Saggi T HE ฀ T E X T UAL ฀ T RADI TION฀OF ฀ DANTE’S฀ « C OMME DI A»฀ AND฀ THE฀«BARBI ฀ LOCI» PETER฀ROBINSON The฀textual฀problem฀with฀the฀Commedia฀is฀simply฀stated.฀As฀its฀most฀ recent฀editor,฀Prue฀Shaw,฀puts฀it:฀«The฀first฀problem฀facing฀any฀editor฀ of฀Dante’s฀Commedia฀is฀the฀richness฀of฀the฀manuscript฀record฀–฀around฀ 600฀copies฀if฀we฀count฀only฀complete฀texts฀of฀the฀poem,฀more฀than฀800฀if฀ we฀include฀partial฀and฀fragmentary฀copies».1฀In฀terms฀of฀size:฀no฀editor,฀ no฀team฀of฀editors฀working฀with฀the฀traditional฀methods฀of฀Italian฀philology฀could฀ever฀find฀the฀resources฀to฀look฀at฀every฀reading฀in฀every฀one฀ of฀the฀14233฀lines฀of฀the฀Commedia฀in฀every฀one฀of฀the฀800฀plus฀manuscripts฀and฀significant฀early฀print฀editions฀of฀the฀poem.฀ Nor฀is฀the฀problem฀just฀the฀number฀of฀manuscripts.฀It฀is฀known฀that฀ three฀early฀and฀very฀influential฀rescensions฀of฀the฀whole฀text฀were฀made฀ by฀Giovanni฀Boccaccio฀between฀1355฀and฀1373,฀and฀that฀all฀three฀primary฀Boccaccio฀copies,฀though฀based฀on฀the฀still-extant฀Vatican฀manuscript฀known฀to฀Commedia฀editors฀as฀«Vat»,฀introduced฀many฀readings฀ from฀many฀other฀manuscripts:฀what฀editors฀know฀as฀«contamination».฀ From฀that฀point฀on,฀the฀manuscripts฀show฀such฀a฀mixture฀of฀readings฀ that฀standard฀stemmatic฀processes,฀depending฀on฀the฀orderly฀copying฀of฀ Earlier฀versions฀of฀this฀paper฀were฀read฀at฀meetings฀of฀the฀Studia฀Stemmatalogica฀working฀ group,฀convened฀between฀2010฀and฀2012฀by฀Tuomas฀Heikkila฀of฀the฀University฀of฀Helsinki.฀ I฀am฀grateful฀to฀members฀of฀the฀working฀group฀for฀their฀comments฀on฀he฀methodology฀of฀ the฀paper,฀and฀especially฀to฀Steven฀Steven฀J.฀Schwager฀and฀Teemu฀Roos฀for฀their฀detailed฀ help฀with฀hypergeometric฀distribution.฀I฀am฀grateful฀also฀to฀Prue฀Shaw฀for฀her฀help฀with฀ the฀discussion฀in฀the฀latter฀part฀of฀the฀paper฀of฀the฀influence฀of฀Contini฀on฀Italian฀textual฀ scholarship฀–฀and,฀as฀always,฀for฀the฀privilege฀of฀working฀so฀closely฀and฀so฀long฀with฀her. 1 ฀«Introduction»฀in฀Prue฀Shaw,฀ed.,฀Dante฀Alighieri.฀Commedia.฀A฀Digital฀Edition,฀Birmingham-Florence,฀Scholarly฀Digital฀Editions-SISMEL-Edizioni฀del฀Galuzzo,฀2010. 2 Peter฀Robinson readings฀from฀one฀manuscript฀into฀another,฀appear฀impossible.2฀Even฀if฀ one฀could฀record฀every฀reading฀in฀every฀witness฀there฀is฀common฀agreement฀among฀many฀scholars฀that฀traditional฀stemmatic฀analysis,฀as฀identified฀with฀Lachmann฀and฀the฀followers฀of฀the฀method฀named฀for฀him,฀ cannot฀deal฀with฀cases฀of฀contamination฀and฀hence฀it฀is฀impossible฀to฀ construct฀a฀useful฀genetic฀hypothesis฀concerning฀the฀relations฀among฀ the฀many฀manuscripts฀of฀the฀Commedia.3฀It฀appears,฀then,฀that฀we฀are฀ faced฀with฀an฀impossible฀situation.฀The฀fundamental฀importance฀of฀the฀ Commedia฀to฀Italian฀and฀world฀literature฀makes฀it฀imperative฀that฀Dante’s฀master-work฀be฀edited,฀and฀edited฀to฀the฀highest฀possible฀standard.฀ Yet,฀either฀difficulty฀–฀the฀sheer฀size฀of฀the฀tradition฀or฀the฀prevalence฀of฀ contamination฀–฀would฀on฀its฀own฀make฀it฀impossible฀to฀achieve฀what฀ one฀could฀argue฀are฀the฀two฀minimal฀requirements฀of฀a฀scholarly฀edition:฀first,฀that฀it฀examine฀all฀the฀evidence฀in฀all฀the฀witnesses;฀second,฀ that฀it฀derive฀from฀all฀this฀evidence฀an฀understanding฀of฀the฀whole฀tradition฀which฀can฀then฀be฀used฀by฀the฀editor฀to฀identify฀which฀manuscripts฀and฀which฀readings฀within฀them฀are฀most฀likely฀nearest฀to฀Dante’s฀own฀text.4 Over฀the฀last฀centuries฀of฀Dante฀scholarship,฀scholars฀have฀tried฀several฀routes฀past฀this฀impossibility.฀Giorgio฀Petrocchi,฀editor฀of฀the฀most฀ signficant฀modern฀edition฀of฀the฀Commedia,฀addressed฀both฀problems฀ with฀a฀single฀strategy:฀that฀of฀basing฀his฀edition฀of฀the฀Commedia฀«secondo฀l’antica฀vulgata»฀only฀on฀the฀27฀manuscripts฀(twenty-four฀complete,฀three฀fragmentary,฀and฀counting฀Martini’s฀collation฀of฀the฀Aldine฀ edition฀as฀a฀manuscript)฀accepted฀at฀the฀time฀of฀his฀edition฀as฀dating฀ securely฀before฀1355.5฀This฀approach฀both฀reduced฀the฀number฀of฀manuscripts฀which฀he฀needed฀to฀examine฀to฀a฀manageable฀number,฀and฀also฀ greatly฀reduced฀the฀problem฀of฀contaminated฀manuscripts฀by฀looking฀ only฀at฀manuscripts฀written฀before฀Boccaccio฀started฀work฀on฀his฀recen฀Indeed,฀manuscripts฀datable฀from฀before฀1335฀already฀show฀clear฀signs฀of฀contamination:฀that฀is,฀the฀import฀of฀readings฀from฀manuscripts฀other฀than฀the฀exemplar.฀The฀ Landino฀manuscript฀(La),฀dated฀to฀1336,฀has฀readings฀scraped฀away฀and฀others฀substituted฀across฀its฀whole฀length.฀The฀even฀earlier฀manuscript,฀not฀now฀extant,฀dated฀to฀13301331฀and฀used฀by฀Luca฀Martini฀in฀his฀collation฀of฀the฀1515฀Aldine฀text,฀also฀included฀ readings฀from฀manuscripts฀other฀than฀the฀exemplar฀(Shaw,฀«Introduction:฀Overview»). 3 ฀Among฀Middle฀English฀textual฀scholars,฀the฀view฀that฀no฀useful฀genetic฀hypotheses฀ can฀be฀created฀for฀large฀manuscript฀traditions฀has฀become฀so฀widely฀accepted฀as฀to฀count฀ as฀gospel:฀see฀George฀Kane’s฀editions฀of฀Piers฀Plowman฀and฀his฀assault฀on฀John฀Manly฀ and฀Edith฀Rickert฀(who฀did฀think฀they฀could฀disentangle฀the฀80-plus฀manuscripts฀of฀the฀ Canterbury฀Tales)฀in฀Paul฀Ruggiers,฀ed.,฀Editing฀Chaucer:฀the฀Great฀Tradition,฀Norman,฀ Oklahoma,฀Pilgrim฀Books,฀1984. 2 The฀textual฀tradition฀of฀Dante’s฀Commedia 3 sions฀of฀the฀Commedia.฀Petrocchi฀stated฀very฀clearly฀that฀no฀edition฀based฀ on฀such฀a฀small฀selection฀of฀manuscripts฀could฀claim฀to฀be฀«un’edizione฀ critica»,฀which฀must฀be฀based฀on฀all฀the฀manuscript฀evidence,฀not฀just฀ some฀of฀it.6฀Accordingly,฀he฀claimed฀only฀the฀more฀modest฀title฀of฀«La฀ Commedia฀secondo฀l’antica฀vulgata»฀for฀his฀edition.฀Nonetheless,฀he฀also฀ argued฀that฀even฀though฀his฀text฀was฀based฀only฀on฀this฀small฀fraction฀of฀ the฀manuscripts,฀all฀the฀later฀manuscripts฀would฀not฀contribute฀a฀single฀ reading฀not฀already฀present฀in฀or฀readily฀deducible฀from฀the฀antica฀vulgata฀manuscripts฀to฀a฀critical฀text.฀Therefore,฀he฀declared,฀a฀sound฀editorial฀text฀based฀only฀on฀the฀antica฀vulgata฀manuscripts฀would฀be฀identical฀ to฀that฀based฀on฀the฀whole฀tradition,฀and฀so฀his฀text฀could฀stand฀for฀the฀ whole฀tradition,฀despite฀the฀provisional฀status฀implied฀by฀the฀«secondo฀ l’antica฀vulgata»฀qualification.฀ Petrocchi’s฀approach฀evades,฀rather฀than฀satisifies,฀the฀need฀to฀form฀ a฀view฀of฀the฀entire฀manuscript฀tradition.฀Indeed,฀his฀assertion฀that฀a฀ sound฀edition฀could฀be฀based฀on฀these฀27฀manuscripts฀alone฀could฀only฀ be฀validated฀by฀looking฀at฀all฀the฀evidence฀in฀all฀the฀manuscripts.฀But฀ how฀can฀this฀be฀done,฀given฀over฀eight฀hundred฀manuscripts,฀more฀than฀ 14000฀lines฀of฀text฀and฀100,000฀words฀in฀a฀complete฀manuscript?฀Before฀ Petrocchi,฀the฀Italian฀Dante฀scholar฀Michele฀Barbi,฀then฀at฀the฀beginning฀of฀a฀remarkable฀fifty-year฀philological฀career,฀and฀at฀the฀instigation฀฀ of฀three฀senior฀Dante฀scholars฀associated฀with฀the฀newly-founded฀Soci฀This฀pessimistic฀view,฀that฀no฀genealogical฀representation฀of฀the฀relationships฀ among฀the฀manuscripts฀of฀the฀Commedia฀is฀possible,฀was฀reached฀by฀the฀nineteenthcentury฀English฀scholar฀Edward฀Moore,฀who฀examined฀hundreds฀of฀manuscripts฀of฀the฀ Commedia฀only฀to฀conclude฀«My฀own฀belief฀would฀be฀that฀owing฀to฀the฀complicated฀ intermixture฀of฀texts,฀such฀a฀genealogy฀never฀can฀be฀constructed.»฀(Edward฀Moore,฀ Contributions฀to฀the฀textual฀criticism฀of฀the฀Divina฀Commedia,฀including฀the฀complete฀collation฀throughout฀the฀Inferno฀of฀all฀the฀mss.฀at฀Oxford฀and฀Cambridge,฀Cambridge,฀Cambridge฀University฀Press,฀1889,฀p.฀xxxi).฀Shaw฀cites฀the฀German฀scholar฀Witte฀as฀implicitly฀reaching฀the฀same฀conclusion฀in฀the฀Prolegomena฀to฀his฀1862฀edition฀of฀the฀poem,฀ based฀on฀his฀own฀independent฀examination฀and฀collation฀of฀hundreds฀of฀manuscripts฀ («Introduction»,฀n.฀22). 5 ฀Dante฀Alighieri,฀La฀Commedia฀secondo฀l’antica฀vulgata,฀a฀cura฀di฀Giorgio฀Petrocchi,฀ in฀Le฀opere฀di฀Dante฀Alighieri.฀Edizione฀Nazionale฀a฀cura฀della฀Società฀Dantesca฀Italiana,฀ VII,฀Milano,฀1966-1967,฀4฀vols.฀Note฀that฀in฀this฀article฀all฀references฀to฀Mart,฀in฀terms฀of฀ readings฀found฀and฀analysis฀based฀on฀those฀readings,฀are฀to฀the฀Martini฀collation฀except฀ where฀otherwise฀specified. 6 ฀«Ai฀fini฀di฀un’edizione฀critica,฀e฀perché฀tale฀veramente฀sia,฀si฀deve฀tuttora฀tener฀fede฀ alla฀legge฀di฀partenza฀e฀di฀fondo฀d’ogni฀ricerca฀testuale:฀l’interrogazione฀integrale฀della฀tradizione.»฀Giorgio฀Petrocchi,฀«Proposte฀per฀un฀testo-base฀della฀Divina฀Commedia»,฀Filologia฀romanza,฀II฀(1955),฀pp.฀337-365฀[p.฀343],฀cited฀by฀Shaw,฀«Introduction:฀Overview». 4 4 Peter฀Robinson età฀Dantesca฀Italiana,฀produced฀a฀list฀of฀lines฀in฀the฀Commedia฀which฀฀ he฀judged฀critical฀for฀the฀establishment฀of฀manuscript฀relations฀across฀the฀฀ whole฀text฀of฀the฀Commedia฀and฀all฀its฀manuscripts.฀These฀are฀known฀ as฀the฀«400฀loci»:฀in฀fact,฀there฀are฀396฀of฀them.7฀Barbi฀explains฀in฀a฀later฀ article฀that฀these฀loci฀were฀chosen฀on฀the฀basis฀of฀considerable฀experience฀ working฀with฀Commedia฀manuscripts฀in฀the฀Florentine฀libraries,฀and฀on฀ consideration฀of฀the฀significance฀of฀particular฀variant฀readings.8฀As฀well฀ as฀reducing฀the฀amount฀of฀effort฀required฀to฀survey฀the฀whole฀tradition฀(one฀need฀look฀at฀only฀396฀lines,฀not฀at฀14233),฀this฀also฀offered฀the฀ promise฀of฀a฀collaborative฀approach:฀the฀work฀could฀be฀divided฀among฀ many฀scholars.฀Accordingly,฀special฀forms฀were฀printed฀off,฀to฀be฀distributed฀to฀scholars฀and฀so฀enable฀a฀complete฀survey฀of฀these฀key฀lines฀in฀ every฀manuscript฀everywhere.฀As฀Shaw฀relates,฀the฀response฀was฀disappointing:฀some฀eleven฀scholars,฀including฀Barbi฀himself,฀examined฀a฀few฀ manuscripts฀and฀reported฀their฀findings.