Dear Susan, I can¹t say that I am proficient enough in statistics to understand your explanation (below), but here are my thoughts: First, we are not talking about a lot of lineages, but about the M222 lineages in areas facing the Atlantic: Ireland, Scotland, North America and Iceland. We are also talking only about whether the generation time was 3/century or 4/century. Since our project is moving fast, this should be a time when people should be encouraged to measure whatever portions of their pedigrees they can dig up and see if we can get a consensus of the most appropriate generation time for M222 people. As a corollary, Brad Knowles has asked the question "Can anyone produce numbers that support 25 years per generation?² You say below "it makes little difference whether or not the generation length is 3 or 4 per century to calculate the TMRCA". I¹m not sure that this is an accurate statement. What you may mean is that 3 vs 4 makes little difference in calculating the TMRCA in GENERATIONs. Here I have pulled out a section from your M222_DS_SNPSummary5.19.14.xls f215509 Ewing,S (prob S603)BIGYS588 f43498Lane, should test YSEQ BIGY S588 f205253 Corbin FTDNA/YSEQ S588 60G about 1500 years ±370 years L & C 48G about 1200 years ±320 years What I think I see is that you equate 60G with 1500 years and 48G with 1200 years ie. 4 generations per century. If the generation time were actually 3/century, then the times in years would be 60G in 2000 years and 48G in 1500 years - a pretty significant difference. David On 2014-05-18, 11:18 AM, "Susan Hedeen" <chantillycarpets@earthlink.net> wrote: >When looking at a population of hundreds of lineages, generations per >century may be a floating concept; some lineages will coalesce around 3 >while others will coalesce around 4 and even 5. > >Each method will incorporate into the method of calculation a generation >length constant; in that regard, however it makes little difference >whether or not the generation length is 3 or 4 per century to calculate >the TMRCA. Most generation length averages will be tied to a mutation >rate constant for calculating purposes and all are subject to a standard >deviation as well as all statistical considerations are. >

Average generation length of the principle tribe of the Uladh, the Dál Fiatach. O'Clery's Book of Genealogies has the ancestry of Ruaidhrí Mac Duinnshléibhe, the last king of the Dál Fiatach who died in 1201, from his ancestor Muiredaigh Muindeirg who died in 489. It is 21 generations, so the average length is about 34 years. If you push it back to Fergus Dubhtach who died in 226, you get 28 generations for an average of 35 years. These names before Saint Patrick are considered sketchy by modern scholars though. -----Original Message----- From: dna-r1b1c7-bounces@rootsweb.com [mailto:dna-r1b1c7-bounces@rootsweb.com] On Behalf Of David Maclennan Sent: Sunday, May 18, 2014 9:16 PM To: Susan Hedeen; dna-r1b1c7@rootsweb.com Subject: Re: [R-M222] Generations per century Dear Susan, I can¹t say that I am proficient enough in statistics to understand your explanation (below), but here are my thoughts: First, we are not talking about a lot of lineages, but about the M222 lineages in areas facing the Atlantic: Ireland, Scotland, North America and Iceland. We are also talking only about whether the generation time was 3/century or 4/century. Since our project is moving fast, this should be a time when people should be encouraged to measure whatever portions of their pedigrees they can dig up and see if we can get a consensus of the most appropriate generation time for M222 people. As a corollary, Brad Knowles has asked the question "Can anyone produce numbers that support 25 years per generation?² You say below "it makes little difference whether or not the generation length is 3 or 4 per century to calculate the TMRCA". I¹m not sure that this is an accurate statement. What you may mean is that 3 vs 4 makes little difference in calculating the TMRCA in GENERATIONs. Here I have pulled out a section from your M222_DS_SNPSummary5.19.14.xls f215509 Ewing,S (prob S603)BIGYS588 f43498Lane, should test YSEQ BIGY S588 f205253 Corbin FTDNA/YSEQ S588 60G about 1500 years ±370 years L & C 48G about 1200 years ±320 years What I think I see is that you equate 60G with 1500 years and 48G with 1200 years ie. 4 generations per century. If the generation time were actually 3/century, then the times in years would be 60G in 2000 years and 48G in 1500 years - a pretty significant difference. David On 2014-05-18, 11:18 AM, "Susan Hedeen" <chantillycarpets@earthlink.net> wrote: >When looking at a population of hundreds of lineages, generations per >century may be a floating concept; some lineages will coalesce around 3 >while others will coalesce around 4 and even 5. > >Each method will incorporate into the method of calculation a >generation length constant; in that regard, however it makes little >difference whether or not the generation length is 3 or 4 per century >to calculate the TMRCA. Most generation length averages will be tied >to a mutation rate constant for calculating purposes and all are >subject to a standard deviation as well as all statistical considerations are. > ------------------------------- To unsubscribe from the list, please send an email to DNA-R1B1C7-request@rootsweb.com with the word 'unsubscribe' without the quotes in the subject and the body of the message

