Phylogenetics of Medieval Manuscripts Christopher Howe Dept of Biochemistry and Corpus Christi...

Phylogenetics of Medieval Manuscripts

Christopher Howe

Dept of Biochemistryand Corpus Christi College

University of Cambridge

Russell Gray & Quentin Atkinson (2003) “Language-tree divergence times support the Anatolian theory of Indo-European origin” Nature 426: 435-439

Phylogenetic analysis can be applied to many things other than sequence data

Hengwrt Canterbury Tales, National Library of Wales, (MS Penarth 392D)

Application of molecular genetic techniques to the study of manuscripts

• Use of computer programs from phylogenetics to study copying history. Which extant versions of the manuscript were copied from the same earlier version? (‘Stemmatology’ - Lachmann 1830s).

• Study of the manuscript itself

Stemmatology might be used:

• to understand more about the history of the manuscripts themselves - those copied from the same version may have been copied in the same place.

• to understand more about the history of particular variants - how they came to be introduced and why?

• to get a better idea of the original text.

Jean Mielot, scribe and translator to the Duke of Burgundy

(from “Scribes and Illuminators”, C de Hamel, British Museum Press)

(from “Scribes and Illuminators”, C de Hamel, British Museum Press)

Proof-reading - corrections sometimes highlighted to confirm accuracy

Humpty Dumpty sat on a wall

Humpty Dumpty sat on a table

Humpty Dumpty stood by the door

Mr Dumpty stood by the door

Humpty Dumpty lay by the door

A 0 0 0 0 0 0 B 0 0 0 0 0 1 C 0 0 1 1 1 2 D 1 0 2 1 1 2

Manuscripts Transcription

Transcripts

Alignment algorithm

Aligned texts e.g.

A Humpty Dumpty sat on a wall B Humpty Dumpty sat on a table C Humpty Dumpty lay by the door D Mr Dumpty stood by the door

Encoding to produce NEXUS file

A 0 0 0 0 0 0 B 0 0 0 0 0 1 C 0 0 1 1 1 2 D 1 0 2 1 1 2

A B C D A - B 1 - C 4 4 - D 5 5 2 -

Construction of distance matrix

Construction of tree

A B

C D

0.5

3

0.5

1.5 0.5

Humpty Dumpty sat on a wall

Humpty Dumpty lay by the door Mr Dumpty stood

by the door

Humpty Dumpty sat on a table

Canterbury Tales

• Written by Geoffrey Chaucer

• About 80 different manuscript versions

• Prologue to The Wife of Bath’s Tale in 58 extant 15th century MS versions

Wife of Bath, from Chaucer’s Canterbury Tales(GG.4.27(1) University Library, Cambridge

SplitsTree analysis of the Prologue to the

Wife of Bath’s Tale

Nl

Cx1

Ry1

Ds

Bo1

He

Ii

Ln

En3

Tc2

Ph2

Si

Ne

Mg

Pw

Gg

Ry2

Tc1

En1 MaRa3

Ha5

Sl1

Ht

Cn

Fi

Ld1

Lc

Bw

Dd

Cp

Ad1

To

Sl2La

Ld2

Ph3

Mm

Ad3

Dl

Ch

Bo2

HgGroup E

Group O

GroupC/D

Group A

GroupB

Group F

Group O

Group O

Barbrook et al. (1998) Nature 394 839

‘Hengwrt Chaucer’

National Library of Wales (MS Penarth 392D) Around 1400

Comparison of conclusions based on evolutionary methods with those based on conventional approaches:

John Lydgate’s ‘Kings of England Sithen William the Conqueror’ (1426)

105 lines, 35 extant manuscripts & 3 early printed copies

Analysed by occurrence of variants and other information (e.g. TCC601-1 and TCC601-2 are two copies bound in the same MS by the same “Trinity Anthologies” scribe)

4 groups, A, B, C and ‘others’

Ashmole 59 Buryed at Caane thus seythe the Croniculer

Digby 186 Beryed att Cane & thus says the cronyclere

BL Ad 31042 Beryed at caene so seyth the cronyclere

Lansd. 762 Buried at cane this saith the croneclere

de Worde And is buried at Cane as the Cronycle sayes

R. Wyer And buryed at cane as the Cronycle sayes

William Rufus

Kings of England II

(British Library, Harley

4205 f.1v copyright)

• Parsimony tree (weighted)

• Reproduces groups (Bodley 48, 686 were questionable in manual stemma)

• All pairings reproduced

Contamination - scribe used more than one exemplar

• Contamination - scribe used more than one exemplar in making a copy

• Can be successive (changes at a point) or simultaneous/incidental

• Analogous to lateral (horizontal) gene transfer or genetic recombination

With successive contamination, trees constructed from different parts of the text, e.g. first quarter versus last quarter should be different

1st quarter Last quarter

Prologue to the Wife of Bath’s Tale

Hg

El

El

Hg

Relationship of Hengwrt and Ellesmere manuscripts

So Ellesmere and Hengwrt had different exemplars for the first quarter, but shared the same exemplar for the last (Ellesmere may have been copied from Hengwrt).

