The Basis For Assumed Relationships Between Versions of Tam Lin
The relationships between versions of Tam Lin presented on this website is based upon the analysis of discrete data by phylogenetic software in the phylip package. The data set is the table created to track the variations in the storyline, with weights given to events that are judged to be of more significance in determining the lineage of a particular version. The specific software used was Mix, using the wagner method of analysis, and assuming that ancestral states are unknown.
For those not familiar with phylogenetic analysis, it is a method of determining the relationship between items based on their degree of similarity, most often by genetic sequence analysis. It is most commonly used to determine evolutionary relationship between organisms, and most of the software is designed with this specific purpose in most. If, for example, you have ten organisms from which you have extracted genetic material, the analysis of common stretches of DNA (those sections known to code for the same protein or occupy analogous chromosomal locations) may help you to determine which organisms are most closely related. Organisms with a recent common ancestor will have fewer differences between their genetic sequences than those whose common ancestor is more distant. Discrete characteristics (those whose value can be either 1 or 0, or in more common terms, can be answered yes or no) can also be analyzed in this manner, which is what I have done to generate the relationships portrayed on this website.
However, the analysis is far from accurate. Folklore is a very different beast from, say, a beast. The lineage of organisms tends to be, well, linear. Folklore, on the other hand, rarely has such a straightforward history as living organisms. A performer may hear two very different versions of the same tale, and the version they perform as a result may be a mix of the two. In biological terms, this would be like inheriting a genetic trait from your second cousin, not a possibility this software was designed to handle. Likewise, verses from other stories may be introduced into a folktale, which in biological terms is like a creature having a horse for a mother and a bear for a father, another biological impossibility, and therefore not something the software was designed to spot. Furthermore, the software I have used to determine these relationships assumes that all individuals included in the analysis are contemporaries, while some of the versions of Tam Lin I have on this site are doubtlessly descended from other versions. These relationships are not reflected in the order of the listing of the versions. The software also makes basic assumptions about the nature of mutation that may be not be applicable to folklore, but are too basic to the nature of phlyogenetic analysis for me to easily work around them. I am not sufficiently knowledgeable about folklore to know if what I am doing makes sense, nor a I sufficiently knowledgeable about phylogeny to know how to do it correctly.
For example, when I tried to use the analysis program dollop to generate an 'ancestral' version of Tam Lin, the software concluded that over sixty-seven percent of the events possible according to the storyline table had occurred. Many of those events are mutually exclusive, and no recorded version of which I am aware has more than about fifty-seven percent of the events occuring. In this case, the flaws of the attempt to fit the software to an 'off-label' use was flawed. In the case of the software I have chosen to use, the flaws are not as obvious. I am fully aware that this method is far from perfect, but it seemed at least somewhat less arbitrary than merely listing the versions in the order in which I had found them, a method which is of no significance whatsoever.
The method I have used to generate these relationships may not be the best method, but it has provided the clearest results of the methods I have tried. When I originally analysed the table of date with the entries unweighted, the software generated eight possible tables. When I weighted the options (in effect, to tell the software to consider some of the variations as more important than others) the software generated two tables. I consulted the original eight tables in choosing between these two tables, and that is how I selected the one I used. As I've added new versions, I've played with the software a little more. I usually find that if I'm getting a lot of different tables, then I need to play with the weights a little to get a tighter result.