฀Among฀these฀eleven฀were฀Guiseppe฀Vandelli฀and฀Mario฀Casella:฀yet฀when฀these฀two฀scholars฀came฀to฀ produce฀their฀own฀editions฀of฀the฀Commedia฀in฀1921฀and฀1923฀respectively,฀neither฀used฀the฀collation฀of฀the฀Barbi฀loci฀as฀the฀base฀for฀an฀analysis฀of฀the฀whole฀tradition,฀and฀hence฀their฀edition.฀ As฀we฀have฀seen,฀the฀next฀great฀edition฀after฀Vandelli,฀that฀of฀Petrocchi฀ in฀1965,฀also฀made฀no฀use฀of฀the฀Barbi฀loci,฀choosing฀a฀different฀rationale฀ for฀a฀selection฀of฀manuscripts฀on฀which฀to฀base฀the฀edition.฀Lanza’s฀1995฀ edition฀took฀an฀even฀more฀extreme฀approach฀to฀the฀problem฀of฀how฀to฀ derive฀a฀single฀text฀from฀a฀vast฀number฀of฀manuscripts:฀he฀used฀just฀one฀ manuscript,฀the฀Trivulziano฀1080฀(«Triv»),฀one฀of฀the฀very฀earliest฀surviving฀Florentine฀manuscripts฀and฀long฀famous฀for฀the฀quality฀of฀its฀text฀and฀ the฀beauty฀of฀its฀decoration.฀In฀this฀choice,฀Lanza฀was฀acting฀consciously฀ in฀the฀tradition฀of฀«best-text฀editing»฀associated฀with฀Joseph฀Bédier.฀Following฀his฀work฀on฀Le฀Lai฀de฀l’ombre,฀Bédier฀asserted฀the฀impossibility฀of฀ creating฀a฀useful฀genetic฀hypothesis฀about฀any฀large฀and฀complex฀manuscript฀tradition.9฀In฀his฀view,฀Lachmannian฀stemmatics฀is฀an฀impossibil7 ฀The฀three฀scholars฀were฀Adolfo฀Bartoli,฀Alessandro฀D’Ancona,฀Isidoro฀Del฀Lungo,฀ writing฀in฀«Per฀l’edizione฀critica฀della฀Divina฀Commedia»,฀Bullettino฀della฀Società฀Dantesca฀Italiana,฀5-6฀(1891),฀pp.฀25-27,฀and฀followed฀by฀Barbi’s฀«Canone฀di฀luoghi฀scelti฀per฀ lo฀spoglio฀dei฀mss.฀della฀Divina฀Commedia»฀on฀pages฀28-38.฀All฀the฀Barbi฀loci฀are฀listed฀ in฀Appendix฀A฀of฀Shaw’s฀edition,฀with฀convenient฀hypertext฀links฀to฀the฀edition’s฀collation฀of฀each฀line฀in฀the฀seven฀manuscripts฀and฀two฀editions฀included฀within฀it. 8 ฀«Fu฀frutto฀di฀lunghi฀studi,฀e฀fissata฀quindi฀non฀a฀priori,฀cioè฀a฀caso฀...฀»;฀Barbi,฀«Ancora฀ sul฀testo฀della฀Divina฀Commedia»,฀Studi฀danteschi,฀XVIII฀(1934),฀p.฀56. 9 ฀«La฀tradition฀manuscrite฀du฀Lai฀de฀l’ombre»,฀Romania,฀54฀(1928),฀pp.฀161-296,฀321356;฀reprinted฀Paris,฀E.฀Champion,฀1970.฀ The฀textual฀tradition฀of฀Dante’s฀Commedia 5 ity฀in฀such฀circumstances.฀The฀only฀recourse฀then฀is฀to฀identify฀the฀best฀ single฀witness฀to฀the฀text,฀by฀whatever฀criteria฀are฀available,฀and฀use฀that฀ as฀the฀base฀for฀an฀edition.฀As฀Lanza฀also฀argued,฀such฀an฀edition฀can฀(if฀ rigorously฀faithful฀to฀the฀source)฀present฀a฀text฀which฀actually฀existed฀ at฀a฀single฀historic฀moment,฀unsullied฀by฀editorial฀impulse฀or฀linguistic฀ revision.฀In฀fact,฀Lanza฀did฀permit฀emendations฀from฀other฀manuscripts฀ into฀his฀edition,฀but฀only฀under฀stringent฀circumstances฀and฀from฀a฀very฀ limited฀range฀of฀other฀Florentine฀manuscripts.฀Of฀course,฀this฀approach฀ frees฀the฀editor฀completely฀from฀the฀need฀to฀survey฀the฀entire฀tradition,฀ whether฀by฀using฀the฀Barbi฀loci฀or฀by฀any฀other฀means. Around฀1996,฀then,฀Barbi’s฀loci฀looked฀like฀a฀road-not-taken฀in฀Commedia฀textual฀scholarship:฀an฀interesting฀idea฀but฀never฀implemented.฀ However,฀in฀2001฀a฀new฀edition฀of฀the฀Commedia฀appeared,฀edited฀by฀ Federico฀Sanguineti,฀in฀which฀the฀Barbi฀loci฀moved฀back฀to฀the฀centre฀ stage฀of฀Commedia฀textual฀scholarship.10฀This฀edition฀was฀startling฀for฀ several฀reasons.฀First,฀where฀Petrocchi,฀Lanza฀and฀other฀editors฀agreed฀ that฀it฀was฀impossible฀to฀form฀a฀useful฀genetic฀hypothesis฀of฀the฀entire฀ tradition,฀Sanguineti฀claimed฀that฀it฀was฀possible฀to฀achieve฀this฀for฀the฀ Commedia.฀In฀opposition฀to฀Bédier’s฀gloomy฀pessimism฀about฀stemmatics,฀Sanguineti฀asserted฀that฀Lachmannian฀stemmatics฀could฀be฀practiced฀ rigorously฀and฀usefully฀on฀the฀whole฀Commedia฀tradition.฀Second,฀Sanguineti฀declared฀that฀not฀only฀could฀traditional฀stemmatics฀be฀applied฀to฀ the฀whole฀tradition,฀but฀he฀had฀done฀it.฀He฀had฀looked฀at฀all฀the฀Barbi฀ loci฀in฀every฀one฀of฀the฀800฀manuscripts฀(so฀achieving฀on฀his฀own,฀with฀ virtually฀no฀support,฀what฀scholars฀had฀failed฀to฀achieve฀in฀over฀a฀hundred฀years),฀and฀from฀analysis฀of฀the฀readings฀at฀these฀loci฀he฀had฀created฀ a฀comprehensive฀account฀of฀the฀whole฀tradition,฀and฀isolated฀just฀seven฀ manuscripts฀as฀necessary฀and฀sufficient฀for฀the฀creation฀of฀a฀critical฀text.฀ Sanguineti’s฀identification฀of฀the฀«Sanguineti฀seven»฀was฀the฀starting฀ point฀of฀Shaw’s฀edition.฀It฀seemed฀possible฀to฀produce฀a฀digital฀edition฀ which฀would฀do฀for฀these฀seven฀manuscripts฀what฀Shaw’s฀edition฀of฀Dante’s฀Monarchia฀had฀done฀for฀the฀manuscripts฀of฀that฀work:฀that฀is,฀offer฀ full฀machine-readable฀transcripts฀of฀them,฀create฀an฀extremely฀precise฀ collation฀of฀the฀transcripts,฀and฀then฀use฀various฀methods฀(including฀phylogenetic฀analysis,฀derived฀from฀evolutionary฀biology)฀to฀create฀a฀view฀of฀ the฀relationships฀among฀the฀manuscripts.11฀At฀first,฀Shaw฀(and฀I)฀planned฀ 10 ฀Dantis฀Alagherii฀Comedìa,฀edizione฀critica฀per฀cura฀di฀Federico฀Sanguineti,฀Firenze,฀ Edizioni฀del฀Galluzzo,฀2001. 11 ฀Prue฀Shaw,฀ed.,฀Dante฀Alighieri,฀Monarchia.฀A฀Digital฀Edition,฀Birmingham-Florence,฀ Scholarly฀Digital฀Editions-Società฀Dantesca฀Italiana,฀2006.฀See฀the฀review฀article฀by฀Paolo฀ 6 Peter฀Robinson to฀work฀with฀Sanguineti฀in฀the฀making฀of฀this฀digital฀edition,฀and฀there฀ were฀several฀substantive฀discussions฀towards฀this฀aim.฀However,฀Sanguineti฀left฀the฀partnership฀at฀an฀early฀stage,฀and฀almost฀all฀the฀work฀of฀the฀฀ edition฀proceeded฀without฀his฀involvement.฀In฀turn,฀the฀aim฀of฀the฀edition฀ changed:฀it฀became฀a฀test฀of฀Sanguineti’s฀arguments฀about฀the฀relation-฀ ships฀among฀these฀seven฀manuscripts.฀Shaw฀examines฀Sanguineti’s฀claims฀฀ in฀considerable฀detail,฀and฀there฀is฀no฀need฀here฀to฀do฀more฀than฀summarize฀her฀criticisms฀of฀them,฀both฀in฀terms฀of฀his฀conclusions฀and฀of฀ his฀methodology.฀Briefly:฀it฀appears฀from฀her฀analysis,฀based฀not฀on฀any฀ selection฀of฀variants฀but฀on฀the฀entire฀body฀of฀variation฀in฀the฀seven฀ manuscripts,฀firstly฀that฀Sanguineti’s฀key฀conclusion฀is฀wrong,฀and฀secondly฀that฀the฀methodology฀he฀used฀to฀reach฀this฀conclusion฀was฀flawed.฀ Sanguineti’s฀key฀conclusion฀is฀that฀one฀manuscript,฀and฀one฀manuscript฀ alone฀–฀Vatican฀Library฀ms.฀Urbinate฀latino฀366฀(Urb)฀–฀represents฀a฀pure฀ line฀of฀descent฀from฀Dante’s฀original฀text฀independent฀of฀every฀other฀ manuscript.฀Thus฀his฀stemma฀of฀the฀seven฀manuscripts฀(figure฀1): figure฀1 Sanguineti’s฀stemma฀of฀the฀«Sanguineti฀seven»฀manuscripts฀of฀the฀Commedia.฀ Note฀the฀positioning฀of฀Rb฀as฀sharing฀an฀ancestor฀α฀with฀the฀five฀manuscripts฀ LauSC฀Mart฀Triv฀Ash฀Ham,฀leaving฀«U»฀(Urb)฀as฀independent฀of฀Rb฀and฀all฀ other฀manuscripts. ω α [y] [β] x z a b LauSC฀฀฀฀Mart฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Triv฀฀Ash฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Ham฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Rb฀฀฀฀฀฀฀Urb (L)฀฀฀฀฀฀฀(M)฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀(T)฀฀(A)฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀(H)฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀(R)฀฀฀฀฀฀฀(U) Trovato฀(with฀an฀appendix฀on฀the฀Shaw฀Commedia฀edition)฀in฀«La฀Doppia฀“Monarchia”฀di฀ Prue฀Shaw฀(con฀una฀postilla฀sulla฀Commedia)»,฀Ecdotica,฀7฀(2010),฀pp.฀193-207. The฀textual฀tradition฀of฀Dante’s฀Commedia 7 Here,฀Dante’s฀original฀is฀ω,฀and฀Sanguineti฀argues฀that฀six฀of฀the฀seven฀ descend฀from฀a฀single฀copy฀of ω,฀labelled฀α,฀while฀Urb฀is฀on฀its฀own,฀des-฀ cended฀independently฀from฀ω.฀Accordingly,฀the฀one฀manuscript฀Urb฀ has฀ the฀ same฀ authority฀ as฀ the฀ other฀ six฀ combined.฀Where฀ most฀ editions฀have฀centred฀their฀text฀around฀the฀Florentine฀manuscripts฀in฀the α branch฀of฀this฀stemma฀(thus,฀Lanza฀used฀Triv฀as฀his฀base,฀Casella฀LauSC;฀ Petrocchi฀often฀favoured฀readings฀in฀Triv฀and฀Mart฀ahead฀of฀those฀in฀ Urb)฀Sanguineti฀based฀his฀entire฀edition฀on฀Urb.฀Sanguineti’s฀argument฀ depends฀on฀Urb฀sharing฀no฀ancestor฀below฀ω฀with฀any฀other฀manuscript,฀and฀so฀contradicts฀Petrocchi,฀who฀argued฀that฀Rb฀and฀Urb฀share฀ an฀ancestor฀(his฀e)฀below฀the฀original฀(figure฀2).฀As฀Shaw฀asserts,฀«the฀ position฀of฀ms.฀Rb฀in฀the฀stemma฀is฀critical฀for฀Sanguineti’s฀argument:฀฀ figure฀2 Part฀of฀Petrocchi’s฀stemma฀of฀the฀antica฀vulgata฀manuscripts฀of฀the฀Commedia.฀ Note฀the฀positioning฀of฀Rb,฀as฀sharing฀an฀ancestor฀with฀Urb฀(and฀Mad). β ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀e ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Rb ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Urb ฀฀฀฀฀฀฀฀฀฀Mad if฀he฀is฀wrong฀about฀that,฀tutto฀crolla฀–฀the฀whole฀edifice฀collapses.»฀Her฀ analysis,฀supported฀by฀phylogenetic฀analysis฀(see฀further฀below),฀is฀decisive:฀Petrocchi฀is฀right,฀Rb฀and฀Urb฀do฀share฀an฀ancestor,฀and฀Sanguineti฀ is฀wrong.฀Further,฀one฀can฀trace฀Sanguineti’s฀error฀directly฀to฀his฀meth- 8 Peter฀Robinson odology.฀Although฀Sanguineti฀claimed฀that฀he฀based฀his฀analysis฀on฀the฀ 396฀Barbi฀loci,฀Shaw฀points฀out฀that฀he฀actually฀uses฀very฀few฀of฀them฀to฀ support฀his฀conclusions.฀Sanguineti฀identifies฀over฀four฀hundred฀manuscripts฀as฀members฀of฀a฀single฀group฀(«la฀cosiddetta฀“tradizione฀β”»)฀on฀ the฀basis฀of฀just฀four฀readings.฀Further,฀one฀of฀these฀readings฀(Par.฀xxiii฀ 103,฀spiro฀for฀giro)฀has฀the฀character฀of฀a฀classic฀polygenetic฀error,฀which฀ could฀readily฀occur฀independently฀in฀unconnected฀scribal฀copies.฀Elsewhere,฀he฀justifies฀discarding฀the฀evidence฀of฀this฀large฀group฀(expanded฀ to฀over฀600฀manuscripts฀on฀the฀basis฀of฀a฀very฀few฀readings)฀because฀it฀is฀ contaminated฀by฀readings฀drawn฀from฀the฀distinct฀traditions฀which฀he฀ sees฀underlining฀the฀seven฀manuscripts:฀thus,฀the฀appearance฀of฀a฀single฀ reading฀from฀Rb฀in฀some฀of฀these฀600฀manuscripts฀(ale฀for฀aer฀at฀Purg.฀ii฀ 35)฀is฀used฀to฀argue฀that฀this฀group฀is฀contaminated฀from฀Rb฀or฀a฀manuscript฀closely฀related฀to฀it. Manifestly,฀arguments฀based฀on฀just฀a฀few฀readings฀in฀a฀text฀of฀over฀ 100,000฀words฀will฀lack฀conviction:฀a฀reason฀why฀Barbi฀suggested฀analysis฀of฀some฀400฀lines฀of฀text,฀not฀just฀twenty฀or฀so.