Wednesday, September 30, 2009

Oxfordians and the 1604 Question by Peter Farey

The Earl of Oxford is known to have died in 1604, when according to virtually all Shakespearean scholars at least a quarter of Shakespeare's plays were yet to be written. Oxfordians - who unlike Marlovians generally accept the date upon which their candidate is supposed to have died - imagine that they can overcome this problem by claiming that these plays must have all been written before then, and that the scholars have simply shifted them into the period c.1592-1613 because that corresponds with what would have been the writing career of their author, William Shakespeare of Stratford. Oxfordian chronologies would have him writing everything some ten or more years earlier.

Unfortunately for them, however, it is not quite as simple as that. How much the blank verse used in English drama changed over the years cannot simply be ignored in this way. Let us look at one such change - the gradual but steady move away from a procession of end-stopped lines in regular iambic pentameter.

The opposite of an end-stopped line, an open line - or a run-on line, or enjambment - is one in which the meaning flows on to the next line without punctuation. For example, see Prospero's famous lines in Act IV scene 1 of The Tempest:
Our revels now are ended: these our actors,
As I foretold you, were all spirits, and
Are melted into air, into thin air.
It would simply not be possible to put any sort of punctuation at the end of the second line without wrecking the sense. That's an open line. And a feminine ending? The rhythm of Shakespeare's blank verse is the iambic (di-dum) pentameter (repeated 5 times). This is how lines two and three scan, but not line one, which has acquired an extra "di" syllable at the end. That's a feminine ending.

In a post to SHAKSPER ("The Global Electronic Shakespeare Conference") on March 2, 1996, David Kathman wrote: "The use of feminine endings in blank verse increased steadily among English poets in general between the 1580s and the 1620s, just as the use of enjambment increased steadily in the same period; Shakespeare followed both of these trends"; and as Ward Elliott and Robert Valenza tell us: "The percentages of both indicators tripled over Shakespeare's lifetime."1

Here, using Elliott and Valenza's figures, are details of plays by Beaumont, Chapman, Daniel, Dekker, Fletcher, Greene, Heywood, Jonson, Kyd, Herbert (Mary), Lyly (that lone one at the bottom), Marlowe, Middleton, Munday, Nashe, Peele and Porter.2As we can see, even though the data come from many different playwrights, there is a clearly discernable trend, as indicated by the computer-generated trendline. Note in particular that before 1600 only two plays exceeded a usage rate of 35, whereas after 1600 only two plays did not exceed that level. By "usage rate," we mean the number of open lines plus feminine endings there are on average per hundred lines of verse.

Although there is no generally agreed chronology for the works of Shakespeare, the differences between the dates suggested by various scholars are relatively minor. Given that the figures being used in these graphs are those of Elliott and Valenza, it seems appropriate to use their dates too.

Here is a graph in which the usage rate for each Shakespeare play according to Elliott and Valenza is plotted against the date they suggest for it. It is worth bearing in mind that their intention in assembling this information had nothing to do with our reason for using it here. Even without the trendline, one can see how Shakespeare's figures increase over the years as David Kathman reported - in fact at a quite extraordinarily constant rate. What should be pointed out is that every one of the plays usually given a date after the death of the Earl of Oxford in 1604 has a usage rate greater than 45. Every other Shakespeare play (i.e. 1604 or earlier) except Measure for Measure (50) and All's Well That Ends Well (51) has a rate of less than that.

Although these dates are the ones given by Elliott and Valenza, the trend would have been no less evident if we had used the dates published by any other Shakespearean scholar. No matter which of the various orthodox chronologies one chooses, there is a highly significant trend by which these figures increase over the years. In other words we can be very confident that most if not all of those plays given a date after the death of the Earl of Oxford were indeed written after those given a date before it.

Some Oxfordians argue that the whole timetable could be moved back ten years or so, which would result in precisely the same trend in Shakespeare's works, and simply mean that the other writers took longer than we thought to catch on to the changing approach. Unfortunately for this argument, however, it would mean them bumping straight into the buffer which is Francis Meres.

