., there is a pattern visible in many words, of e.g., s- to h-, and so on.

When we come to Witzel's Haraxwati, however, it goes a little beyond linguistics, if I remember correctly. He requires that two Afghan rivers, the Arghandab and the Helmand switched names. (I may be misremembering).

[gandharva]

Is it possible for you to explain in a bit non techy language why the conclusions of two underhill papers are different?

I'll try. I am writing this from memory so pardon me for any errors.

First thing to understand is nature of the research in population genetics. Genes by themselves are real thing. But they don't provide direct ancestry and chronological information. Over the years researchers have developed different analytics method to infer ancestry information from genes. There has been gradual improvement in accuracy and details derived from genes. Unfortunately reachers are slow to incorporate new learning and we would see published papers long after the method is discredited. So to reconcile findings from different papers, one need to consider limitations of used method.

SNP and STR:

Initial phase started with ability to sample certain mutation locations (SNP) within Y DNA and mtDNA. Study of the mutation was very effective in providing uni-parental ancestry. This is when different haplogroup (like R1a, R1b, R1a1a etc) were identified and showed that all modern humans traces back to single group of parental group in Africa. Over the years more and more sub groups were identified providing additional clarity. Please refer to https://en.wikipedia.org/wiki/Haplogrou ... 1a1a1b1.29 to see how details emerged within R1a clade over the years.

Reachers used mutation variance (STR) to infer coalescing dates for happlogroups. Dates derived from such analysis are now shown to overestimate by ~3 times. Coalescing dates also depended on number of nodes in branch. As more nodes are identified, dates changed. One thing to keep in mind is that variance is function of both effective population size and time depth.

Very early papers for Indian genetics were published using this method and before Z93 and Z283 were discovered. For all practical purpose they are now outdated and shouldn't be used by layman.

Autosomal DNA:

As ability to sample DNA improved, reachers started to analyze segment of DNA for all chromosomes. David Reich's lab came up with methodology to identify admixed component from DNA. They applied it first for India to identify ANI and ASI components. Later they established that Europeans are also admixed population (WHG, EHG, EEF, CHS etc).

Reich's lab then came up with method to identify admixture dates and published http://genetics.med.harvard.edu/reich/R ... a_Date.pdf which we are all familiar with. One limitation of this method is that it only dates recent admixture. So if ANI admixed ANI+ASI population after long time, it would only date recent admixture.

Some papers derive relative chronology of component in different population using a variance. It also requires assumption about effective population size and mutation rates. One need to consider mutation rates used to compare dates from different papers.

Whole Genome:

It is now possible to sample and analyze whole genome. Whole genome analysis is the most accurate method today for modern DNA analysis. Recently published paper http://www.nature.com/ng/journal/vaop/n ... .3559.html from 1000 genome would be most up to date. This paper shows that expansion of Z93 occurred in last 5500 years.

Ancient DNA:
With ability to sample DNA from ancient remains, one can now apply all above methods on ancient DNA. Ancient DNA have added benefit of independent dating and location. IMHO, finding from ancient DNA trumps any from modern DNA.

I hope this helps.

A dumb question. Lucotte paper which dates Z93 and Z280 says this.

"We have refound in our samples the clear distinction initially
established by Pamjav et al. [21] between Indian Z93 populations and
European Z280 populations: all our South Asian populations are Z93,
while almost all our European populations are Z280. Datations show
that the Z93 Pakistano-Indian group is the most ancient (about 15,5 K
years); in Europe, the Eastern populations are the most ancient (about
12,5 K years) and the Northern ones the most recent (about 6,9 Kyears"

So if Z93 goes to 5500 from 15.5Kyears, is there any data on Z280? I mean what's it accurate age compared
to 12.5K years in Lucotte's paper?

[hanumadu]

From Underhill 2014

We caution against ascribing findings from a contemporary
phylogenetic cluster of a single genetic locus to a particular
pre-historic demographic event, population migration, or cultural
transformation. The R1a TMRCA estimates we report have wide
confidence intervals and should be viewed as preliminary; one must
sequence tens of additional R1a samples to high coverage to uncover
additional informative substructure and to bolster the accuracy of the
branch lengths associated with the more terminal portions of the
phylogeny

[ukumar]

A dumb question. Lucotte paper which dates Z93 and Z280 says this.