฀Ideally,฀one฀would฀ base฀analysis฀on฀the฀whole฀text,฀on฀every฀variant฀at฀every฀word฀in฀every฀ line.฀To฀do฀this฀for฀all฀800฀manuscripts฀would฀be฀impossible฀with฀the฀ resources฀ available฀ to฀ any฀ current฀ project;฀ but฀ one฀ could฀ do฀ it฀ for฀ a฀ smaller฀number:฀indeed,฀for฀the฀seven฀manuscripts฀identified฀by฀Sanguineti,฀and฀this฀was฀what฀Shaw฀did.฀The฀base฀of฀her฀edition฀was฀a฀complete฀transcription฀of฀the฀seven฀manuscripts,฀and฀creation฀of฀a฀very฀precise฀collation฀of฀every฀word฀of฀every฀one฀of฀these฀seven฀with฀each฀other,฀ and฀ with฀ two฀ major฀ edited฀ texts:฀ those฀ of฀ Petrocchi฀ and฀ Sanguineti.฀ While฀the฀edition’s฀eventual฀aim฀was฀to฀use฀this฀collation฀and฀the฀analysis฀built฀on฀it฀to฀explore฀Sanguineti’s฀hypothesis,฀as฀explained฀in฀the฀ last฀paragraphs,฀this฀same฀collation฀can฀be฀used฀for฀another฀purpose:฀to฀ test฀the฀efficacy฀of฀the฀Barbi฀loci฀for฀their฀declared฀aim฀(as฀accepted฀by฀ Sanguineti),฀as฀a฀base฀for฀an฀analysis฀of฀the฀entire฀tradition.฀Of฀course,฀฀ in฀this฀edition฀we฀have฀only฀seven฀manuscripts.฀However,฀the฀key฀place฀in฀฀ the฀tradition฀of฀at฀least฀six฀of฀these,฀accepted฀by฀every฀scholar฀in฀the฀฀ last฀century,฀makes฀these฀manuscripts฀a฀good฀place฀to฀start.฀Moreover,฀ as฀we฀have฀seen,฀Sanguineti฀based฀his฀identification฀of฀these฀seven฀manuscripts฀as฀critical฀to฀the฀understanding฀of฀the฀whole฀tradition฀of฀the฀ Commedia฀on฀his฀collation฀of฀these฀396฀lines.฀Accordingly,฀one฀should฀ expect฀that฀these฀loci฀should฀be฀pre-eminently฀informative฀in฀the฀analysis฀of฀these฀seven฀manuscripts. Shaw’s฀edition฀proceeded฀as฀follows.฀First,฀an฀exact฀word-by-word฀collation฀of฀the฀whole฀text฀in฀the฀seven฀manuscripts฀was฀created฀as฀follows: The฀textual฀tradition฀of฀Dante’s฀Commedia 9 1.฀All฀seven฀manuscripts฀were฀transcribed฀in฀full,฀word฀by฀word฀and฀ line฀by฀line,฀into฀machine-readable฀form. 2.฀The฀transcription฀recorded฀layers฀of฀scribal฀revision฀in฀each฀manuscript,฀so฀that฀one฀could฀distinguish฀at฀any฀point฀what฀was฀written฀by฀ the฀original฀scribe,฀changed฀by฀him฀or฀her,฀then฀changed฀again฀by฀later฀ scribes.฀This฀is฀particularly฀useful฀with฀respect฀to฀Mart,฀where฀it฀is฀actually฀the฀readings฀recorded฀by฀Luca฀Martini฀from฀a฀now-lost฀manuscript฀ as฀alterations฀to฀the฀Aldine฀edition฀which฀are฀of฀primary฀interest.฀Thus,฀ in฀Purg.฀vii.฀51,฀where฀the฀Aldine฀edition฀prints฀«o฀pur฀sarria»,฀Martini฀ writes฀«o฀non฀saria»฀in฀the฀right฀margin฀(figure฀3),฀both฀readings฀are฀ recorded฀in฀our฀transcription.฀Further,฀the฀layers฀of฀readings฀are฀marked฀ so฀they฀can฀be฀compared฀separately. figure฀3 Martini’s฀annotations฀to฀the฀Aldine฀edition฀at฀Purg.฀vii.฀51. figure฀4 The฀«Literal»฀view฀of฀the฀transcription,฀showing฀both฀the฀original฀text฀and฀Martini’s฀annotations฀as฀marked฀in฀the฀right฀margin. figure฀5 The฀«Aldine฀Original»฀view฀of฀the฀transcript,฀showing฀the฀text฀as฀originally฀ printed. figure฀6 The฀«Martini’s฀Collation»฀view,฀showing฀the฀text฀with฀Martini’s฀annotations฀ applied. 10 Peter฀Robinson 3.฀The฀transcripts฀were฀then฀compared฀with฀one฀another,฀and฀with฀ the฀texts฀of฀the฀Petrocchi฀and฀Sanguineti,฀by฀Shaw฀and฀her฀team,฀using฀the฀฀ computer฀program฀Collate.฀The฀collation฀removed฀orthographic฀and฀ spelling฀variants,฀as฀can฀be฀seen฀in฀this฀collation฀of฀the฀second฀word฀of฀ the฀first฀line฀of฀the฀Inferno฀canto฀1฀shown฀in฀figure฀3: figure฀7 Collation฀of฀mezzo,฀Inf.฀i.฀1. Here,฀the฀eight฀different฀spellings฀of฀mezzo฀in฀the฀nine฀witnesses฀(seven฀ manuscripts฀plus฀Sanguineti฀[FS]฀and฀Petrocchi฀[PET])฀are฀regularized฀ to฀mezzo,฀so฀that฀no฀variant฀shows฀in฀this฀line. The฀collation฀also฀preserved฀information฀about฀the฀layers฀of฀scribal฀ revision,฀so฀that฀one฀could฀compare฀the฀different฀states฀of฀the฀text฀within฀ a฀manuscript,฀and฀between฀that฀manuscript฀and฀others.฀Thus,฀the฀collation฀at฀Purg.฀vii.฀51฀in฀figure฀8฀distinguishes฀the฀Aldine฀text฀(«Martorig»)฀and฀Martini’s฀collation฀(«Mart-c2»),฀and฀shows฀too฀that฀the฀reading฀«o฀pur»฀is฀found฀in฀Urb,฀and฀hence฀in฀Sanguineti฀(«FS»),฀while฀four฀ other฀manuscripts฀(Ash฀Ham฀Rb฀Triv)฀have฀the฀same฀reading฀as฀Martini฀ (this฀is฀also฀the฀reading฀of฀Petrocchi)฀and฀LauSC฀alone฀reads฀«o฀uer.» figure฀8 The฀collation฀of฀«o฀non»฀in฀Purg.฀vii.฀51. At฀the฀end฀of฀this฀process,฀a฀complete฀record฀of฀every฀significant฀variant฀at฀every฀word฀of฀the฀Commedia฀at฀every฀layer฀of฀writing฀in฀these฀ seven฀manuscripts฀was฀available. The฀next฀step฀was฀to฀analyze฀this฀record฀of฀variation฀to฀see฀what฀could฀ be฀deduced฀about฀the฀manuscript฀relations.฀Our฀analysis฀followed฀two฀ paths.฀First,฀Shaw฀used฀the฀traditional฀means฀of฀philology,฀scrutinizing฀all฀฀ the฀variants฀bearing฀on฀her฀argument,฀categorizing฀them,฀and฀seeking฀฀ explanations฀which฀take฀account฀of฀the฀full฀range฀of฀variants.฀Thus,฀in฀ the฀key฀section฀«The฀position฀of฀Rb»฀she฀looks฀at฀twenty-eight฀variants฀฀ The฀textual฀tradition฀of฀Dante’s฀Commedia 11 discussed฀by฀Petrocchi,฀and฀then฀supplements฀her฀discussion฀of฀these฀by฀ examination฀of฀a฀further฀76฀variants.฀Her฀argument฀hinges฀not฀only฀on฀ the฀nature฀of฀the฀variants฀in฀themselves,฀but฀also฀in฀the฀consistent฀agreement฀of฀Urb฀and฀Rb฀against฀all,฀or฀nearly฀all,฀the฀other฀manuscripts: Some฀of฀these฀categories฀are฀not฀especially฀significant฀in฀themselves฀(many฀of฀ them฀are฀included฀in฀Brandoli’s฀categories฀of฀polygenetic฀error).฀Singly,฀they฀ mean฀next฀to฀nothing.฀But฀it฀is฀the฀presence฀of฀a฀long฀series฀of฀them฀uniformly฀ right฀across฀the฀text฀in฀a฀very฀small฀number฀of฀manuscripts฀which฀is฀striking฀ (and฀this฀is฀surely฀what฀Petrocchi’s฀phrase฀«foltezza฀di฀statistica»฀refers฀to฀at฀least฀ in฀part).฀It฀is฀simply฀impossible฀to฀imagine฀that฀copyists฀working฀independently฀ would฀make฀precisely฀these฀small฀changes฀at฀precisely฀these฀same฀points฀right฀ across฀a฀text฀of฀this฀length.฀The฀most฀economical฀hypothesis฀is฀that฀they฀are฀working฀from฀a฀common฀exemplar฀and฀inherit฀these฀readings฀from฀that฀exemplar. figure฀9 The฀unrooted฀phylogram฀for฀the฀seven฀manuscripts,฀from฀the฀Shaw฀edition.฀The฀฀ arrow฀pointing฀to฀the฀node฀linking฀all฀of฀Ash/Ham,฀Urb/Triv,฀LauSC฀and฀Mart/ Triv฀has฀been฀added฀to฀this฀figure. ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Rb All฀mss,฀whole฀text ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀LauSC ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Urb ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Mart ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Triv Ash ————฀1000฀changes ฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀฀Ham 12 Peter฀Robinson In฀the฀second฀path฀of฀analysis,฀I฀took฀the฀complete฀record฀of฀variation฀created฀by฀the฀collation฀and฀submitted฀that฀to฀analysis฀by฀the฀program฀PAUP฀(«Phylogenetic฀Analysis฀Using฀Parsimony»),฀used฀by฀evolutionary฀biologists฀to฀create฀hypotheses฀about฀the฀genetic฀relationships฀฀ of฀organisms฀based฀on฀the฀characteristics฀they฀share฀and฀do฀not฀share.฀฀ A฀full฀account฀of฀how฀phylogenetic฀analysis฀works,฀and฀why฀and฀how฀it฀is฀ suited฀to฀the฀analysis฀of฀textual฀traditions,฀with฀evidence฀of฀its฀successful฀ use฀in฀many฀contexts฀(including฀on฀artificial฀traditions฀specially฀devised฀ to฀test฀its฀adequacy)฀may฀be฀found฀elsewhere.12฀figure฀9฀gives฀the฀«phylogram»฀for฀the฀whole฀Commedia฀generated฀by฀PAUP฀based฀on฀some฀ 94,000฀places฀of฀variation฀(typically฀a฀single฀word;฀but฀also฀phrases)฀in฀ the฀seven฀manuscripts. This฀representation฀of฀manuscript฀relations฀differs฀from฀a฀traditional฀ stemma฀(as฀for฀example,฀that฀given฀by฀Petrocchi).฀Firstly,฀it฀is฀«unrooted»:฀ that฀is,฀it฀represents฀the฀groupings฀of฀the฀manuscripts฀without฀any฀presumption฀as฀to฀originality,฀or฀the฀direction฀of฀variation.฀Secondly,฀the฀ manuscripts฀are฀shown฀as฀related฀to฀one฀another฀through฀shared฀nodes,฀ with฀three฀pairs฀each฀sharing฀a฀common฀node฀(Mart/Triv;฀Urb/Rb;฀Ash/ Ham,฀and฀the฀seventh฀manuscript฀(LauSC)฀descending฀from฀a฀node฀ on฀the฀line฀between฀Mart/Triv฀and฀the฀arrowed฀node฀linking฀the฀other฀ four฀manuscripts.฀The฀relative฀lengths฀of฀the฀lines฀between฀nodes,฀and฀ between฀nodes฀and฀manuscripts,฀are฀significant,฀and฀one฀can฀measure฀ approximately฀the฀differences฀between฀the฀manuscripts฀using฀the฀scale฀฀ «1000฀changes»฀(or,฀ten฀changes฀per฀canto,฀one฀every฀14.23฀lines)฀on฀the฀ bottom฀left.฀Thus:฀the฀line฀between฀the฀node฀joining฀Ash/Ham฀up฀to฀ the฀arrowed฀node฀which฀links฀to฀the฀other฀five฀manuscripts฀is฀about฀the฀ same฀length฀as฀the฀scale฀line฀(thus,฀c.1000฀changes),฀and฀approximately฀ double฀the฀length฀of฀the฀line฀from฀the฀arrowed฀node฀to฀the฀shared฀ancestor฀of฀Urb/Rb฀(thus,฀c.500฀changes),฀and฀approximately฀the฀same฀as฀from฀ the฀arrowed฀node฀to฀the฀ancestor฀of฀Mart/Triv฀(thus,฀c.1000฀changes฀฀ again). Experience฀using฀these฀phylograms฀has฀taught฀us฀to฀be฀careful.฀This฀ phylogenetic฀analysis฀appears฀to฀suggest฀that฀Sanguineti฀is฀wrong:฀Urb฀ and฀Rb฀do฀indeed฀share฀a฀common฀ancestor,฀as฀both฀Petrocchi฀and฀Shaw฀ assert.฀However,฀the฀software,฀left฀to฀itself,฀will฀always฀place฀a฀manuscript฀somewhere฀on฀the฀phylogram:฀sometimes฀this฀can฀suggest฀a฀rela฀See,฀for฀example,฀the฀publicatons฀listed฀at฀http://www.textualscholarship.org/newstemmatics/bibliography/index.html;฀notably฀the฀collection฀of฀articles฀in฀Pieter฀van฀ Reenen,฀August฀den฀Hollander฀and฀Margot฀van฀Mulken,฀eds.,฀Studies฀in฀Stemmatology฀ II,฀Amsterdam,฀John฀Benjamins,฀2004. 12 The฀textual฀tradition฀of฀Dante’s฀Commedia 13 tionship฀which฀in฀fact฀does฀not฀exist.฀Also,฀these฀phylograms฀cannot฀ represent฀contamination.