In 1598 Francis Meres published his Palladis Tamia which, had it not been for its relevance to Shakespeare, might have been largely forgotten long before now. In this he gave a list of plays by Shakespeare, which clearly indicated that they must have been performed by that year. He said that Shakespeare was the "most excellent" English dramatist for both comedy and tragedy, and gave as examples of the former The Two Gentlemen of Verona, The Comedy of Errors, Love's Labour's Lost, A Midsummer Night's Dream, The Merchant of Venice and the now untraceable Love's Labour's Won. For tragedy he cited Richard II, Richard III, Henry IV, King John, Titus Andronicus, and Romeo and Juliet.

As Oxfordians are quick to point out, however, it's not a complete list of the plays Shakespeare had written by then, since it is fairly clear that the three parts of Henry VI and The Taming of the Shrew - although not necessarily with the same titles - had also been performed before 1598. So is there a specific reason for those plays to have been omitted which would not apply to any other? Yes, there certainly is.

This is what Edward Burns says in his Arden (3rd series) edition of 1 Henry VI (p.8), describing the first page of the theatre owner Philip Henslowe's accounts: "What follows is a day-to-day calendar of the plays performed that year at the Rose theatre by a company under the patronage of Lord Strange - hence 'my lord stranges mene'. About a third of the way down the page a line reads 'ne - Rd at harey the vj the 3 of marche 1591...iijli xvjs 8d'." "Harey the vj" is generally accepted as being 1 Henry VI, and there is no record of it ever being performed by any company other than Lord Strange's Men.

In his Arden 3 edition of 2 Henry VI (p.121), Ronald Knowles says: "...it seems probable that, by 1591, 2 and 3 Henry VI had been performed successfully, presumably by Lord Strange's company, since 'Pembroke's Men' appears on the title page of The True Tragedy [the 1595 quarto of 3 Henry VI] and this company is generally considered an offshoot of Strange's." 2 Henry VI had been called "The First Part of the Contention between the two great houses of York and Lancaster." That it was the "first part" may be taken to indicate that both plays were owned and performed by the same company.

All three of the Henry VI plays therefore seem to have been the property of the Strange/Pembroke companies and there is no record of their being associated in any way with the Lord Chamberlain's company. The reason for their omission from Meres's list is therefore fairly clear - he didn't know that they were by Shakespeare, resident playwright for the Lord Chamberlain's Men.

Similarly, The Taming of the Shrew had a quarto version called The Taming of A Shrew and, as Brian Morris (Arden 2, p.45) says, its title-page states, "As it was sundry times acted by the Right honorable, the Earle of Pembroke his servants." The next time we hear of it is in the First Folio of 1623.

Titus Andronicus was similar to this, although Meres did include it in his list. What we find in this case, however - despite evidence of it having passed through the hands of Strange's, Pembroke's and even Sussex's Men - is that when Q2 appeared in 1600 (only two years after Meres's list) it had added the Lord Chamberlain to the noblemen whose servants had played it, most probably before 1598.

There is no evidence of any of the rest of the plays on Meres's list having been played by companies other than the Lord Chamberlain's. It's been suggested that Richard III might have originated with the Strange/Pembroke group, but the fact that the 1598 quarto is ascribed to Shakespeare - and that there was a bawdy 1602 joke about Burbage playing Richard - makes it fairly certain that it was in the Lord Chamberlain's Men's repertoire by the time Meres mentioned it.

So Meres's list seems to consist only of those plays said to have been by Shakespeare which were in the Lord Chamberlain's Men's repertory by 1598. Most scholars therefore quite reasonably claim that the first plays written by Shakespeare were those Meres listed, plus the other four, and that any other plays of his made their first appearance after this date.

In considering whether this claim is justified it is interesting to see how much the two techniques are used in those first 15 plays of Shakespeare - the 11 on Meres's list for which we have data plus the four missing ones - when compared with the rest of his plays. If we do so we find that (with the exception of 1 Henry IV and The Merchant of Venice) every one of them has a usage rate lower than 35. In contrast, every other play of his (with the exception of The Merry Wives of Windsor which has hardly any verse to speak of) has a rate exceeding 35. This offers massive support for the Shakespearean scholars' claim.

That this matches the trend observed among Shakespeare's contemporaries is reinforced by the fact that in the first graph the contemporaries' trendline is only two points away from his level in 1598. By 1604, however, he is ahead by some ten points on average, and by 1613 that difference between them has grown to twenty per hundred lines of verse. Part of what made Shakespeare's verse so much greater than the rest was this appreciation of just how much freedom the two techniques offered.