"We have refound in our samples the clear distinction initially
established by Pamjav et al. [21] between Indian Z93 populations and
European Z280 populations: all our South Asian populations are Z93,
while almost all our European populations are Z280. Datations show
that the Z93 Pakistano-Indian group is the most ancient (about 15,5 K
years); in Europe, the Eastern populations are the most ancient (about
12,5 K years) and the Northern ones the most recent (about 6,9 Kyears"

So if Z93 goes to 5500 from 15.5Kyears, is there any data on Z280? I mean what's it accurate age compared
to 12.5K years in Lucotte's paper?

These methods have margin of error in 100s of years so don't take this literary but z280 should be few centuries younger than z93.

[ukumar]

From Underhill 2014

We caution against ascribing findings from a contemporary
phylogenetic cluster of a single genetic locus to a particular
pre-historic demographic event, population migration, or cultural
transformation. The R1a TMRCA estimates we report have wide
confidence intervals and should be viewed as preliminary; one must
sequence tens of additional R1a samples to high coverage to uncover
additional informative substructure and to bolster the accuracy of the
branch lengths associated with the more terminal portions of the
phylogeny

TMRCA has been work in progress. However whole genome and ancient DNA is allowing better calibration and estimates are getting better. With R1a clade markers found in European ancient DNA, IMO date debate is more or less closed. There may be minor correction in future but not in millennia.

[hanumadu]

The method of using STR to estimate TMRCA seems to be invented by Underhill himself.

Zhivotovsky LA, Underhill PA, Cinnioglu C et al: The effective mutation rate at
Y chromosome short tandem repeats, with application to human population-divergence
time. Am J Hum Genet 2004;

[Prem Kumar]

ukumar:

1) Not only is M417 date important, M17 (R1a1a*) is important as well. Underhill's paper assigns a coalescence date of 14 kya for R1a1a* (xM458) with its highest diversity in Indus Valley! Its presence in Poland is 11.2kya. So, its clear that M17 moved from Indus Valley area to Europe.

2) In the German graves, R1a1a* (M458) was found. This is a very significant finding . This indicates that these individuals were descendants of some folks from the Indus area, either directly or indirectly

3) In response to Shiv, you said that Z93 came into India. Where did you get that from? You also mentioned that Z93* is found in Steppes. Yes, it is. But Underhill's 2014 paper makes it clear that its a descendant, consistent with a founder effect (what I like to call Punjabi Rishi impregnating a lot of Steppe women!)

lower diversities occur in south Siberian paragroup R1a-Z93* (H=0.921), in Jewish R1a-M582 (H=0.844) and in Roma R1a-M780 (H=0.759), consistent with founder effects that are evident in the network patterns for these populations (Supplementary Figure 2)

[vishvak]

2) In the German graves, R1a1a* (M458) was found. This is a very significant finding . This indicates that these individuals were descendants of some folks from the Indus area, either directly or indirectly

Actually given aryan terminology, India should be sending genetic research fellows to find out who killed the aryans in Europe and how. May be ancestors of those peddling AIT (& connecting Sanskrit to Nazism) were the ones doing actual invasion and killng of aryans.

It won't be difficult, or not, to find out what happened to aryans in Europe, and importantly, if there is any genetic connection at all between the current post enlightened lot and aryans in Europe, and what happened before enlightenment, and so on and so forth.

[ukumar]

1) Not only is M417 date important, M17 (R1a1a*) is important as well. Underhill's paper assigns a coalescence date of 14 kya for R1a1a* (xM458) with its highest diversity in Indus Valley! Its presence in Poland is 11.2kya. So, its clear that M17 moved from Indus Valley area to Europe.

2) In the German graves, R1a1a* (M458) was found. This is a very significant finding . This indicates that these individuals were descendants of some folks from the Indus area, either directly or indirectly

3) In response to Shiv, you said that Z93 came into India. Where did you get that from? You also mentioned that Z93* is found in Steppes. Yes, it is. But Underhill's 2014 paper makes it clear that its a descendant, consistent with a founder effect (what I like to call Punjabi Rishi impregnating a lot of Steppe women!)

lower diversities occur in south Siberian paragroup R1a-Z93* (H=0.921), in Jewish R1a-M582 (H=0.844) and in Roma R1a-M780 (H=0.759), consistent with founder effects that are evident in the network patterns for these populations (Supplementary Figure 2)

#1- M17 date is too back in past and I don't know if language and religion can be projected that back. 10k year is long long time. I do believe that people did migrate out of India after last ice age. One thing I would mention is that diversity is function of effective population size and time depth. Higher diversity by itself is not sufficient to prove that mutation first occurred in India.