฀Therefore,฀we฀have฀to฀turn฀to฀other฀methods฀ to฀confirm,฀deny,฀or฀qualify฀the฀relationships฀suggested฀by฀the฀phylogenetic฀analysis. In฀our฀analysis฀of฀the฀tradition฀of฀the฀Commedia,฀one฀of฀the฀«other฀ methods»฀Shaw฀and฀I฀used฀was฀database฀analysis฀of฀the฀distribution฀ of฀variants฀across฀the฀manuscripts฀and฀across฀the฀whole฀length฀of฀the฀฀ tradition.13฀Our฀hypothesis฀is฀this:฀consider฀the฀manuscripts฀descending฀directly฀from฀a฀single฀node฀(for฀example,฀Urb฀and฀Rb).฀If฀the฀manuscripts฀descended฀directly฀from฀this฀single฀node฀really฀share฀a฀common฀ ancestor฀below฀the฀archetype,฀then฀one฀should฀be฀able฀to฀identify฀a฀set฀ of฀variants฀likely฀to฀have฀been฀introduced฀into฀that฀shared฀ancestor,฀and฀ then฀descending฀to฀those฀manuscripts.฀Thus,฀there฀should฀be฀variants฀ which฀satisfy฀the฀following฀four฀conditions: 1.฀They฀should฀be฀present฀in฀the฀manuscripts฀descended฀directly฀from฀ the฀shared฀ancestor฀(thus,฀in฀both฀Urb฀and฀Rb); 2.฀They฀should฀be฀likely฀not฀to฀have฀been฀present฀in฀the฀archetype; 3.฀They฀should฀be฀found฀rarely฀in฀other฀manuscripts฀outside฀those฀ descended฀from฀the฀shared฀ancestor฀(thus,฀rarely฀in฀any฀of฀Ash฀Ham฀ LauSC฀Mart฀Triv); 4.฀There฀should฀be฀a฀signficant฀number฀of฀such฀variants. We฀used฀a฀database-like฀search฀tool,฀«VBase»,฀to฀find฀the฀variants฀which฀ satisfied฀these฀conditions.฀This฀tool฀is฀available฀on฀the฀DVD฀publication,฀ where฀it฀is฀pre-supplied฀with฀key฀searches฀we฀found฀useful฀in฀the฀course฀of฀ our฀analysis.฀Here฀is฀our฀search฀for฀variants฀likely฀to฀have฀been฀introduced฀ by฀the฀joint฀ancestor฀of฀Urb/Rb,฀and฀whose฀existence฀therefore฀supports฀ Petrocchi’s฀view฀that฀Urb฀and฀Rb฀share฀an฀ancestor฀below฀the฀archetype,฀ and฀contradicts฀Sanguineti’s฀view฀that฀they฀do฀not: ฀The฀database฀analysis฀was฀carried฀out฀by฀myself,฀employing฀all฀the฀data฀generated฀ by฀the฀transcription฀and฀collation฀directed฀by฀Shaw฀and฀using฀the฀criteria฀based฀on฀the฀ Shaw฀analysis.฀Shaw฀and฀I฀discussed฀and฀reviewed฀the฀results,฀and฀jointly฀wrote฀an฀article฀«Phylogenetic฀Analysis»,฀included฀in฀the฀Shaw฀edition,฀drawing฀heavily฀on฀the฀database฀analysis.฀Henceforth,฀this฀database฀analysis฀is฀described฀as฀the฀«Shaw/Robinson»฀ analysis,฀and฀the฀variant฀lists฀produced฀by฀it฀of฀the฀distinct฀groups฀as฀the฀«Shaw/Robinson»฀lists. 13 14 Peter฀Robinson figure฀10 VBase฀search฀for฀variants฀evidencing฀a฀common฀ancestor฀below฀the฀archetype฀ for฀Urb฀Rb. The฀first฀line฀of฀this฀search฀requests฀variants฀only฀present฀in฀both฀Rb฀ and฀Urb,฀and฀hence฀likely฀to฀have฀been฀present฀in฀their฀common฀ancestor฀(condition฀1,฀above).฀The฀next฀line฀asks฀only฀to฀see฀those฀variants฀฀ present฀in฀fewer฀than฀five฀(that฀is,฀in฀Urb฀Rb฀and฀no฀more฀than฀two฀ other)฀witnesses,฀and฀hence฀not฀likely฀to฀have฀been฀present฀in฀the฀ancestor฀of฀the฀whole฀tradition฀(condition฀2).฀The฀likelihood฀is฀that฀if฀a฀variant฀is฀present฀in฀three฀or฀more฀other฀witnesses฀besides฀Urb฀and฀Rb,฀then฀ that฀variant฀would฀very฀probably฀have฀been฀present฀in฀the฀archetype.฀ Indeed,฀if฀one฀increases฀the฀number฀from฀<5฀to฀<6,฀or฀more,฀we฀see฀the฀ search฀returning฀more฀and฀more฀variants,฀as฀it฀catches฀readings฀likely฀ present฀in฀the฀archetype฀(thus:฀488฀readings฀<6;฀806฀readings฀<7;฀1881฀ readings฀<8;฀6109฀<9).฀Accordingly,฀this฀restriction฀(returning฀only฀read- The฀textual฀tradition฀of฀Dante’s฀Commedia 15 ings฀in฀fewer฀than฀five฀witnesses)฀satisfies฀the฀second฀condition฀given฀ above.฀Note฀that฀because฀we฀include฀the฀Sanguineti฀edition฀(«FS»)฀in฀our฀ collation,฀and฀Sanguineti฀frequently฀chooses฀the฀Urb฀reading฀over฀that฀ of฀other฀manuscripts,฀as฀in฀all฀three฀cases฀from฀Inferno฀1฀given฀here,฀the฀ restriction฀to฀five฀is฀effectively฀Urb฀+฀RB฀+฀FS฀plus฀one฀other฀manuscript฀ only฀(in฀fact,฀fully฀247฀of฀these฀308฀readings฀returned฀by฀this฀search฀are฀฀ present฀in฀FS).฀ The฀next฀condition฀(condition฀3)฀is฀that฀the฀readings฀evidencing฀an฀ exclusive฀common฀ancestor฀for฀Urb฀and฀Rb฀should฀be฀rarely฀found฀in฀ any฀other฀manuscript.฀This฀table฀gives฀the฀number฀of฀occurrences฀of฀any฀ of฀these฀308฀readings฀across฀all฀the฀witnesses฀here฀collated: table฀1 Occurrence฀of฀Urb/Rb฀readings฀in฀all฀witnesses. Manuscript Number฀of฀Urb/Rb฀ readings฀(max.฀308) Percent฀of฀Urb/Rb฀ readings฀present Ash 40 13 Ham 46 15 LauSC 11 4 Mart 4 1 Rb 308 100 Triv 9 3 Urb 308 100 FS 247 80 PET 19 6 Because฀of฀the฀operation฀of฀co-incident฀(or฀convergent)฀variation,฀it฀ is฀to฀be฀expected฀that฀by฀simple฀chance,฀a฀proportion฀of฀readings฀introduced฀by฀one฀scribe฀in฀one฀copying฀might฀be฀found฀in฀a฀quite฀unrelated฀ copy฀made฀by฀a฀different฀scribe.฀One฀might฀hypothesize฀that฀in฀any฀two฀ acts฀of฀copying,฀the฀different฀scribes฀might฀introduce฀the฀same฀mistake฀ some฀ten฀percent฀of฀the฀time,฀or฀in฀one฀in฀ten฀variants.฀On฀this฀calculation,฀the฀incidence฀of฀the฀308฀Urb/Rb฀in฀all฀except฀FS฀Ash฀Ham฀can฀be฀ explained฀as฀simple฀coincidence฀–฀and฀the฀high฀number฀of฀Urb/Rb฀variants฀present฀in฀FS฀results฀from฀Sanguineti’s฀decision฀to฀base฀his฀edition฀ 16 Peter฀Robinson on฀Urb.฀The฀number฀of฀Urb/Rb฀variants฀in฀Ash฀and฀Ham฀is฀marginally฀ higher฀than฀one฀might฀expect,฀and฀this฀may฀be฀the฀result฀of฀contamination฀between฀the฀Urb/Rb฀and฀Ash/Ham฀groups฀(see฀below).฀Accordingly,฀this฀group฀of฀variants฀satisfies฀the฀third฀condition฀given฀above:฀ readings฀from฀this฀group฀are฀rarely฀found฀in฀other฀manuscripts.฀ The฀last฀condition฀is฀that฀the฀number฀of฀variants฀should฀be฀«significant»:฀that฀is,฀not฀just฀a฀random฀agreement฀between฀these฀two฀witnesses฀ but฀the฀likely฀result฀of฀shared฀descent฀from฀a฀single฀ancestor฀containing฀ these฀readings.฀Again,฀one฀can฀use฀VBase฀searches฀to฀explore฀what฀levels฀of฀ agreement฀between฀any฀two฀witnesses฀might฀be฀the฀result฀of฀co-incident฀ variation.฀From฀the฀phylogram฀given฀in฀figure฀9,฀one฀may฀hypothesize฀ that฀the฀following฀pairs฀of฀manuscripts฀do฀not฀share฀a฀common฀ancestor฀below฀the฀archetype:฀Rb/Mart,฀Rb/Triv,฀Urb/Mart,฀Urb/Triv,฀Urb/ LauSC,฀Rb/LauSC,฀Rb/Ham,฀Urb/Ham,฀Rb/Ash,฀Urb/Ash,฀Ash/LauSC,฀ Ham/LauSC,฀Mart/Ash,฀Mart/Ham,฀Triv/Ash,฀Triv/Ham,฀Mart/LauSC,฀฀ Triv/LauSc.฀Here฀are฀the฀results฀of฀VBase฀searches฀on฀these฀pairs฀of฀manuscripts,฀using฀the฀same฀criteria฀employed฀for฀the฀search฀given฀in฀figure฀10: table฀2 Number฀of฀readings฀found฀in฀hypothetically฀unrelated฀witness฀pairs,฀likely฀to฀be฀ the฀result฀of฀chance฀agreement.฀All฀searches฀were฀for:฀witness฀pair฀+฀<4฀of฀all฀witnesses,฀i.e.฀witness฀pair฀+฀one฀other,฀except฀for฀pairs฀including฀Urb฀(marked฀*),฀ where฀searches฀were฀for฀witness฀pair฀+฀<5฀of฀all฀witnesses,฀i.e.฀witness฀pair฀+฀FS฀+฀฀ one฀other. Witness฀Pair Number฀of฀readings฀found฀by฀Vbase Rb/Mart 37 Rb/Triv 91 Urb/Mart* 28 Urb/Triv* 58 Urb/LauSC* 99 Rb/LauSC 70 Rb/Ham 219 Urb/Ham* 203 Rb/Ash 201 The฀textual฀tradition฀of฀Dante’s฀Commedia Urb/Ash* 167 Ash/LauSC 103 Ham/LauSC 91 Mart/Ash 45 Mart/Ham 57 Triv/Ash 102 Triv/Ham 123 Mart/LauSC 35 Triv/LauSc 107 17 For฀all฀except฀the฀pairings฀involving฀Urb/Rb/Ash/Ham,฀these฀searches฀ typically฀return฀fewer฀than฀one฀hundred฀variants฀(ranging฀from฀35฀for฀ Mart/LauSC฀up฀to฀123฀for฀Triv/Ham).฀One฀may฀infer฀from฀this฀that฀for฀this฀฀ text,฀any฀two฀witnesses฀may฀happen฀to฀share฀up฀to฀around฀100฀readings฀ (or฀one฀a฀canto)฀and฀it฀will฀mean฀nothing฀whatever:฀the฀agreement฀is฀just฀ chance.฀Conversely,฀one฀may฀deduce฀that฀chance฀cannot฀explain฀two฀wit-฀ nesses฀sharing฀308฀variants฀(or฀three฀a฀canto):฀this฀is฀likely฀to฀be฀the฀ result฀of฀shared฀descent฀from฀a฀common฀ancestor.฀Thus,฀these฀308฀readings฀ may฀ satisfy฀ the฀ fourth฀ condition:฀ their฀ number฀ is฀ significant฀ of฀ more฀than฀chance฀agreement.฀ I฀except฀here฀the฀pairings฀involving฀Urb/Rb฀with฀Ash/Ham.฀The฀number฀฀ of฀agreements฀between฀any฀two฀of฀these฀ranges฀from฀167฀(Urb/Ash)฀to฀ 219฀(Rb/Ham).฀This฀seems฀higher฀than฀one฀might฀expect฀from฀random฀ agreement,฀and฀we฀recall฀that฀the฀number฀of฀Urb/Rb฀readings฀found฀in฀ Ash฀and฀Ham฀given฀in฀Table฀1฀was฀also฀rather฀higher฀than฀one฀might฀ expect฀from฀random฀agreement.฀Shaw฀and฀I฀examine฀this฀question฀in฀ our฀«Phyologenetic฀analysis»฀argument฀and฀suggest฀that฀this฀agreement฀ across฀the฀pairs฀Ash/Ham฀and฀Urb/Rb฀is฀likely฀to฀be฀the฀result฀of฀con-฀ tamination,฀ particularly฀ in฀ Inferno฀ 1-7฀ and฀ the฀ last฀ thirteen฀ canti฀ of฀฀ Paradiso. To฀summarize:฀by฀this฀process฀of฀VBase฀searching,฀built฀upon฀the฀ results฀of฀the฀phylogenetic฀analysis,฀we฀isolated฀a฀list฀of฀308฀variants฀ which฀we฀believe฀are฀evidence฀of฀a฀family฀relationship฀between฀the฀two฀ manuscripts฀Rb฀and฀Urb.฀The฀phylogram฀(figure฀9)฀suggests฀that฀there฀ are฀two฀other฀manuscript฀pairs฀among฀the฀seven:฀Ash/Ham฀and฀Mart/ Triv.฀Further,฀it฀may฀be฀that฀the฀five฀witnesses฀Ash/Ham/Mart/Triv/ 18 Peter฀Robinson LauSC฀(that฀is,฀the฀two฀pairs฀Ash/Ham฀and฀Mart/Triv฀plus฀LauSC)฀may฀ also฀descend฀from฀a฀single฀copy฀below฀the฀archetype,฀and฀therefore฀also฀ form฀a฀family฀grouping฀within฀the฀tradition.฀Following฀Petrocchi,฀we฀ call฀this฀group฀α.฀For฀all฀these฀groupings,฀we฀carried฀out฀similar฀VBase฀ searches฀to฀that฀we฀described฀above฀for฀Urb/Rb,฀identifying฀for฀each฀฀ a฀set฀of฀variants฀characteristic฀of฀each฀group฀and฀likely฀to฀have฀been฀intro-฀ duced฀by฀the฀exclusive฀common฀ancestor฀of฀each.฀The฀groups฀(hereafter,฀the฀Shaw/Robinson฀lists),฀the฀number฀of฀variants฀we฀found฀for฀each฀ group,฀and฀the฀number฀of฀lines฀in฀which฀these฀variants฀occur฀(rather฀ fewer฀than฀the฀number฀of฀variants,฀as฀some฀lines฀have฀more฀than฀one฀ variant)฀are฀given฀in฀table฀3. table฀3 Numbers฀of฀variants฀evidencing฀hypothetical฀manuscript฀groups,฀and฀numbers฀ of฀lines฀in฀which฀they฀occur. Group No.฀Variants No.