To sum up:

1) Whether we are talking about Elizabethan or Jacobean playwrights in general or about Shakespeare in particular, there is a very clear trend in which the use of open lines and feminine endings tended to increase throughout Shakespeare's lifetime.

2) All of the orthodox Shakespearean chronologies show ten or eleven plays written after Oxford's death.

3) Every one of the eleven plays considered here has a frequency of open lines plus feminine endings which is more than that of any other play ascribed to Shakespeare bar two. The median (midpoint) value of the figures for these plays is 64 per hundred lines of verse.

4) Of the plays listed by Francis Meres as having been written by 1598 - plus those missing from his list for which there is evidence of earlier performance by the Strange/Pembroke companies - every one of them other than The Merchant of Venice and 1 Henry IV has a frequency of open lines plus feminine endings which is less than any of the rest of the plays attributed to him other than The Merry Wives of Windsor. The median value for all of them is only 26 per hundred lines of verse.

5) No matter which of the various orthodox chronologies one chooses, there is a highly significant trend by which these figures increase over the years. However, as the reasoning explained in Gary Taylor's chronology shows,3 the high correlation of the figures with the dates is not due to the latter being based upon the former, which plays little or no part in how the dates are arrived at. So the clear trend adds considerable support to the orthodox datings being substantially correct.

6) This leaves Oxfordians with the following options.

• Find reasons for shifting every one of those ten or eleven plays back before 1604, and pretend that the resulting demolition of the trend doesn't matter.

• Shift the whole canon back the requisite amount in time, which would retain the trend, but then explain why Meres had inexplicably omitted most of Shakespeare's finest plays from his list.

• Create an entirely new chronology for the post-1598 plays which retains the trend, but squeezes all 23 of them into the six-year period between 1598 and 1604. Without any justification for the "new" dates being based either upon internal or external evidence, however, this would of course be cheating.

• Hope that nobody notices and that it will all just go away, or

• Think it possible that they just might be backing the wrong candidate?

Peter Farey

© Peter Farey, September 2009
  Emmerich Anonymous Shakespeare
Peter Farey, a founding member of the International Marlowe-Shakespeare Society, is the 2007 recipient of the Calvin & Rose G. Hoffman Prize, administered annually by The King's School in Canterbury for a "distinguished publication on Christopher Marlowe." He was also a founding member (with Derek Jacobi) of the UK's National Youth Theatre. Click here to reach Peter's website.

Are Oxfordians backing the wrong candidate? Click here for Peter's fascinating analysis of how Marlowe compares with de Vere on Shakespeare authorship criteria.

1Elliott, Ward and Robert Valenza. "And Then There Were None: Winnowing the Shakespeare Claimants." Computers and the Humanities 30, 1996. p.198.
2Ibid. pp.211-211. All of their counting was done by computer, and the authors acknowledge (p.215) that it is not as accurate as manual counting for feminine endings. For the present exercise, however, this is of less importance than having the same method used throughout.
3Taylor, Gary. "The Canon and Chronology of Shakespeare's Plays," in Stanley Wells, Gary Taylor et. al., William Shakespeare, A Textual Companion (1987, 1997). pp.81-2.

Click here for the blog's home page and recent content.

25 comments:

Achitophel said...

Not a good day for Oxfordians, if I may be so frank.

Lee said...

I dunno Achitophel, it's a pretty good day for this Oxfordian.

So what happens btw if you try dating the chronology of the sonnets via this same criteria of feminine endings and use of enjambment? If it's so effective, why hasn't it been done to any acclaim?

Also wouldn't there be a relationship between the subject matter of the play and the percentage point reached? If a royal subject is employed then Shakespeare used a lot more blank verse.

Also the gravity of the situation affects the type of blank verse he used. So wouldn't the subject chosen and ratio of comedy to tragedy affect the data, and hasn't this been pretty much ignored in the study?

So I guess the last possibility Oxfordians could arrive at is the whole test was performed incompetently and is therefore meaningless?

Daryl Pinksen said...

Devastating hit to the Oxford case, Peter.