#2- But even ancestral mutation of M458, z283 and Z93 is found in ancient steppe.

#3- that is my guess based on current situation because only Z93 is found India where as its ancestral and cousin mutations are found in steppe. I personally still not believe that Z93 migrated out of steppe. I think it spread from somewhere in middle. Underhill's paper you sited also puts origination of R1a1a in northern Iran. Also note that paper is using modern DNA and steppe population turn over has occurred several time in past. Ancient DNA of Andronovo to date is predominantly Z93.

[peter]

Is it possible for you to explain in a bit non techy language why the conclusions of two underhill papers are different?

I'll try. I am writing this from memory so pardon me for any errors.

First thing to understand is nature of the research in population genetics. Genes by themselves are real thing. But they don't provide direct ancestry and chronological information. Over the years researchers have developed different analytics method to infer ancestry information from genes. There has been gradual improvement in accuracy and details derived from genes. Unfortunately reachers are slow to incorporate new learning and we would see published papers long after the method is discredited. So to reconcile findings from different papers, one need to consider limitations of used method.

SNP and STR:

Initial phase started with ability to sample certain mutation locations (SNP) within Y DNA and mtDNA. Study of the mutation was very effective in providing uni-parental ancestry. This is when different haplogroup (like R1a, R1b, R1a1a etc) were identified and showed that all modern humans traces back to single group of parental group in Africa. Over the years more and more sub groups were identified providing additional clarity. Please refer to https://en.wikipedia.org/wiki/Haplogrou ... 1a1a1b1.29 to see how details emerged within R1a clade over the years.

Reachers used mutation variance (STR) to infer coalescing dates for happlogroups. Dates derived from such analysis are now shown to overestimate by ~3 times. Coalescing dates also depended on number of nodes in branch. As more nodes are identified, dates changed. One thing to keep in mind is that variance is function of both effective population size and time depth.

Very early papers for Indian genetics were published using this method and before Z93 and Z283 were discovered. For all practical purpose they are now outdated and shouldn't be used by layman.

Autosomal DNA:

As ability to sample DNA improved, reachers started to analyze segment of DNA for all chromosomes. David Reich's lab came up with methodology to identify admixed component from DNA. They applied it first for India to identify ANI and ASI components. Later they established that Europeans are also admixed population (WHG, EHG, EEF, CHS etc).

Reich's lab then came up with method to identify admixture dates and published http://genetics.med.harvard.edu/reich/R ... a_Date.pdf which we are all familiar with. One limitation of this method is that it only dates recent admixture. So if ANI admixed ANI+ASI population after long time, it would only date recent admixture.

Some papers derive relative chronology of component in different population using a variance. It also requires assumption about effective population size and mutation rates. One need to consider mutation rates used to compare dates from different papers.

Whole Genome:

It is now possible to sample and analyze whole genome. Whole genome analysis is the most accurate method today for modern DNA analysis. Recently published paper http://www.nature.com/ng/journal/vaop/n ... .3559.html from 1000 genome would be most up to date. This paper shows that expansion of Z93 occurred in last 5500 years.

Ancient DNA:
With ability to sample DNA from ancient remains, one can now apply all above methods on ancient DNA. Ancient DNA have added benefit of independent dating and location. IMHO, finding from ancient DNA trumps any from modern DNA.

I hope this helps.

It does. But still not clear and the fault is mine.
Say Z93. Two individuals A and B:
AB
00 (none have it)
01 (B has it)
10 (A has it)
11 (both have it)

How does this determine who is older? Can you give an example with another gene and Z93 to help clarify the example. In other words if we use two genes can we determine who came first? How?

[shiv]

Unfortunately generations of Indians have swallowed this fudging whole - it amazes me that people will jump up and say "Why Saraswati? Why not Haraxwati?" while they laugh derisively at the idea of "Tejo Mahalaya, not Taj Mahal". Witzels Haraxwati and Tejo Mahalaya should be equally credible

The linguists' claim is supposed to be somewhat more credible than Tejo Mahalaya, because the sound shifts from Saraswati to Harauvati (the old Persian cognate) are systematic, i.e., there is a pattern visible in many words, of e.g., s- to h-, and so on.