฀Lines Urb/Rb 308 297 Mart/Triv 874 842 Ash/Ham 770 722 α 327 314 total 1885 Taken฀together,฀this฀gives฀a฀total฀number฀of฀1885฀lines฀containing฀variants฀which฀our฀analysis฀indicates฀are฀significant฀for฀establishing฀family฀ relationships฀among฀these฀seven฀witnesses.14฀That฀is:฀of฀the฀14233฀lines฀ in฀the฀whole฀Commedia,฀it฀appears฀that฀1885฀give฀evidence฀of฀family฀relationships.฀Earlier฀in฀this฀essay฀I฀described฀the฀Barbi฀loci:฀his฀list฀of฀396฀lines฀ which฀he฀chose฀as฀indicative฀for฀the฀establishment฀of฀manuscript฀relations฀ across฀the฀whole฀14223฀lines฀of฀the฀Commedia.฀We฀have฀1885฀lines฀which฀ we฀find฀are฀indicative;฀Barbi฀suggests฀396฀which฀he฀believed฀to฀be฀indicative. This฀gives฀us฀a฀standpoint฀from฀which฀we฀can฀assess฀Barbi’s฀choice฀ of฀lines.฀Both฀Barbi฀and฀ourselves฀(Shaw฀and฀myself)฀were฀seeking฀the฀ same฀thing.฀Accordingly,฀one฀would฀expect฀a฀high฀degree฀of฀correspond14 ฀If฀one฀adds฀the฀numbers฀of฀lines฀(297+842+722+314)฀the฀sum฀is฀2175,฀not฀1885.฀ This฀is฀because฀290฀lines฀contain฀variants฀from฀more฀than฀one฀group.฀ The฀textual฀tradition฀of฀Dante’s฀Commedia 19 ence฀between฀the฀two฀lists฀of฀lines:฀say,฀with฀very฀many฀of฀Barbi’s฀396฀ lines฀appearing฀in฀our฀larger฀list฀of฀1885฀lines.฀Of฀course,฀the฀same฀lines฀ might฀appear฀on฀both฀lists฀and฀mean฀nothing฀more฀than฀chance฀agreement.฀In฀fact,฀there฀are฀well-known฀probabilistic฀methods฀for฀estimating฀in฀a฀case฀such฀as฀this฀what฀proportion฀of฀agreement฀between฀the฀lists฀ might฀be฀simple฀chance,฀and฀what฀more฀than฀chance. Consider฀that฀we฀have฀an฀urn฀containing฀14233฀balls,฀each฀with฀a฀ number฀from฀1฀to฀14233฀drawn฀on฀it:฀our฀14233฀lines฀of฀the฀Commedia฀.15฀ We฀choose,฀purely฀at฀random,฀1885฀balls฀from฀the฀urn:฀this฀is฀the฀Shaw/ Robinson฀list.฀We฀put฀all฀1885฀balls฀back,฀and฀we฀choose,฀purely฀at฀random,฀ 396฀balls฀from฀the฀urn:฀this฀is฀the฀Barbi฀list.฀Now,฀for฀any฀one฀ball฀that฀ we฀happen฀to฀choose,฀the฀chance฀that฀it฀is฀on฀the฀Shaw/Robinson฀list฀is:฀ 1885฀/฀14333฀(=0.13244฀approx) So฀if฀we฀choose฀396฀balls฀from฀the฀urn,฀then฀by฀simple฀chance฀the฀number฀ of฀balls฀we฀are฀likely฀to฀get฀which฀are฀on฀the฀Shaw/Robinson฀list฀will฀be: 396฀x฀1885฀/฀14233฀(=396฀x฀0.13244)฀=฀52.446 That฀is:฀by฀simple฀chance฀agreement,฀if฀we฀chose฀396฀lines฀at฀random฀ several฀times฀over฀(taking฀care฀to฀put฀back฀all฀the฀balls฀back฀in฀the฀urn฀ between฀choices)฀we฀would฀expect฀the฀average฀number฀of฀lines฀chosen฀ to฀approximate฀to฀52.446.฀Say฀we฀did฀this฀exercise฀of฀choosing฀two฀sets฀of฀ balls฀(one,฀1885฀balls,฀another฀396฀balls)฀a฀hundred฀times,฀two฀hundred฀ times,฀each฀time฀noting฀down฀the฀number฀of฀balls฀in฀common฀between฀ the฀two฀choices.฀Sometimes฀there฀would฀be฀45฀in฀common,฀sometimes฀ 59,฀sometimes฀49,฀sometimes฀55,฀etc.:฀but฀the฀more฀times฀we฀chose฀the฀ two฀sets฀of฀balls,฀the฀more฀the฀average฀number฀of฀lines฀in฀both฀lists฀would฀ approach฀52.446.฀Hence,฀52.466฀is฀the฀«mean฀number». Probability฀can฀do฀even฀more฀for฀us.฀The฀«mean฀number»฀is฀useful,฀ as฀we฀can฀presume฀that฀if฀we฀find฀that฀the฀number฀of฀lines฀in฀common฀ on฀the฀two฀lists฀is฀close฀to฀this฀number,฀it฀is฀mere฀random฀chance.฀Thus,฀ ฀For฀the฀discussion฀that฀follows,฀with฀the฀analogy฀of฀the฀urn฀containing฀14233฀balls,฀ the฀explanation฀of฀hypergeometric฀distribution,฀and฀many฀of฀the฀calculations฀of฀probabilities฀here฀given,฀I฀am฀deeply฀indebted฀to฀Steven฀J.฀Schwager,฀now฀Professor฀Emeritus฀ in฀the฀Departments฀of฀Biological฀Statistics฀and฀Computational฀Biology,฀and฀of฀Statistical฀Science,฀at฀Cornell฀University.฀I฀have฀taken฀over฀much฀of฀his฀phrasing,฀and฀his฀key฀ statistical฀results.฀I฀am฀grateful฀also฀to฀Teemu฀Roos฀of฀the฀University฀of฀Helsinki฀for฀his฀ corrections฀and฀suggestions. 15 20 Peter฀Robinson if฀we฀find฀52฀lines฀in฀common฀between฀the฀two฀lists,฀this฀is฀just฀chance.฀ But฀what฀if฀we฀find฀51฀lines,฀or฀53?฀Instinctively,฀we฀will฀think฀this฀is฀ likely฀to฀be฀chance฀too.฀But฀what฀of฀70฀lines,฀or฀100,฀or฀30?฀At฀what฀point฀ does฀it฀become฀highly฀unlikely฀that฀a฀given฀number฀of฀lines฀will฀occur฀ on฀both฀lists฀by฀chance?฀Probability฀can฀give฀us฀answers฀to฀these฀questions฀too.฀This฀situation,฀where฀we฀can฀assign฀statistical฀probabilities฀to฀ the฀range฀of฀numbers฀of฀possible฀lines฀in฀common฀between฀the฀two฀lists฀ (between฀0,฀for฀no฀numbers฀in฀common,฀and฀396,฀with฀every฀number฀ on฀the฀Barbi฀list฀appearing฀on฀our฀list)฀is฀an฀instance฀of฀what฀statisticians฀call฀a฀«hypergeometric฀distribution»,฀and฀it฀is฀possible฀to฀calculate฀ various฀probabilities฀for฀each฀number฀of฀lines฀between฀0฀and฀396,฀thus: 1.฀The฀probability฀that฀exactly฀this฀number฀of฀lines฀should฀be฀in฀com-฀ mon฀in฀the฀two฀lists.฀For฀52฀lines,฀that฀is฀0.0599,฀or฀slightly฀better฀than฀ one฀chance฀in฀twenty฀(for฀51฀it฀is฀0.0592,฀for฀53฀0.0594) 2.฀The฀probability฀that฀there฀should฀be฀this฀number฀of฀lines฀or฀fewer฀ in฀common฀in฀the฀two฀lists.฀For฀52฀lines,฀that฀is฀0.5102,฀or฀just฀better฀than฀ one฀chance฀in฀two฀–฀the฀equivalent฀of฀tossing฀a฀coin฀and฀getting฀heads. 3.฀The฀probability฀that฀there฀should฀be฀this฀number฀of฀lines฀or฀more฀in฀ common฀in฀the฀two฀lists.฀For฀52฀lines,฀that฀is฀0.5498,฀or฀just฀better฀than฀one฀ chance฀in฀two฀–฀again,฀the฀equivalent฀of฀tossing฀a฀coin฀and฀getting฀heads. We฀can฀use฀these฀figures฀to฀tell฀us,฀for฀any฀given฀number฀of฀lines,฀not฀ just฀what฀the฀chances฀are฀of฀exactly฀that฀number฀of฀lines฀occurring฀in฀both฀ lists,฀but฀also฀what฀are฀the฀chances฀of฀a฀lesser฀or฀greater฀number฀of฀lines฀ occurring.฀That฀is:฀we฀can฀use฀the฀hypergeometric฀distribution฀table฀to฀tell฀ us฀(say)฀what฀are฀the฀chances฀of฀there฀being฀30฀or฀less฀lines฀in฀common฀in฀ the฀two฀lists฀(the฀number฀is฀0.0002,฀or฀around฀2฀chances฀in฀10,000),฀or฀of฀70฀ lines฀or฀more฀in฀the฀two฀lists฀(the฀number฀is฀0.0066,฀or฀six฀chances฀in฀1000).฀ Even฀more฀usefully:฀we฀can฀examine฀the฀hypergeometric฀distribution฀ to฀assign฀«upper»฀and฀«lower»฀bounds.฀In฀statistical฀terms,฀anything฀ with฀a฀less฀than฀0.05฀probability,฀that฀is฀less฀than฀one฀in฀twenty,฀may฀be฀ regarded฀as฀unlikely.฀The฀«upper»฀bound฀will฀be฀the฀lowest฀value฀with฀a฀ probability฀of฀less฀than฀0.025฀that฀there฀will฀be฀so฀many฀lines฀in฀common฀ between฀the฀two฀lists;฀the฀«lower»฀bound฀will฀be฀the฀highest฀value฀with฀a฀ probability฀of฀less฀than฀0.025฀that฀there฀will฀be฀so฀many฀lines฀in฀common฀ between฀the฀two฀lists.฀For฀the฀Barbi฀list฀of฀396฀lines,฀the฀probability฀of฀ there฀being฀66฀or฀more฀lines฀in฀common฀with฀the฀Shaw/Robinson฀list฀ of฀1885฀lines฀is฀0.0275฀(for฀67฀lines฀or฀more,฀it฀is฀0.020):฀this฀gives฀an฀ «upper฀bound»฀of฀66.฀The฀probability฀of฀there฀being฀39฀or฀less฀lines฀in฀ common฀with฀the฀Shaw/Robinson฀list฀is฀0.023฀(for฀40฀lines฀or฀less,฀it฀is฀ 0.039):฀this฀gives฀a฀«lower฀bound»฀of฀39.฀Hence฀one฀can฀declare฀that฀the฀ The฀textual฀tradition฀of฀Dante’s฀Commedia 21 chances฀are฀around฀19฀in฀20฀that฀by฀simple฀chance,฀there฀will฀be฀between฀ 39฀and฀66฀lines฀in฀common฀on฀the฀two฀lists.฀Correspondingly,฀if฀we฀find฀ the฀number฀of฀lines฀in฀common฀is฀outside฀the฀range฀39฀to฀66,฀then฀it฀is฀ unlikely฀that฀this฀is฀simple฀chance;฀and฀the฀further฀outside฀this฀range฀ is฀the฀number฀of฀lines฀in฀common,฀the฀more฀unlikely฀it฀is฀that฀this฀is฀ simple฀chance. Armed฀ with฀ these฀ formulae,฀ we฀ may฀ look฀ at฀ the฀ correspondences฀ between฀the฀Barbi฀loci฀and฀the฀Shaw/Robinson฀lists฀of฀lines฀we฀found฀ significant฀for฀the฀stemmatic฀analysis฀of฀the฀Commedia฀(Table฀4). table฀4 Numbers฀ of฀ lines฀ on฀ both฀ the฀ Shaw/Robinson฀ and฀ Barbi฀ lists฀ (second฀ last฀ column),฀compared฀to฀the฀number฀expected฀by฀simple฀chance฀(third฀column),฀ the฀upper฀and฀lower฀numbers฀consistent฀with฀chance฀(<0.05,฀or฀one฀in฀twenty),฀ and฀the฀probability฀of฀that฀number฀of฀lines฀or฀greater฀in฀common฀(last฀column). Group Shaw Barbi฀and฀ Shaw/Rob.฀ Mean Lower bound Upper bound Barbi฀and฀ Shaw/Rob.฀ actual16 Probability (>=) Urb/Rb 297 8 3 14 15฀(6) 0.0194฀(0.838) Mart/Triv 842 23 14 34 64฀(64) <1x10-12฀ (superscript฀-12) Ash/Ham 722 20 11 30 49฀(37) <1x10-10฀ (superscript฀-10) (0.00025) α 314 9 3 16 19฀(14) 0.0013฀ (0.0566) totals 1885 52 39 66 121฀ <1x10-13 (superscript฀-13) 16 ฀Many฀of฀the฀Barbi฀lines฀correspond฀with฀more฀than฀one฀group฀in฀the฀Shaw/Robinson฀analysis.฀That฀is:฀while฀a฀total฀of฀121฀lines฀from฀the฀Barbi฀lists฀appear฀among฀the฀ 1885฀lines฀identified฀in฀the฀Shaw฀analysis,฀26฀of฀the฀121฀lines฀appear฀in฀more฀than฀one฀ of฀the฀Shaw/Robinson฀lists.฀Accordingly,฀one฀should฀adjust฀the฀count฀of฀corresponding฀lines฀for฀each฀group฀to฀take฀account฀of฀this฀overlap:฀the฀adjusted฀line฀counts฀are฀ given฀in฀parentheses฀in฀this฀column,฀and฀the฀corresponding฀probability฀in฀parentheses฀in฀the฀last฀column.฀To฀explain:฀presuming,฀as฀I฀argue,฀that฀Barbi’s฀choice฀of฀lines฀was฀฀ conditioned฀by฀his฀awareness฀of฀Triv,฀then฀lines฀which฀evidence฀both฀Mart/Triv฀and฀ another฀grouping฀should฀not฀be฀counted฀among฀lines฀found฀by฀Barbi฀for฀that฀other฀฀ grouping.฀Therefore,฀one฀should฀remove฀from฀the฀Urb/Rb฀count฀of฀15฀the฀four฀lines฀ also฀found฀in฀the฀Mart/Triv฀grouping,฀hence฀11.฀Triv฀is฀also฀a฀prime฀witness฀for฀variants฀ found฀in฀the฀hypothetical฀α฀group,฀and฀so฀one฀should฀also฀remove฀the฀five฀lines฀shared฀ by฀Rb/Urb฀with฀α:฀hence,฀six฀variants฀(15-4-5),฀as฀given฀in฀parentheses.฀Similarly,฀one฀ 22 Peter฀Robinson Look฀first฀at฀the฀bottom฀«Totals»฀line฀in฀table฀4.฀This฀suggests฀that฀ given฀the฀Shaw/Robinson฀list฀of฀1885฀lines฀and฀the฀Barbi฀list฀of฀396฀lines,฀ both฀drawn฀from฀the฀14233฀lines฀of฀the฀Commedia,฀one฀would฀expect฀฀ that฀by฀simple฀chance฀alone฀between฀39฀and฀66฀lines฀might฀appear฀on฀฀ both฀lists.฀The฀last฀column฀tells฀us฀that฀Barbi฀in฀fact฀has฀121฀of฀the฀Shaw/ Robinson฀list,฀considerably฀higher฀than฀one฀might฀expect฀from฀simple฀ chance.฀Indeed,฀the฀chance฀of฀there฀being฀121฀(or฀more)฀lines฀in฀common฀ is฀0฀to฀twelve฀decimal฀places;฀or,฀to฀put฀it฀another฀way,฀roughly฀equal฀to฀฀ the฀chance฀of฀throwing฀a฀coin฀forty฀times฀and฀getting฀heads฀each฀time.