Lee's response is the only one available to the Oxfordians: insist that the conclusions are meaningless, insist that the data was assembled by incompetent researchers, insist that the science was bad.

None of these is true. If genre were as strong an influence as the date, then the graph would not display a neat linearity, it would be a scattered mess. That the data does show a clear trend is proof that date is the strongest determining factor, and genre much less significant.

Ignoring the serious injury Peter has done to the Oxfordian case is akin to the Black Knight from Monty Python's Holy Grail insisting "It's just a flesh wound!"

Peter Farey said...

Thanks Daryl, that illustrates it beautifully.

Lee asks whether this technique has been applied to the Sonnets. No, but I have analysed them in a similar way based on the frequency with which certain common words are used. With some exceptions (the "marriage" and "dark lady" sonnets) it seemed to indicate that Thorpe published them more or less in their original order. See the essay Shakespeare's Sonnet Sequence on my website.

As for the effect of subject matter, genre, and all sorts of other factors on the results, we see this in the variations from the trend line. The average (mean) of the results for all of the Shakespeare plays is in fact 42 (to the nearest), and each play differs from that to a certain extent. It is possible to calculate how much of the overall deviation from that mean is due to those sort of things, and how much is due to the trend. In the case of Shakespeare it works out that an astonishing 89% of the deviation is due to the trend, and only 11% to such other factors. In other words, as Daryl said, date is the strongest determining factor by far.

Why not "to greater acclaim"? Because I am a non-Stratfordian writing for other non-Stratfordians. You work it out!

Peter Farey

Luanna said...

This is a brilliant and excellent piece! I always enjoy reading Peter Farey's writing because he relys on logic and hard evidence rather than trying to prove what he wishes were true through faulty reasonining, as so many anti-Stratfordians (and, for that matter, Stratfordians) do.

SeattleSlew said...

FANTASTIC!

PatrickJ said...

This is very impressive--had to read it a few times since I haven't thought of feminine endings in years!

MariaVLentini said...

I'm fairly new to this whole issue, but I definitely can't be an Oxfordian based on some of the things I have read on this very nice blog. I am very impressed by Mr. Farey's articles.

Santo said...

Dear Lee,

Rather than smugly (and weakly) challenging the test, how about challenging the facts? Or is it that you also read Farey's "Wrong Candidate" and became very agitated with how weak your candidate really is?

Dave Herber said...

How does the latest claim by Sir Brian Vickers affect the Marlovian Shakespeare story?

http://blogs.usatoday.com/ondeadline/2009/10/a-program-used-to-detect-plagiarism-indicates-that-william-shakespeare-was-at-least-a-co-author-of-an-unattributed-play-abou.html

My take is simple, if Shakespeare were indeed Marlowe wouldn't the collaboration of Kyd and Marlowe be almost a given considering their close friendship. This really helps the Marlovian cause and I would like to see some more on Vickers' research.

Peter Farey said...

Dave Herber said...
How does the latest claim by Sir Brian Vickers affect the Marlovian Shakespeare story?

Way back before either the internet or the British Library played any part in my life, I had great difficulty in getting hold of a copy of Edward III, but eventually found one included in a floppy disk (remember them?) of Marlowe's collected works I had obtained from the Oxford Computing Service. We do perhaps need to recall that for many reasons this play or parts of it have often been attributed to Marlowe in the past.

Like Dave, I would like to see the methodology and results as related to Shakespeare and Kyd, and satisfy myself of their validity. Having done this, however, I would then invite Sir Brian to apply the same method in comparing Marlowe's later plays with the "Shakespeare" parts of Edward III. I can have no idea what the results might be, of course, but the degree of similarity might offer an interesting new light on whether the Marlovian theory is right or not!

Peter Farey

Peter Farey said...

Two weeks have passed, and Lee (bless him) is the only Oxfordian to have responded here to the argument I presented. Nobody is going to persuade me that Lee is the only Oxfordian to have read my post, so where are they all? Why wouldn't they want to be the first publicly to prove my conclusion wrong? Unless...

Peter Farey

Dave Herber said...

Peter, I too hope that this is not the entirety of Brian's work and that in order to at least further verify his findings he has indeed matched Marlowe with Kyd and Shakespeare.

I am making inquiries as to how I can contact him as I would like to know more. If I get a reply I will certainly pass on what I find.