When we come to Witzel's Haraxwati, however, it goes a little beyond linguistics, if I remember correctly. He requires that two Afghan rivers, the Arghandab and the Helmand switched names. (I may be misremembering).

Sound shifts in my view are simply cooked up linguistic rules to create some theory on the spot. Sound shift theories about Sanskrit have been fairly dumb - attributing to Sanskrit "older sounds' and "newer sounds" and in between sounds with no consistency to a level where even linguists disagree.

Having said that here is a simple question.

If Rig Veda is older than Avestan ( a cooked up non existent language - but that is a different issue) the sound has changed from sa to ha and moved from India to Greece. If cooked up Avestan is older, then the sound shift may have moved from somewhere else to Iran and Greece stating with "ha" and changing to "sa". Both Greek and cooked up Avestan have the same ha-sa sound change

Why is Avestan cooked up? Because the entire language has been imagined by linguists using Sanskrit texts because all that remains of the Parsi language is Sanskrit translations of a Middle Persian text by one Neriosangh apart from what Parsi Priest in Gujarat told one fellow called Anquetil du Perron

The linguistic "proof" of sound change is at best only a hypothesis. And the only old language that exists in abundant detail is sanskrit, apart from a few scraps found in Syria. Avestan is made up from Sanskrit records that are over 2000 years after the AIT dated for Rig veda. What is the proof that there were no "sound changes" in 2000 years?

Rest of the world has zero evidence of any old language or comparable antiquity to Rig veda. Everything is cooked up by linguists. The lack of data has been filled up by a bunch of buggers who simply could not accept the fact that a language might possibly moved from heathen to holy lands

[shiv]

#1- M17 date is too back in past and I don't know if language and religion can be projected that back.

No disrespect meant, but that statement is based on mental conditioning only. Even genetics papers have authors who are mentally conditioned to accept existing racist language spread claims in existing textx based on 19th century theories. But few people including you seem to realize that there are astronomical dates in ancient texts that are very credible and consistent. Forum member Nilesh Oak has written a book on the issue.

So yes, language and religion have both been projected back to as far back as 8000 years ago or more.

[A_Gupta]

It does. But still not clear and the fault is mine.
Say Z93. Two individuals A and B:
AB
00 (none have it)
01 (B has it)
10 (A has it)
11 (both have it)

How does this determine who is older? Can you give an example with another gene and Z93 to help clarify the example. In other words if we use two genes can we determine who came first? How?

The very simplified mental model I have is this: suppose each man hands down a copy of a book to his son, but the copy of the book has a small rate of random errors and no error-correction or proof-reading, so that when the son gives the grandson a copy, the errors in the book he received from his father are copied, of course, possibly with more copy errors. Moreover the book is so long and the rate of error is such that the probability of the same error being created twice are small.

Now, you have a collection of such books from a set of contemporary men. Can you say anything about how the men are related, and how?

One is, we can group men by how many errors they have in common in their books. The more errors they have in common, the closer they are related. Of course, we have a few thousand samples from a very rich family tree, so it is a sparse family tree that we can construct.

Two - let's say an event where the 30th word in the book was misspelled happened long ago. In books that have that error, there will likely be a variety of other errors, accumulated since that time, and they will differ quite a bit. On the other hand, the set of books that have a recent error in copying will be much more similar to each other. Thus we can put the errors in a relative time line.

The principle does not change even if we do not know what the original book was.

Another possibility is we notice that there is a particular feature in the errors, repeated words, e.g., "and and", "and and and". Now, copy errors that happened long ago are more likely to show repeats in some or other version of the book available today, while recent errors will not have had a lot of time to have repeat errors. We can then put when errors originally occurred into a time line.

If you think about it, all of the above relies on having a sufficiently representative sample of books.

[A_Gupta]

Rest of the world has zero evidence of any old language or comparable antiquity to Rig veda.

The decipherment of ancient cuneiform texts should then be considered bogus?

[peter]

It does. But still not clear and the fault is mine.
Say Z93. Two individuals A and B:
AB
00 (none have it)
01 (B has it)
10 (A has it)
11 (both have it)

How does this determine who is older? Can you give an example with another gene and Z93 to help clarify the example. In other words if we use two genes can we determine who came first? How?