฀฀ It฀cannot฀be฀chance฀that฀there฀are฀so฀many฀lines฀appearing฀on฀both฀lists.฀It฀฀ must฀be฀because฀both฀lists฀seek฀to฀represent฀the฀same฀phenomena,฀in฀this฀ case:฀lines฀useful฀for฀genetic฀reconstruction฀of฀the฀tradition.฀For฀these฀ 121฀lines,฀then,฀Barbi’s฀hypothesis฀that฀these฀lines฀would฀be฀useful฀for฀ genetic฀reconstruction฀appears฀justified฀by฀our฀analysis.฀Accordingly,฀an฀ analysis฀based฀on฀the฀396฀lines฀selected฀by฀Barbi฀will฀give฀a฀better฀picture฀ of฀the฀manuscript฀relations฀than฀(say)฀just฀choosing฀any฀396฀lines.฀To฀ that฀extent,฀one฀might฀indeed฀use฀the฀Barbi฀loci฀towards฀a฀genetic฀reconstruction฀of฀the฀history฀of฀the฀tradition,฀as฀Barbi฀planned฀and฀as฀Sanguineti฀attempted. However,฀not฀all฀genetic฀reconstructions฀are฀equal.฀A฀closer฀look฀at฀ Table฀4฀shows฀some฀problems.฀Consider฀the฀numbers฀of฀variants฀found฀ in฀the฀Barbi฀list฀which฀also฀appear฀on฀the฀Shaw/Robinson฀lists฀for฀each฀฀ of฀the฀four฀groups.฀The฀number฀of฀lines฀for฀the฀Mart/Triv฀group฀found฀on฀฀ the฀Barbi฀list฀is฀much฀higher฀than฀one฀would฀expect฀by฀simple฀chance:฀ 64,฀over฀double฀the฀number฀which฀could฀be฀reasonably฀explained฀by฀ chance฀alone฀(taking฀the฀upper฀bound฀of฀34฀as฀the฀most฀chance฀might฀ reasonably฀give).฀For฀Ash/Ham,฀the฀number฀is฀significantly฀higher฀than฀ chance฀might฀give฀(though฀less฀so฀than฀for฀the฀Mart/Triv).฀But฀for฀α the฀number฀is฀only฀marginally฀higher฀than฀chance฀might฀suggest฀(19,฀ compared฀to฀an฀upper฀limit฀of฀16)฀and฀for฀the฀crucial฀pair฀Urb/Rb฀the฀ number฀of฀15฀lines฀is฀only฀just฀higher฀than฀chance฀would฀explain. It฀seems฀then฀that฀analysis฀based฀on฀the฀Barbi฀loci฀will฀be฀very฀effective฀ at฀revealing฀the฀Mart/Triv฀and฀Ash/Ham฀pairs,฀but฀much฀less฀effective฀for฀฀ α and฀ for฀ Urb/Rb.฀ The฀ prominence฀ of฀ Mart/Triv฀ in฀ the฀ Barbi฀ loci฀ is฀฀ likely฀to฀be฀no฀accident.฀The฀Trivulziano฀1080฀manuscript฀–฀exceptional฀ for฀the฀extraordinary฀beauty฀of฀its฀script,฀with฀page฀after฀page฀free฀from฀ should฀discount฀the฀49฀Ash/Ham฀lines฀by฀the฀10฀lines฀found฀in฀Mart/Triv฀and฀the฀two฀ found฀in฀α,฀hence฀12;฀and฀discount฀also฀the฀19 α฀lines฀by฀the฀three฀found฀in฀Mart/Triv฀ and฀the฀two฀found฀in฀Ash/Ham,฀hence฀14. The฀textual฀tradition฀of฀Dante’s฀Commedia 23 any฀blemish,฀and฀famous฀for฀its฀scribe฀Francesco฀di฀Ser฀Nardo,฀its฀age,฀the฀ consistency฀of฀its฀language฀(see฀the฀description฀in฀Shaw)฀–฀was฀already฀ well-known฀to฀Barbi.฀It฀is฀reasonable฀to฀assume฀that฀Barbi’s฀knowledge฀of฀ Triv฀was฀a฀factor฀in฀his฀choice฀of฀loci;฀hence฀the฀high฀number฀of฀Mart/Triv฀ lines฀found฀in฀the฀Barbi฀list.฀Accordingly,฀we฀should฀discount฀the฀numbers฀of฀lines฀given฀in฀the฀last฀column฀of฀table฀4฀as฀suggested฀in฀the฀footnote฀to฀the฀table,฀and฀use฀the฀numbers฀given฀in฀brackets฀as฀the฀base฀for฀ our฀assessment.฀This฀(of฀course)฀has฀no฀effect฀on฀the฀efficacy฀of฀the฀Barbi฀ list฀as฀far฀as฀Mart/Triv฀is฀concerned,฀and฀the฀Ash/Ham฀count฀(37฀adjusted฀ variants฀against฀an฀upper฀chance฀bound฀of฀30)฀is฀also฀still฀beyond฀what฀ could฀be฀explained฀by฀chance.฀But฀the฀number฀of฀Urb/Rb฀lines฀on฀the฀ Barbi฀list฀now฀falls฀to฀six:฀lower฀even฀than฀the฀expected฀number฀which฀ simple฀chance฀would฀give. Sanguineti,฀as฀we฀have฀seen,฀based฀his฀entire฀stemmatic฀analysis฀on฀ the฀Barbi฀loci.฀Accordingly,฀he฀would฀have฀collated฀just฀six฀lines฀which฀ (in฀the฀Shaw฀analysis)฀alone฀contain฀readings฀likely฀to฀have฀been฀introduced฀by฀the฀common฀ancestor฀of฀Urb/Rb.฀Indeed,฀in฀his฀analysis฀he฀ refers฀to฀only฀four฀of฀these฀lines.17฀Shaw฀is฀able฀to฀base฀her฀assertion฀ that฀Rb฀and฀Urb฀share฀a฀common฀ancestor฀below฀the฀archetype฀on฀the฀ 308฀variant฀readings,฀found฀in฀297฀lines,฀identified฀by฀VBase฀working฀ on฀the฀results฀of฀the฀phylogenetic฀analysis.฀Sanguineti,฀it฀appears,฀following฀Barbi,฀examined฀only฀a฀handful฀of฀lines฀likely฀to฀contain฀readings฀which฀evidence฀common฀descent฀of฀Urb฀and฀Rb฀from฀an฀ancestor฀ below฀the฀archetype฀(Petrocchi’s฀β).฀Hence,฀his฀argument฀that฀Urb฀and฀ Rb฀do฀not฀share฀any฀such฀ancestor,฀that฀Urb฀descends฀directly฀from฀the฀ archetype,฀sharing฀no฀intermediate฀ancestor฀with฀any฀other฀extant฀manuscript.฀This฀gives฀Urb,฀in฀his฀analysis,฀extraordinary฀and฀unique฀status:฀ 17 ฀The฀four฀are:฀Inf.฀xi฀84฀men฀Dio฀offende฀e฀men฀biasimo฀accatta;฀Inf.฀xiii฀63฀tanto฀ ch’฀i’฀ne฀perde’฀li฀sonni฀e’฀polsi;฀Purg.฀iv฀72฀che฀mal฀non฀seppe฀carreggiar฀Fetòn;฀Purg.฀xvi฀ 145.฀Così฀tornò,฀e฀più฀non฀volle฀udirmi.฀The฀two฀he฀appears฀not฀to฀have฀used฀are฀Inf.฀xxvi฀ 15฀rimontò฀’l฀duca฀mio฀e฀trasse฀mee฀and฀Purg.฀v฀88฀Io฀fui฀di฀Montefeltro,฀io฀son฀Bonconte.฀Another฀two฀appear฀in฀Mart/Triv฀as฀well฀as฀in฀Urb/Rb,฀and฀are฀used฀by฀Sanguineti:฀Inf.฀i฀28฀Poi฀ch’฀èi฀posato฀un฀poco฀il฀corpo฀lasso;฀Purg.฀xviii฀57฀e฀de’฀primi฀appetibili฀ l’฀affetto;฀two฀appear฀in฀Mart/Triv฀but฀are฀not฀used฀by฀Sanguineti:฀Purg.฀xviii฀58฀che฀sono฀฀ in฀voi฀sì฀come฀studio฀in฀ape;฀Purg.฀xxiv฀61฀e฀qual฀più฀a฀gradire฀oltre฀si฀mette;฀two฀appear฀in฀฀ α and฀are฀used฀by฀Sanguineti:฀Inf.฀xvii฀50฀or฀col฀ceffo฀or฀col฀piè,฀quando฀son฀morsi;฀฀ Purg.฀xiii฀121฀tanto฀ch’฀io฀volsi฀in฀sù฀l’฀ardita฀faccia;฀three฀appear฀in฀α฀and฀are฀not฀used฀by฀ Sanguineti:฀Inf.฀i฀116฀vedrai฀li฀antichi฀spiriti฀dolenti;฀Purg.฀xxiii฀2฀ficcava฀ïo฀sì฀come฀far฀ suole;฀Par.฀i฀54฀e฀fissi฀li฀occhi฀al฀sole฀oltre฀nostr’฀uso.฀It฀should฀be฀noted฀that฀Sanguineti฀ also฀does฀not฀use฀another฀line฀on฀the฀Barbi฀list,฀Par.฀I฀25,฀which฀both฀Petrocchi฀and฀Shaw฀ regard฀as฀significant฀for฀analysis฀of฀Urb/Rb. 24 Peter฀Robinson indeed,฀the฀authority฀of฀this฀one฀manuscript฀Urb฀is฀equivalent฀to฀that฀ of฀all฀the฀other฀800-plus฀manuscripts฀of฀the฀Commedia฀put฀together.฀ Accordingly,฀Sanguineti฀consistently฀prefers฀the฀readings฀of฀Urb฀to฀those฀ of฀other฀manuscripts,฀accepting฀the฀Urb฀reading฀in฀every฀case฀except฀ where฀there฀is฀clear฀error. Sanguineti’s฀assertion฀of฀the฀unique฀authority฀of฀Urb฀affects฀his฀text฀ significantly.฀In฀some฀1406฀places,฀he฀accepts฀the฀reading฀of฀Urb฀over฀that฀ found฀in฀other฀manuscripts฀and฀accepted฀by฀Petrocchi.฀If฀you฀read฀the฀฀ Commedia฀in฀Sanguineti’s฀edition,฀about฀one฀line฀in฀ten฀will฀be฀different฀from฀that฀given฀in฀the฀widely-accepted฀Petrocchi฀text฀(or,฀indeed,฀ from฀any฀edition฀based฀on฀Triv฀or฀other฀manuscripts฀from฀the฀Florentine฀tradition).฀All฀this฀follows฀from฀his฀assertion฀that฀Urb฀is฀not฀genetically฀related฀to฀Rb,฀which฀itself฀can฀be฀traced฀to฀his฀apparent฀examination฀of฀just฀a฀handful฀of฀the฀many฀variants฀which฀in฀Shaw’s฀analysis฀(as฀ in฀Petrocchi’s)฀evidence฀this฀relationship. How฀did฀this฀situation฀arise?฀It฀appears฀that฀Sanguineti฀put฀all฀his฀trust฀ in฀the฀Barbi฀loci,฀as฀the฀basis฀for฀a฀complete฀account฀of฀the฀entire฀tradition.฀However,฀the฀Barbi฀loci฀include฀very฀few฀of฀the฀variants฀identified฀ by฀the฀VBase฀analysis฀as฀evidence฀for฀the฀genetic฀relationship฀of฀Urb฀and฀ Rb.฀Accordingly,฀he฀failed฀to฀see฀this฀relationship,฀and฀this฀led฀directly฀to฀ his฀hypothesis฀of฀the฀unique฀authority฀of฀Urb.฀It฀can฀be฀argued฀that฀this฀ failure฀is฀exacerbated฀because฀it฀appears฀Sanguineti฀did฀not฀use฀all฀the฀ Barbi฀loci:฀he฀used฀only฀four฀of฀the฀six฀lines฀given฀by฀Barbi฀which฀in฀our฀฀ analysis฀evidence฀the฀Urb/Rb฀relationship฀alone,฀and฀also฀used฀only฀four฀of฀฀ the฀nine฀lines฀given฀by฀Barbi฀which฀evidence฀Urb/Rb฀and฀also฀either฀฀ or฀both฀of฀Mart/Triv฀and฀α.฀The฀Barbi฀loci฀contain฀a฀total฀of฀fifteen฀lines฀ which฀evidence฀Urb/Rb,฀either฀on฀its฀own฀(six฀lines)฀or฀in฀combination฀ with฀other฀groups฀(nine฀lines);฀Sanguineti฀uses฀only฀eight฀of฀these.฀One฀ is฀reminded฀that฀key฀parts฀of฀Sanguineti’s฀argument฀rest฀on฀a฀very฀few฀ readings:฀he฀uses฀just฀four฀to฀eliminate฀more฀than฀four฀hundred฀manuscripts฀from฀consideration฀(see฀Shaw’s฀discussion,฀«Introduction฀–฀Sanguineti:฀The฀Tradition»).฀Yet,฀even฀if฀Sanguineti฀had฀used฀all฀the฀Urb/ Rb฀variants฀included฀in฀the฀Barbi฀loci,฀it฀is฀still฀likely฀that฀he฀would฀have฀ reached฀the฀same฀conclusion. Indeed,฀Petrocchi฀warned฀explicitly฀against฀use฀of฀the฀Barbi฀loci฀as฀a฀ base฀for฀the฀textual฀reconstruction฀of฀the฀whole฀tradition.฀In฀his฀«Proposte฀per฀un฀testo-base฀della฀Divina฀Commedia»,฀Filologia฀Romanza,฀ II฀(1955),฀pp.฀337-365,฀he฀argues฀that฀while฀Barbi’s฀hypothesis฀(that฀a฀ genetic฀reconstruction฀of฀the฀tradition฀is฀possible฀on฀the฀basis฀of฀a฀small฀ number฀of฀chosen฀lines)฀was฀plausible฀in฀1890,฀research฀into฀the฀tradi- The฀textual฀tradition฀of฀Dante’s฀Commedia 25 tion฀in฀the฀years฀since฀Barbi’s฀article฀showed฀that฀the฀complexities฀in฀the฀ tradition฀are฀of฀such฀a฀nature฀that฀any฀such฀partial฀analysis฀can฀yield฀only฀ «un฀primo฀e฀generico฀orientamento฀nel฀fitto฀dei฀rapporti฀tra฀i฀codici»:฀ «a฀preliminary฀and฀general฀sense฀of฀direction฀within฀the฀thicket฀of฀relationships฀among฀the฀manuscripts».฀The฀analysis฀in฀this฀article฀suggests฀ that฀Petrocchi฀is฀exactly฀right.฀Analysis฀based฀on฀the฀Barbi฀loci฀shows฀the฀ broadest฀features฀of฀the฀tradition:฀thus฀the฀clearly฀marked฀pairs฀Ash/Ham฀ and฀Mart/Triv.฀But฀it฀fails฀to฀show฀more฀elusive,฀yet฀real,฀relationships,฀ as฀we฀have฀seen฀for฀Urb/Rb.฀Such฀maps฀are฀dangerous.฀As฀we฀noted฀earlier฀,฀Petrocchi฀accordingly฀took฀a฀quite฀different฀direction.฀Rather฀than฀ sample฀a฀few฀readings฀across฀very฀many฀manuscripts,฀as฀Barbi฀suggests,฀ he฀chose฀to฀limit฀the฀number฀of฀manuscripts฀surveyed฀and฀examined฀all฀ the฀readings฀in฀those฀manuscripts.฀The฀Shaw฀edition฀follows฀this฀path,฀฀ narrowing฀the฀number฀of฀manuscripts฀still฀further฀to฀just฀seven฀(including฀one,฀LauSC,฀not฀among฀the฀antica฀vulgata฀manuscripts฀used฀by฀Petrocchi),฀and฀using฀an฀array฀of฀analytic฀techniques฀not฀available฀to฀Petrocchi.