Statistical considerations here are indeed at the heart of your not understanding a statistical model which involves many considerations other than the generation length you have seized on. In this circumstance the model I used for calculating these values uses a "_statistical"_ generation tied to a mutation rate constant. That statistical generation is indeed 25 years. If I wanted to change the generation length I could, and I'd also need to change the mutation rate constant. I have done this and demonstrated it on the L21 forum last year. Troll through the archives and see if you can find it. I prefer, however, to use the model as it was designed. It works, and IMO the most accurate in comparison to the multitude of other models out there. This is why I prefer the model, and since I am doing the work I guess it is my option to choose the model or models I prefer. There are others out there and I don't limit myself. Most of them are far more simple to use than this one. I often will cross check using other models. Those that I choose normally will be within margin...yet despite the fact that this particular model has multiple steps, a didactic approach, isn't a plug and play, I choose this one because the information it produces is trust worthy, consistent, address nuances that others do not, and is reproducible. David, you are welcome to run TMRCA calculations any way you see fit. Pick a model or develop one yourself and work into it all the considerations you believe need be there. Calibrate it against known genealogies as this one has been. Take it then and start using it. Do lineage analyses, sub-clade analyses, etc. as I've been doing with this one and a few assorted other models for the better part of 3 years now. Get proficient. I hope you do not consider it rude, but I'll not engage further in what I suggested I preferred not to engage in when I added a note approaching the difference between using a statistical model to calculate the TMRCA of haplotypes representing many different lineages and your approach to counting generation for 2 in line lineages, one of them being of a fabricated nature as revealed by one of our historical experts, Paul Duffy. Susan On 5/18/2014 9:16 PM, David Maclennan wrote: > Dear Susan, > > I can¹t say that I am proficient enough in statistics to understand your > explanation (below), but here are my thoughts: > First, we are not talking about a lot of lineages, but about the M222 > lineages in areas facing the Atlantic: Ireland, Scotland, North America > and Iceland. We are also talking only about whether the generation time > was 3/century or 4/century. Since our project is moving fast, this should > be a time when people should be encouraged to measure whatever portions of > their pedigrees they can dig up and see if we can get a consensus of the > most appropriate generation time for M222 people. > As a corollary, Brad Knowles has asked the question "Can anyone produce > numbers that support 25 years per generation?² > You say below "it makes little difference whether or not the generation > length is 3 or 4 per century to calculate the TMRCA". I¹m not sure that > this is an accurate statement. What you may mean is that 3 vs 4 makes > little difference in calculating the TMRCA in GENERATIONs. > > Here I have pulled out a section from your M222_DS_SNPSummary5.19.14.xls > > > > > > > > > > > > > f215509 Ewing,S (prob S603)BIGYS588 > > > f43498Lane, should test YSEQ BIGY S588 > > > f205253 Corbin FTDNA/YSEQ S588 > > > > 60G about 1500 years ±370 years > > > > L & C 48G about 1200 years ±320 years > > What I think I see is that you equate 60G with 1500 years and 48G with > 1200 years ie. 4 generations per century. If the generation time were > actually 3/century, then the times in years would be 60G in 2000 years and > 48G in 1500 years - a pretty significant difference. > > David > > > On 2014-05-18, 11:18 AM, "Susan Hedeen" <chantillycarpets@earthlink.net> > wrote: > >> When looking at a population of hundreds of lineages, generations per >> century may be a floating concept; some lineages will coalesce around 3 >> while others will coalesce around 4 and even 5. >> >> Each method will incorporate into the method of calculation a generation >> length constant; in that regard, however it makes little difference >> whether or not the generation length is 3 or 4 per century to calculate >> the TMRCA. Most generation length averages will be tied to a mutation >> rate constant for calculating purposes and all are subject to a standard >> deviation as well as all statistical considerations are. >> >