This is analogous to recombination and lateral gene transfer.

We have a method for mapping recombination in DNA sequences. (Maximum chi-squared method of John Maynard Smith - compares observed distribution of differences between sequences with expected if no recombination occurred.) Can we use this with texts?

X X X X X X X X X X

X X X X X X X X X X

Numbers of differences on each side of arrow

2,8 observed

1,9 expected if evenly distributed

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

10,0 observed


10,0 observed


10,0 observed


6,4 observed


(O-E)2/E

Score

1.1

4.3

1.1

4.3

10.0

Character number

Chi-square

Maximum chi-squared analysis for Wife of Bath’s Prologue indicates breakpoint at character 3384 (line 404) in Hengwrt/Ellesmere comparison. (Manly & Rickert - line 400)

The Mahabharata

One of the major Sanskrit epics of India

Ascribed to the mythological sage Vyasa – ‘the compiler or arranger’

It describes the dynastic struggle between two sets of cousins fighting for the rights to the Bharata kingdom.

Mahabharata – a long poem!

100,000 stanzas

18 Books or Cantos

1200 chapters

Northern - simpler and shorterSouthern - 13,000 stanzas longer

Sanskrit Texts

North-West(Kashmiri)

North-East

South

Sarada (Z)Devanagari version of Sarada (K)

Telugu (T)Grantha (G)Malayalam (M)

Devanagari (D)Bengali (B)Nepali (J)Maithili (V)Devanagari version of vulgate (Dn)

NeighborNet of complete data

Kashmiri

North-East

South

NeighborNet of Section 1

D6 and Dn1 closely related


D6 and Dn1 closely related


D6 and D5 closely related

Manuscripts D6 and D5

Max chi-squared value = 71.0Chr no. (max chi-squared) = 3735

P = 1.5 x 10-8

(significance threshold = 8.0 x 10-4)

Manuscripts D6 and Dn1

Max chi-squared value = 57.8Chr no. (max chi-squared) = 3757

P = 2.7 x 10-6

(significance threshold = 8.0 x 10-4)

Predicted break-point D6/D5 = chr 3735 (end Ch51)

Predicted break-point D6/Dn1 = chr 3757(Start Ch59)

(52-58 not included in analysis)

Start of chapter 59 is a key part of the narrative where the protagonists begin a dicing match, staking their brothers, their wives and eventually their kingdoms.

Phylogenetic methods offer a powerful tool for studying the copying history of manuscripts.

Conclusions are consistent with those reached by conventional scholarship. Results can be obtained much faster.

We can identify situations where scribes used more than one version in making their copy.• Canterbury Tales• Mahabharata

Parchment is, in effect, dried animal skin, often with minimal chemical treatment (unlike tanning).

It is likely to contain DNA from the organism used to make it.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Between species:

• What species was used? Cow, sheep, rabbit, seal?

• Mitochondrial DNA

Within species:

• Manuscript of uncertain origin was written on parchment genetically indistinguishable from a manuscript of known origin

• Mediaeval husbandry

• Nuclear DNA



Mitochondrial DNA analysis of 1763, 1829 & modern parchments

(D-loop, primers1F, 2R)

Siz

e st

and

ards

176

3

182

9

Mod

ern

Bla

nks

Expected size

Successful amplification of mitochondrial DNA





Successful amplification of a nuclear microsatellite

Siz

e st

and

ards

176

3

182

9

Mod

ern

Note size difference

Microsatellite INRA005 allows differentiation between source of modern parchment and 1763, 1829 sources

Composite figure

We can successfully isolate mitochondrial and nuclear DNA from parchment.

Need to develop less (or non-) destructive methods

Offers:

• unambiguous identification of source species

• insights into medieval animal breeding

• identification of close relationships between different parchments

Textual Analysis

Heather Windram, Wendy Phillips-Rodriguez, Adrian Barbrook


Matt Spencer

University of Liverpool

Peter Robinson, Barbara Bordalejo

University of Birmingham

Linne Moooney

University of York

Parchment DNA

Mim Bower, Michael Campana & colleagues


Christopher de Hamel

Corpus Christi College, Cambridge

Funding

Leverhulme Trust

Arts & Humanities Research Council

Phylogenetics of Medieval Manuscripts Christopher Howe Dept of Biochemistry and Corpus Christi...

Documents

Transcript of Phylogenetics of Medieval Manuscripts Christopher Howe Dept of Biochemistry and Corpus Christi...