฀The฀close฀agreement฀between฀Shaw฀and฀Petrocchi฀on฀almost฀all฀points฀ may฀be฀seen฀as฀a฀vindication฀of฀Petrocchi’s฀method.18 It฀could฀be฀argued฀that฀it฀is฀no฀surprise฀that฀Sanguineti฀failed฀to฀create฀ an฀adequate฀account฀of฀the฀manuscript฀relations฀on฀the฀basis฀of฀the฀Barbi฀ loci.฀One฀might฀question฀(as฀Petrocchi฀did)฀the฀fundamental฀methodology,฀of฀trying฀to฀create฀a฀manuscript฀stemma฀of฀so฀vast฀a฀tradition,฀and฀฀ so฀long฀a฀text,฀on฀a฀sample฀of฀less฀than฀3%฀of฀the฀material.฀One฀might฀also฀฀ point฀to฀weaknesses฀in฀Sanguineti’s฀use฀of฀the฀loci:฀his฀dependence฀on฀a฀ very฀few฀readings฀to฀classify฀very฀many฀manuscripts;฀his฀apparent฀failure฀to฀use฀all฀the฀loci.฀But฀there฀is฀one฀aspect฀of฀this฀comparison฀of฀the฀ Shaw฀results฀with฀the฀Barbi฀loci฀which฀might฀be฀surprising.฀I฀noted฀above฀ that฀Barbi฀designated฀396฀of฀the฀14233฀lines฀of฀the฀Commedia฀as฀most฀ likely฀to฀be฀useful฀for฀tracing฀the฀textual฀families฀to฀the฀original฀(«rintracciare฀le฀famiglie฀dei฀testi฀a฀penna»).฀The฀Shaw/Robinson฀analysis฀identified฀1885฀lines฀as฀containing฀variants฀distinctive฀of฀family฀groupings.฀ One฀might฀expect฀that฀a฀high฀proportion฀of฀Barbi’s฀396฀lines฀would฀be฀฀ among฀the฀1885฀identified฀in฀the฀Shaw฀analysis:฀in฀fact,฀there฀are฀121฀lines฀ in฀common.฀While฀this฀is฀far฀more฀than฀simple฀chance฀might฀predict฀(as฀ explained฀above),฀it฀seems฀lower฀than฀one฀might฀expect.฀Barbi฀has฀a฀nearlegendary฀status฀among฀Italian฀philologists.฀He฀was฀in฀his฀early฀20s฀at฀the฀ ฀But฀not฀on฀all฀points:฀Petrocchi฀argues฀that฀Ham฀is฀a฀descendant฀of฀Ash;฀Shaw฀ shows฀that฀Ash฀and฀Ham฀are฀both฀independently฀descended฀from฀the฀same฀ancestor฀ within฀the฀tradition. 18 26 Peter฀Robinson time฀of฀publication฀of฀the฀loci฀article,฀at฀the฀beginning฀of฀an฀illustrious฀ career฀as฀editor,฀philologist฀and฀Dante฀scholar,฀in฀which฀he฀came฀to฀know฀ the฀business฀of฀editing฀and฀the฀manuscripts฀and฀text฀of฀the฀Commedia฀ as฀few฀have฀ever฀done฀or฀will฀do.฀If฀one฀were฀to฀choose฀any฀one฀person฀ to฀select฀a฀group฀of฀lines฀most฀likely฀to฀be฀useful฀for฀stemmatic฀analysis,฀Barbi฀would฀be฀the฀choice฀of฀many฀scholars.฀Yet,฀less฀than฀a฀third฀of฀ the฀lines฀he฀selected฀appear฀in฀the฀list฀of฀1185฀lines฀identified฀by฀the฀Shaw฀ analysis.฀That฀is:฀some฀two฀thirds฀of฀the฀lines฀he฀chose฀(275฀of฀396)฀turned฀ out฀not฀to฀be฀productive฀for฀Shaw’s฀analysis.฀One฀could฀make฀some฀allowance,฀that฀some฀of฀these฀275฀lines฀might฀be฀revelatory฀of฀relations฀among฀฀ manuscripts฀not฀studied฀by฀Shaw,฀but฀the฀fundamental฀position฀of฀six฀of฀ the฀seven฀Shaw฀manuscripts฀in฀Petrocchi’s฀account฀of฀the฀early฀history฀฀ of฀the฀tradition฀(the฀area฀of฀most฀concern฀to฀Barbi,฀as฀for฀any฀editor)฀make฀ this฀a฀difficult฀argument฀to฀sustain.19 Why,฀then,฀does฀Barbi฀manage฀to฀identify฀so฀few฀lines฀which฀are฀actually฀productive฀for฀phylogenetic฀analysis?฀Barbi’s฀account฀of฀the฀principles฀on฀which฀he฀chose฀these฀lines฀is฀highly฀compressed,฀just฀a฀few฀ sentences.฀He฀remarks฀that฀it฀would฀be฀«inutile฀affatto»฀to฀set฀out฀the฀ reasons฀for฀the฀choice฀verse฀by฀verse,฀and฀states฀the฀general฀principle฀ 19 ฀Petrocchi฀divides฀the฀manuscripts฀of฀the฀antica฀vulgata฀into฀two฀major฀groups:฀α฀ and฀β.฀He฀divides฀β into฀two฀groups,฀d฀and฀e,฀with฀Rb฀and฀Urb฀as฀two฀of฀the฀three฀representatives฀of฀e.฀He฀divides฀α฀into฀three฀groupings:฀a฀b฀c.฀Mart฀and฀Triv฀are฀the฀sole฀representatives฀of฀a;฀Ash฀and฀Ham฀represent฀two฀(in฀Shaw’s฀analysis;฀Petrocchi฀sees฀Ham฀ as฀descended฀from฀Ash)฀of฀some฀five฀branches฀of฀b;฀the฀original฀text฀of฀Mart฀(«M0»฀in฀ Shaw)฀is฀close฀to฀Vat,฀a฀primary฀representative฀of฀Petrocchi’s฀c.฀Thus,฀both฀primary฀and฀ all฀three฀secondary฀branches฀of฀the฀tradition฀defined฀by฀Petrocchi฀are฀represented฀in฀the฀ six฀manuscripts฀Ash฀Ham฀Mart฀Triv฀Urb฀Rb฀studied฀by฀Shaw.฀The฀only฀branch฀of฀Petrocchi’s฀stemma฀(if฀it฀is฀a฀branch)฀not฀represented฀is฀that฀of฀the฀anomalous฀Landino฀manuscript,฀which฀Petrocchi฀places฀as฀a฀blend฀of฀readings฀drawn฀from฀c฀(via฀«c1»)฀and฀β฀(via฀ «d»).฀It฀should฀be฀noted฀that฀Shaw’s฀study฀also฀shows฀that฀the฀Laurenziano฀Santa฀Croce฀ manuscript,฀much฀favoured฀by฀early฀editors฀of฀the฀Commedia฀(Witte฀thought฀it฀the฀ best฀surviving฀manuscript฀and฀usually฀chose฀its฀readings฀ahead฀of฀those฀of฀other฀manuscripts฀for฀his฀1862฀edition;฀Moore฀and฀Casella฀also฀rated฀it฀highly;฀see฀Shaw฀«Introduction»,฀notes฀31฀and฀47)฀and฀chosen฀by฀Sanguineti฀as฀one฀of฀his฀seven,฀is฀actually฀a฀classically฀contaminated฀manuscript,฀combining฀readings฀in฀its฀original฀text฀drawn฀from฀four฀ of฀the฀five฀sub-archetypes฀identified฀by฀Petrocchi฀(that฀is:฀a฀b฀c฀e),฀and฀the฀contamination฀฀ is฀compounded฀by฀the฀corrections฀introduced฀into฀the฀manuscript,฀with฀the฀«c2»฀corrector฀apparently฀introducing฀readings฀from฀another฀witness฀which฀appears฀to฀combine฀readings฀from฀Petrocchi’s฀c฀and฀e฀branches฀(Robinson฀and฀Shaw฀«Phylogenetic฀ Analysis»,฀in฀Shaw฀Commedia.)฀This฀analysis฀of฀LauSC฀accordingly฀confirms฀Petrocchi’s฀argument:฀that฀manuscripts฀written฀after฀1355฀are฀likely฀to฀contain฀such฀mixes฀of฀ readings฀drawn฀from฀multiple฀branches฀as฀to฀be฀of฀no฀value฀towards฀reconstruction฀ of฀the฀earliest฀phases฀of฀the฀tradition. The฀textual฀tradition฀of฀Dante’s฀Commedia 27 guiding฀the฀choice:฀there฀should฀be฀two฀or฀more฀variants฀present,฀each฀ in฀several฀manuscripts.฀Beyond฀this,฀he฀sets฀out฀the฀readings฀which฀he฀ excluded:฀readings฀which฀show฀transcribers฀and฀commentators฀exhibiting฀critical฀judgement;฀and฀those฀where฀separate฀scribes฀are฀highly฀likely฀ to฀introduce฀the฀same฀reading฀for฀reasons฀of฀«eufonia,฀sinonimia,฀correlazione฀grammaticale»,฀and฀notes฀that฀he฀paid฀special฀attention฀to฀the฀ help฀which฀major฀errors฀can฀afford฀towards฀the฀classification฀of฀manuscripts฀(«Abbiamo฀anche฀pensato฀all’aiuto฀che฀porgono฀per฀la฀classificazione฀dei฀manoscritti,฀gli฀errori฀grossolani฀di฀lezione»). We฀may฀compare฀these฀criteria฀with฀some฀of฀the฀readings฀identified฀ by฀the฀Shaw/Robinson฀analysis฀as฀indicative฀of฀family฀relations.฀Here฀ are฀five฀of฀the฀308฀readings฀identified฀as฀likely฀to฀have฀been฀introduced฀฀ by฀the฀common฀ancestor฀of฀Urb/Rb;฀none฀of฀these฀five฀lines฀appear฀ among฀the฀Barbi฀loci.฀ Inf.฀i฀89:฀aiutami฀da฀lei,฀famoso฀saggio, famoso฀e฀saggio฀ ฀ LauSC฀Rb฀Urb฀FS famoso฀saggio฀ ฀ ฀ Ash฀Ham฀Mart฀Triv฀PET Inf.฀ii฀71:฀vegno฀del฀loco฀ove฀tornar฀disio;฀ di฀ ฀ ฀ ฀ LauSC฀Rb฀Urb฀FS del฀ ฀ ฀ ฀ Ash฀Mart฀Triv฀PET dal฀ ฀ ฀ ฀ Ham Inf.฀ii฀110:฀a฀far฀lor฀pro฀o฀a฀fuggir฀lor฀danno, pro฀e฀a฀ ฀ ฀ Mart-orig฀Rb฀Urb฀FS pro฀ne฀a฀ ฀ ฀ Ash฀LauSC-c2฀Triv prode฀et฀a฀ ฀ ฀ Ham pro฀[..]฀a฀ ฀ ฀ LauSC-orig pro฀o฀a฀ ฀ ฀ Mart-c2฀PET Inf.฀iii฀3:฀per฀me฀si฀va฀tra฀la฀perduta฀gente.฀ ne฀la฀ ฀ ฀ ฀ Rb฀Urb฀FS tra฀la฀ ฀ ฀ ฀ Ash฀Ham฀LauSC฀Mart฀Triv-c1฀PET tra฀ ฀ ฀ ฀ Triv-orig Inf.฀iii฀22:฀Quivi฀sospiri,฀pianti฀e฀alti฀guai฀ altri฀ ฀ ฀ ฀ Ash-orig฀Rb฀Urb alti฀ ฀ ฀ ฀ Ash-c2฀Ham฀LauSC฀Mart฀Triv฀FS฀PET 28 Peter฀Robinson We฀can฀see฀why฀Barbi฀would฀not฀have฀chosen฀these฀readings.฀All฀have฀ the฀slight฀character฀of฀variants฀which฀might฀have฀appeared฀readily฀at฀any฀ point฀of฀the฀tradition,฀and฀hence฀are฀likely฀(to฀use฀the฀term฀employed฀ by฀Petrocchi฀and฀Shaw)฀to฀be฀«polygenetic»;฀in฀Shaw’s฀definition฀«an฀ error฀liable฀to฀arise฀independently฀in฀independent฀manuscripts».฀One฀ can฀imagine฀that฀no฀textual฀scholar,฀educated฀to฀reject฀errors฀likely฀to฀be฀ polygenetic฀as฀indicative฀of฀anything,฀would฀select฀these฀as฀the฀basis฀for฀ any฀kind฀of฀argument:฀indeed,฀only฀one,฀tra฀la/ne฀la฀Inf฀iii.฀3,฀is฀used฀by฀ Shaw฀in฀her฀account฀of฀Rb/Urb.฀Yet฀the฀Shaw/Robinson฀analysis฀of฀the฀ distribution฀of฀the฀variants฀across฀94779฀distinct฀sites฀of฀readings฀(usually,฀one฀word,฀but฀often฀a฀phrase)฀over฀the฀14233฀lines฀of฀the฀Commedia฀suggests฀that฀the฀appearance฀of฀these฀five฀readings฀in฀Urb/Rb฀results฀ from฀their฀introduction฀into฀an฀ancestor฀copy฀shared฀by฀Urb฀and฀Rb.฀ That฀two฀of฀these฀five฀(Inf.฀i฀89฀and฀ii฀71)฀appear฀also฀in฀LauSC฀is฀likely฀ to฀be฀the฀result฀of฀contamination฀in฀that฀manuscript฀(see฀footnote฀20).฀ This฀leaves฀only฀Inf.฀ii฀110฀«pro฀e฀a»฀in฀the฀Aldine฀text฀which฀is฀the฀base฀ of฀the฀Martini฀collation฀(itself฀derived฀from฀Vat)฀and฀Inf.฀iii฀22฀«altri»฀in฀ the฀original฀text฀of฀Ash฀as฀likely฀polygenetic฀errors,฀if฀not฀deriving฀from฀ contamination. To฀sum฀up:฀a฀high฀proportion฀of฀the฀1885฀lines฀identified฀by฀the฀Shaw/ Robinson฀analysis฀as฀indicative฀of฀manuscript฀relations฀contain฀variants฀which฀would฀not฀satisfy฀Barbi’s฀criteria฀as฀being฀unlikely฀a฀priori฀ to฀have฀arisen฀in฀independent฀copies.฀Yet,฀these฀same฀variants฀are฀useful฀ for฀our฀analysis,฀precisely฀because฀as฀a฀group฀they฀did฀not฀arise฀in฀independent฀copies.฀The฀key฀phrase฀here฀is฀«as฀a฀group»:฀individual฀variants฀among฀them฀(say,฀among฀the฀308฀Urb/Rb฀variants)฀did฀appear฀in฀ other฀manuscripts,฀sometimes฀perhaps฀by฀contamination,฀sometimes฀ by฀polygenesis.฀But฀other฀manuscripts฀outside฀each฀group฀appear฀alongside฀the฀members฀of฀the฀group฀in฀a฀scattered,฀random฀fashion,฀as฀we฀see฀ in฀these฀five฀variants:฀Urb/Rb฀agree฀in฀all฀five,฀once฀with฀no฀other฀manuscript฀(iii฀3),฀twice฀with฀LauSC฀(I฀89,฀ii฀71),฀once฀with฀the฀original฀text฀ of฀Mart฀(ii฀110),฀once฀with฀the฀original฀text฀of฀Ash฀(iii฀22).฀This฀is฀exactly฀฀ as฀we฀should฀expect฀in฀a฀real฀manuscript฀tradition:฀that฀the฀scribe฀of฀฀ the฀common฀ancestor฀of฀Urb/Rb฀introduced฀some฀308฀readings฀into฀the฀฀ copy,฀which฀then฀descended฀into฀Urb/Rb.฀Of฀those฀readings,฀we฀should฀ expect฀a฀few฀of฀them฀to฀appear฀in฀other฀manuscripts,฀by฀contamina-฀ tion฀ or฀ by฀ polygenesis:฀ thus,฀ we฀ find฀ 40฀ of฀ the฀ 308฀ in฀ Ash฀ (also,฀ 2฀ Ash-orig),฀11฀in฀LauSC฀(11฀LauSC-orig,฀4฀LauSC-c1,฀5฀LauSC-c2,฀4฀in฀ Mart฀(Mart-orig฀13),฀9฀in฀Triv,฀46฀in฀Ham,฀while฀all฀308฀appear฀in฀both฀฀ Urb฀and฀Rb. The฀textual฀tradition฀of฀Dante’s฀Commedia 29 This฀suggests฀that฀one฀cannot฀determine฀whether฀a฀particular฀variant฀is฀monogenetic฀(as,฀ideally,฀we฀would฀want฀all฀variants฀to฀be)฀or฀ polygenetic฀purely฀on฀the฀basis฀of฀the฀variant฀itself,฀as฀Barbi฀suggests฀ we฀do.