For anyone interested in the weighting system used in the analysis, here is the weighting used:
30 93 W Weights123 322212111121111222131114121212122111142233133213321311121111321112211131111121122121133211111 vechild39a 110011000111011111000001010101111001001110111000110000111010110101011010101000010011110000000 vechild39b 110011000111011111000001010101110001001110111000110000111010110001011010101000001010110100000 vechild39c 000010000101011111000000000000000000000000000001000000000010110001000000001000000000000000000 vechild39D 101000000100111111110100000000000000000000011000110000100110111101001011011010100000000100000 vechild39e 000000000100111111000000000000000000000000000000100000000100000001000010001001000000000100000 vechild39f 000000000101010000000001000001000000010001110000010000000000000001000010001000000010000000000 vechild39g 111000000001011111111011100011100001011011110110110111000110110101001011001010100101000100011 vechild39h 101000000100111000000000000000000000000000010100000000000000000001000010001000000000000001000 vechild39i 110111000111011111100001011101111001001110111001111100011110110101011000001010001010110110000 cechild39j 001000000001011100100000000000000000000000001000110000101100110111011011001010100000000000010 vechild39k 001001000001011111110000010101100000011001110100110101000010110101000010011010000000000000100 childnotes 100111000111011111110001110111000011101100100000100010100111110001000000110110000000100000010 vbhbronson 000000000000111110000100000000000000000000000000110100000000000000100010001000000000000100000 versijholm 110111000111011111100001011101111001001110111000111100011111110101011010001110001010101110000 vecgladish 111000000101011111100101110101111001011001011000100000110110110101011010001110010010110110000 versTheSCA 110011000101011110000001000001100001001000001000100000101000110101000010101000000011001000000 stleyespan 111000000101001110110100000000000000000000000000100000110100110101000011001010000010011010000 convention 110001000101011110000001000001110001001000001000100000101000110101000010001000000011001000000 vrjStewart 110011000101011111000000010101100000001000011000100000101010110101000000101000010010010100000 vWatersons 111010000101011111110000010101000000011001111000100000111000111001000010101110000010010100000 farmstrong 001000000100111111111000010101000000011001111000100000111000000001000010001110000010010000000 vkesselman 110011000101011100000101000001110001001000000000000000101000110101000000101000000011001000000 grigduncan 001000000100111111111000010101110101011001111000110000101110110101001111010010000000000100001 oxfordbook 111011000101011111110001010101111001011001111000110000111110110101011011001010000010110110000 davearthur 000000000000111111110100000000000000000000010000100100000100000001001010001010000010110000000 annebriggs 001000000100110111110000010101000000011001011000100000111000000001000010101110000000010000000 verricklee 001000000101000000000000100001111001001001111000100000100010110101000000001000000010110000000 petemorton 111010000101000111110000010101000000011001111000100000111000111001000010001110000010010100000 mjovanovic 110111111111011111110000001111110010101110111001110000111010110101011010101010001010110110000 veThemovie 100000000000000000000000000000000000000010110000000000000010100100000000000000000000110000000
Resulting Trees
and here is the resulting phylogenetic tree:
((((oxfordbook,(vecgladish,(verricklee,(((annebriggs,farmstrong),(petemorton, vWatersons)),(grigduncan,(vechild39k,vechild39g)))))),(vrjStewart, ((vkesselman,(convention,versTheSCA)),(childnotes,((((vechild39h,(veThemovie, vechild39f)),((davearthur,vbhbronson),vechild39e)),(stleyespan,(cechild39j, vechild39D))),vechild39c))))),((mjovanovic,(versijholm,vechild39i)), vechild39b)),vechild39a)[0.5000];Or to present it in the more conventional format:
+--------------------------oxfordbook
!
+-------------------------------------------23 +-----------------------vecgladish
! ! !
! ! ! +--------------------verricklee
! +-14 !
! ! ! +--annebriggs
! ! ! +----25
! +-26 ! +--farmstrong
! ! +-------20
! ! ! ! +--petemorton
! ! ! +----27
! +-19 +--vWatersons
+-----------6 !
! ! ! +-----grigduncan
! ! +----------22
! ! ! +--vechild39k
! ! +-10
! ! +--vechild39g
! !
! ! +-----------------------------------------vrjStewart
! ! !
! ! ! +-----vkesselman
! ! ! +-------------------------------21
! +----------------------------18 ! ! +--convention
! ! ! +-17
! ! ! +--versTheSCA
! ! !
! ! ! +-----------------------------childnotes
! +-15 !
! ! ! +-----vechild39h
! ! ! +--------7
! ! ! ! ! +--veThemovie
+--2 ! ! ! +-29
! ! ! ! +--------5 +--vechild39f
! ! +-------11 ! !
! ! ! ! ! +--davearthur
! ! ! ! ! +-24
! ! ! +--4 +-------12 +--vbhbronson
! ! ! ! ! !
! ! ! ! ! +-----vechild39e
! ! ! ! !
! ! ! ! ! +-----stleyespan
! ! +--3 +----------------16
! ! ! ! +--cechild39j
--1 ! ! +--9
! ! ! +--vechild39D
! ! !
! ! +--------------------------vechild39c
! !
! ! +-----mjovanovic
! ! +-28
! ! ! ! +--versijholm
! +--------------------------------------------------------------------------8 +-13
! ! +--vechild39i
! !
! +--------vechild39b
!
+--------------------------------------------------------------------------------------vechild39a
For further information on the software used, please see the documentation available on the phylip website. Phylip is the product of Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package) version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle. Any errors produced while using this software package are mine, believe me.
Notes
Added to site July 2003