The very simplified mental model I have is this: suppose each man hands down a copy of a book to his son, but the copy of the book has a small rate of random errors and no error-correction or proof-reading, so that when the son gives the grandson a copy, the errors in the book he received from his father are copied, of course, possibly with more copy errors. Moreover the book is so long and the rate of error is such that the probability of the same error being created twice are small.

Now, you have a collection of such books from a set of contemporary men. Can you say anything about how the men are related, and how?

One is, we can group men by how many errors they have in common in their books. The more errors they have in common, the closer they are related. Of course, we have a few thousand samples from a very rich family tree, so it is a sparse family tree that we can construct.

Two - let's say an event where the 30th word in the book was misspelled happened long ago. In books that have that error, there will likely be a variety of other errors, accumulated since that time, and they will differ quite a bit. On the other hand, the set of books that have a recent error in copying will be much more similar to each other. Thus we can put the errors in a relative time line.

The principle does not change even if we do not know what the original book was.

Another possibility is we notice that there is a particular feature in the errors, repeated words, e.g., "and and", "and and and". Now, copy errors that happened long ago are more likely to show repeats in some or other version of the book available today, while recent errors will not have had a lot of time to have repeat errors. We can then put when errors originally occurred into a time line.

If you think about it, all of the above relies on having a sufficiently representative sample of books.

Thanks the mist is clearing. But not out of the woods yet.
If you look at this picture:
how is it certain that m417 is above z282 and z93?
do you know what logic was used to arrive at this conclusion?
This picture is from underhill paper and based on my poor understanding from ukumar's explanation that this picture did not include the full genome data.

Since the tree shows 417 is one node up from z282 and z93 one possibility is everyone who has z282 or z93 has m417 but not all m417 have z93 or z282?

But in the full genome there would certainly be more such genes showing similar sought of relationships. Have we gone through them? How does one go through this data? And where does such data exist?

Furthermore What if there is a natural process by which mutation m417 reverses itself? In such a scenario how would one distinguish m417 as parent of z93?
In otherwords is it possible someone have m515 and z93 and not m417?

[shiv]

Rest of the world has zero evidence of any old language or comparable antiquity to Rig veda.

The decipherment of ancient cuneiform texts should then be considered bogus?

I have serious doubts about the Hittite language. A couple of years ago I tried to get to the original texts of the decipherment of the cuneiform tablets that were allegedly in Hittite. The information is not available anywhere as far as I can tell. Would be happy if someone can get me the info because it sounds fake and contrived.

Everything is fake and contrived. The "decipherment" and declaration that it is an Indo-Europen language. The creation of grammar and yet an admission that phonetic values are unavailable because the text is "syllabic" in nature.

I have also been unable find out how the dates for that imagined language were attributed - the dates were arrived at be "orthography and paleography"

Akkadian - the language from which cuneiform was translated because of biligual texts with Old Persian itself has dates which I have not been able to confirm. What method was used for dating? When I look into this question it is not all that clear. It is mostly by comparison of various texts and possible dates. These dates are not accepted by all, and are to my mind no more or less accurate than the dates in orally transmitted ancient Indian texts. If the Mahabhatata war is stated to have occurred in 3000 BC in the Aihole inscription. That would make Sanskrit far far older than Hittite or Akkadian. The dating methods for the latter seem as fake and contrived as the phonetics and translations.

Because these texts are ancient and extinct languages and their timelines have been fitted conveniently into the type of history that Europe likes to see there is no questioning of the original sources and methods. Those questions can only be asked by people who have some reason to doubt the genuineness of "Late 19th and early 20th century "research"". For European history there are no doubts. All is well because the "ancient structures" that support European history have all been constructed and are in place with Akkadian and Hititte all translated and deciphered and done

From what I can see, racism an Eurocentricity and "Orientalism" have all played a role in the linguistics of the "near East" ("near" Europe that is) and nothing can be taken as is without validation. All the dates seem to have been made up to fit other parts of a grand story - and I know other parts have been cooked up. It is a mess. The short answer is that I do not trust the decipherment of those tests one bit. I need to see the original arguments made and original texts - some of them were over a century ago. Without revalidation - blind acceptance only suggests GIGO to me.

[shiv]

The Earliest Contributions to the Decipherment of Sumerian and Akkadian
http://cdli.ucla.edu/files/publications ... 11_001.pdf