฀One฀must฀also฀take฀into฀account฀both฀the฀actual฀distribution฀of฀ manuscripts฀which฀have฀the฀variant,฀and฀the฀consistency฀of฀this฀manuscript฀distribution฀across฀many฀other฀variants.฀When฀we฀see฀the฀pair฀ Urb/Rb฀appear฀over฀and฀over,฀sometimes฀on฀their฀own,฀sometimes฀with฀ a฀varying฀cast฀of฀other฀manuscripts,฀then฀we฀are฀entitled฀to฀presume฀that฀ this฀consistency฀of฀distribution฀itself฀may฀suggest฀a฀genetic฀relationship฀ between฀the฀manuscripts.฀Although฀both฀Petrocchi฀and฀Shaw฀pay฀close฀ attention฀to฀variants฀which฀they฀regard฀as฀likely฀to฀be฀monogenetic,฀on฀ the฀basis฀of฀the฀variant฀itself฀(as฀Barbi฀does),฀they฀also฀are฀aware฀of฀the฀ significance฀of฀a฀consistent฀pattern฀of฀agreements,฀found฀in฀variant฀after฀ variant.฀Shaw฀cites฀Petrocchi’s฀argument฀that฀the฀trio฀Urb/Rb/Mad฀have฀ a฀closeness฀«nettamente฀superiore฀per฀foltezza฀di฀statistica฀e฀natura฀dei฀ casi฀a฀qualsiasi฀affinità฀tra฀il฀testimone฀urbinate฀e฀gli฀altri»฀(Introduction,฀ citing฀Petrocchi฀Introduzione฀334-389),฀and฀points฀out฀the฀significance฀ of฀the฀phrase฀«foltezza฀di฀statistica»:฀«statistical฀density».฀As฀I฀noted฀earlier,฀Shaw฀expands฀on฀Petrocchi’s฀perception,฀as฀she฀discusses฀a฀number฀ of฀readings฀supporting฀the฀genetic฀relationship฀of฀Urb/Rb฀which,฀she฀ concedes,฀are฀each฀individually฀possibly฀polygenetic,฀but฀which฀taken฀ together฀evidence฀a฀genetic฀relationship.฀Thus฀her฀conclusion: These฀are฀of฀course฀small฀errors,฀possibly฀polygenetic.฀Each฀one฀in฀itself฀would฀ count฀for฀very฀little.฀But฀it฀is฀the฀cumulative฀picture฀built฀up฀of฀a฀whole฀series฀ of฀such฀small฀errors฀right฀across฀the฀text฀and฀shared฀with฀no฀or฀very฀few฀other฀ manuscripts฀which฀is฀significant฀and฀suggests฀a฀common฀exemplar. Seen฀this฀way,฀Petrocchi฀and฀Shaw’s฀use฀of฀«foltezza฀di฀statistica»฀alongside฀consideration฀of฀the฀monogenetic฀or฀polygenetic฀character฀of฀each฀ variant฀is฀an฀extension฀of฀Barbi’s฀method,฀not฀a฀repudiation฀of฀it.฀Barbi,฀ Petrocchi฀and฀Shaw฀all฀agree฀that฀it฀is฀unsafe฀to฀rely฀on฀certain฀individual฀ variants,฀considered฀on฀their฀own,฀as฀indicative฀of฀relations฀across฀the฀whole฀ tradition.฀Indeed,฀Shaw฀criticizes฀Sanguineti฀(as฀I฀note฀above)฀for฀basing฀ his฀analysis฀on฀a฀few฀variants.฀However,฀both฀Petrocchi฀and฀Shaw฀are฀able฀ to฀use฀«foltezza฀di฀statistica»฀because฀they฀do฀take฀into฀account฀every฀variant,฀every฀reading,฀in฀the฀manuscripts฀they฀survey฀–฀an฀approach฀only฀possible฀because฀they฀choose฀not฀to฀base฀their฀analysis฀on฀a฀sample฀of฀readings฀ from฀a฀great฀many฀manuscripts฀(the฀Barbi฀loci฀methodology,฀followed฀by฀ Sanguineti)฀but฀on฀all฀the฀readings฀from฀a฀small฀number฀of฀manuscripts. 30 Peter฀Robinson Barbi;฀Petrocchi;฀Shaw:฀more฀than฀a฀century฀of฀Dante฀textual฀scholarship,฀a฀century฀which฀saw฀Italian฀textual฀scholarship฀absorb,฀and฀then฀ reshape฀itself฀in฀reaction฀to,฀the฀impact฀of฀Lachmannian฀methods฀and฀ the฀opposition฀to฀these฀methods฀led฀by฀Bédier฀and฀others;฀a฀scholarly฀tradition฀which฀is฀now฀reckoning฀with฀the฀first฀impact฀of฀another฀revolution,฀as฀digital฀methods฀open฀up฀new฀ways฀of฀approaching฀large฀textual฀traditions.฀We฀should฀mention฀two฀more฀names:฀Giogio฀Pasquali,฀ whose฀Storia฀della฀tradizione฀e฀critica฀del฀testo฀(Florence,฀Le฀Monnier฀1934)฀฀ repudiated฀the฀rigid฀methodology฀outlined฀in฀Maas’s฀Textkritik฀(Leipzig,฀฀ Teubner฀1927),฀without฀rejecting,฀as฀Bédier฀did,฀the฀possibility฀of฀making฀ sense฀of฀large฀and฀complex฀traditions;฀and฀Gianfranco฀Contini.฀In฀1935,฀ when฀Contini฀reviewed฀Pasquali’s฀book,฀he฀was฀around฀the฀same฀age฀as฀ Barbi฀when฀he฀was฀compiling฀the฀list฀of฀loci.฀Barbi’s฀own฀La฀nuova฀filologia฀e฀l’edizione฀dei฀nostri฀scrittori฀da฀Dante฀al฀Manzoni฀(Florence:฀Sansoni฀1938)฀follows฀Pasquali฀in฀advocating฀what฀Pugliatti฀describes฀as฀a฀ «virtuous฀middle฀way»฀between฀Lachmann฀and฀Bédier.20฀Contini฀follows฀ this฀«middle฀way»,฀asserting฀throughout฀his฀career฀that฀intelligent฀use฀of฀ every฀available฀method,฀including฀stemmatics฀where฀appropriate,฀joined฀ with฀deep฀knowledge฀of฀the฀manuscripts฀and฀the฀texts฀they฀carry,฀would฀ permit฀a฀useful฀understanding฀of฀whole฀manuscript฀traditions.฀Shaw฀ studied฀with฀Contini,฀first฀as฀a฀post-graduate฀from฀1962฀to฀1964,฀and฀then฀ from฀1966฀as฀she฀wrote฀her฀Dott.฀Lett.฀dissertation,฀an฀edition฀of฀the฀earliest฀Italian฀translation฀of฀the฀Monarchia,฀with฀Contini฀as฀her฀supervisor,฀ thus฀bracketing฀the฀appearance฀of฀Petrocchi’s฀great฀edition฀in฀1966.฀With฀ Contini’s฀encouragement,฀she฀then฀edited฀Marsilio฀Ficino’s฀translation฀of฀ the฀Monarchia.฀Again฀with฀Contini’s฀encouragement,฀in฀1981฀she฀wrote฀ an฀article฀critical฀of฀some฀aspects฀of฀Ricci’s฀edition฀of฀the฀Monarchia.21฀ After฀the฀article฀was฀published,฀Contini฀suggested฀to฀Shaw฀that฀she฀undertake฀an฀edition฀of฀the฀Monarchia.฀Shaw฀hesitated฀at฀first:฀she฀was฀aware฀ that฀a฀factor฀in฀the฀problems฀of฀Ricci’s฀edition฀of฀the฀Monarchia฀was฀the฀ amount฀of฀data฀generated฀by฀a฀text฀in฀many฀versions,฀and฀Ricci’s฀inability฀ to฀manage฀all฀this฀data.฀It฀was฀only฀when฀she฀undertook฀an฀introductory฀ course฀in฀computing฀at฀the฀Cambridge฀University฀Computing฀Centre฀in฀ October฀1985฀and฀saw฀the฀possibility฀of฀using฀a฀computer฀to฀record฀and฀ 20 ฀Paola฀Pugliatti,฀«Textual฀Perspectives฀in฀Italy:฀from฀Pasquali’s฀Historicism฀to฀the฀ Challenge฀of฀“Variantistica”฀(and฀Beyond)»,฀Text,฀11฀(1998),฀pp.฀155-188฀[p.฀162]. 21 ฀Shaw’s฀doctoral฀dissertation฀was฀published฀as฀«Il฀volgarizzamento฀inedito฀della฀ Monarchia»,฀Studi฀danteschi,฀XLVII฀(1970),฀pp.฀59-224;฀the฀edition฀of฀the฀Ficino฀translation฀as฀«La฀versione฀ficiniana฀della฀Monarchia»,฀Studi฀danteschi,฀LI฀(1978),฀pp.฀289-407;฀the฀ article฀on฀Ricci฀in฀«Sul฀testo฀della฀Monarchia»,฀Studi฀danteschi,฀LIII฀(1981),฀pp.฀187-217.฀ The฀textual฀tradition฀of฀Dante’s฀Commedia 31 explore฀all฀the฀data฀generated฀by฀full฀transcripts฀of฀each฀manuscript฀that฀ she฀committed฀herself฀to฀making฀the฀edition฀–฀and,฀to฀using฀computer฀ methods฀to฀do฀so.22฀ Shaw฀discussed฀this฀new฀methodology฀with฀Contini,฀and฀he฀encouraged฀her฀in฀this฀path.฀She฀described฀her฀work฀in฀a฀paper฀at฀the฀conference฀in฀Florence฀in฀1988,฀convened฀to฀celebrate฀the฀centenary฀of฀the฀ founding฀of฀the฀Società฀Dantesca฀Italiana฀in฀1888.23฀Her฀decision,฀to฀base฀฀ the฀edition฀on฀full-text฀transcripts฀of฀the฀manuscripts฀looks฀back฀to฀฀ Petrocchi,฀just฀as฀her฀decision฀that฀the฀transcripts฀be฀made฀in฀machinereadable฀form฀looks฀forward฀to฀the฀development฀of฀computer-based฀ systems฀for฀comparison฀and฀analysis.฀In฀1991,฀Shaw฀had฀completed฀the฀ first฀transcripts฀of฀the฀Monarchia฀manuscripts฀and฀was฀exploring฀how฀ these฀transcripts฀might฀be฀used฀to฀make฀an฀edition.24฀I฀was฀starting฀work฀ on฀what฀became฀the฀Collate฀suite฀of฀software฀tools฀for฀creating฀editions฀ from฀multiple฀transcripts฀of฀works,฀and฀also฀discovering฀how฀tools฀from฀ evolutionary฀biology฀could฀be฀used฀to฀advance฀understanding฀of฀large฀ textual฀traditions.฀Shaw฀came฀to฀see฀me฀(it฀is฀a฀fortunate฀chance฀that฀her฀ dentist฀was฀based฀in฀Oxford,฀where฀I฀was฀then฀working),฀and฀this฀began฀ a฀two-decade฀collaboration฀which฀issued฀in฀her฀editions฀of฀the฀Monarchia฀and฀Commedia.฀ Barbi,฀Pasquali,฀Petrocchi,฀Contini,฀Shaw.฀These฀scholars฀are฀joined฀ in฀a฀common฀belief:฀sense฀can฀be฀made฀of฀large฀manuscript฀traditions฀ using฀good฀method฀and฀expert฀philological฀knowledge,฀and฀useful฀edi฀As฀an฀index฀of฀the฀difficulty฀of฀this฀work,฀in฀those฀days฀before฀computers฀became฀ portable฀and฀personal:฀Shaw฀transcribed฀the฀manuscripts฀onto฀paper฀from฀a฀microfilm฀ reader฀at฀the฀University฀library;฀then฀carried฀the฀transcripts฀across฀the฀Cam฀to฀the฀computing฀centre,฀when฀she฀entered฀them฀into฀the฀Cambridge฀mainframe฀using฀the฀«Zed»฀ line฀editor. 23 ฀Published,฀including฀some฀printouts฀of฀her฀transcripts,฀in฀«Per฀un฀nuovo฀testo฀critico฀ della฀Monarchia»,฀in฀La฀Società฀Dantesca฀Italiana฀1888-1988.฀Atti฀del฀Convegno฀Internazionale฀di฀Studi฀Danteschi฀tenuto฀a฀Firenze฀nel฀novembre฀1988,฀Firenze,฀1995,฀pp.฀435-444. 24 ฀Other฀scholars฀at฀this฀time฀were฀coming฀to฀the฀view฀that฀full-text฀computer-readable฀ transcripts฀ were฀ the฀ way฀ forward฀ for฀ textual฀ scholarship฀ applied฀ to฀ large฀ manuscript฀ traditions.฀ In฀ particular,฀ Francesco฀ Mazzoni,฀ Contini’s฀ successor฀ as฀ President฀฀ of฀the฀Società฀Dantesca฀Italiana,฀encouraged฀his฀students฀to฀make฀full-text฀transcripts฀of฀฀ Commedia฀manuscripts,฀and฀developed฀a฀set฀of฀transcription฀guidelines฀for฀this฀work.฀ These฀«principi฀di฀trascrizione»฀may฀be฀found฀at฀http://www.danteonline.it/italiano/ risorse.htm฀(with฀an฀English฀translation฀by฀Shaw).฀Transcriptions฀of฀several฀manuscripts฀are฀available฀on฀the฀danteonline฀site.฀Mazzoni฀himself,฀and฀other฀associates฀of฀ the฀SDI฀(particularly฀Paula฀Laurella),฀were฀extremely฀supportive฀of฀Shaw’s฀continuing฀ work฀on฀the฀Monarchia฀(her฀print฀edition฀was฀published฀in฀the฀SDI’s฀Edizione฀Nazionale฀ series;฀the฀SDI฀co-published฀the฀DVD฀version)฀and฀on฀the฀Commedia. 22 32 Peter฀Robinson tions฀of฀works฀can฀be฀founded฀on฀that฀sense.฀Full-text฀machine฀readable฀transcripts฀of฀whole฀manuscripts,฀passed฀through฀an฀array฀of฀computer-based฀methods,฀are฀now฀part฀of฀a฀philologist’s฀armory,฀to฀be฀used฀ in฀conjunction฀with฀(and฀not฀in฀opposition฀to)฀knowledge฀of฀the฀manuscripts,฀of฀the฀text฀itself฀and฀its฀contexts,฀and฀with฀editorial฀intuition.฀ This฀seems฀a฀useful฀starting-point฀for฀further฀explorations฀into฀the฀Com-฀ media฀tradition. University฀of฀Saskatchewan