Anonymous said...

I have proved Peter's arguments wrong on numerous occasions over the past decade on HLAS.

Firstly, his mathematical operations would give the screaming heebie-geebees to anyone with a familiarity with Statistical method (more on this later).
Secondly, (and not too different) he forgets that Shake-speare (whoever he was) was both human and a supreme artist. That artist consciously decided whether or not to end-stop or to employ a feminine ending on each occasion. He was not (as the Stratfordians would have us believe) an automaton, obliged to follow some pre-set pattern.
Thirdly, Peter adopts another (exceedingly bad) Stratfordian assumption that "Shake-speare" was a follower and not a leader.
Fourthly, he ignores the Oxfordian theory that the canonical plays were written primarily for the royal court, and Oxford (as "Shake-speare") would have been setting trends in works not usually available to fellow-playwrights for a decade or more.

The mathematical and conceptual problems in Peter's thesis can best be seen with an analogy. Suppose we had a great painter who, apparently following contemporary trends during his career, (a) used more of the colour blue in his paintings, and (b) generally increased the size of his canvases with time. We _could_ present two graphs showing these trends, but no one would want to claim that we could confidently date a particular work of this artist by looking at the extent of the colour blue, or by the size of the canvas. The painter might on any occasion have broken the supposed "rule". He was an independent agent.

A statistical horror here would be to add a measure of 'blueness' to one of size, claim that you had an entity called 'blueness + size' and produce a graph setting that 'entity' against a timescale.

But that is exactly what Peter does. There is no more an entity of 'feminine endings + run-on lines' in any literary work, than there can be one of 'blueness + size' in a set of paintings.

Peter found that the 'run-on lines' scale, on its own, did not provide a good enough graph; nor was the feminine-endings graph good enough on its own. So he created a 'super variable' by adding them together. With more research (even just by asking me) he could have found other variables that apparently changed over time (e.g. in the use of particular words). Had he added in a third scale, he would have had a 'super-super variable'; and with a fourth a 'super-super-super-variable'.

There are, of course, appropriate ways of dealing with multiple measurements of a population -- one of them is called 'multiple regression' (readily available on standard spreadsheets like Excel). But, Peter does not bother with them -- no matter how often told. His only concern is to produce a nice-looking graph.

However, there are even more basic problems to his argument. To continue my analogy, let's suppose that an art critic were to claim that he could date a set of paintings (of a great artist) from their 'blueness' and their size alone. He would, of course, be laughed out of court. We would immediately ask why he had not considered other aspects of the artist's style: say, the kind of brush-strokes, the subject-matter, the lighting, and so on.

There are enormous numbers of qualities in any art (including painting and literature) which we can use to make judgements -- such as on the dating of particular works. Most of them can be measured on some scale, and many of those measures will broadly change in one direction or the other over time. But many will not. It is not good enough to 'cherry-pick' from the huge number of possible measures a mere two -- ignoring all the rest.

Even if Peter had used respectable mathematical techniques with his two "cherry-picked" variables, his results would not have been meaningful. Of course, he could have cherry-picked a few more that roughly fitted his desired pattern, and finished up with a (satisfactory-to-him) biased result.

But what he did was mathematically monstrous.

Paul Crowley

Peter Farey said...

Paul Crowley said...
I have proved Peter's arguments wrong on numerous occasions over the past decade on HLAS. Firstly, his mathematical operations would give the screaming heebie-geebees to anyone with a familiarity with Statistical method.

No Paul, you have simply told me they are wrong, and as countless people there (the humanities.lit.authors.shakespeare newsgroup) have explained to you throughout that time, merely asserting something is not the same as proving it. In fact a great many people "with a familiarity with Statistical method" have been exposed to my "mathematical operations" and not a single objection - let alone a case of "the screaming heebie-geebees" - has been reported so far. Furthermore I do seem to recall in this context a university Professor of Mathematics responding to a comment (not mine) that you displayed "a wide and deep ignorance of statistics" with "True; however, statistics is not thereby singled out -- indeed, it would be prohibitively difficult to name a field that Mr. Crowley has NOT previously displayed a wide and deep ignorance of."

Your analogy of this with the painter displays this ignorance quite nicely. In what units would one measure "blueness" which could be added to units of "size"? Can't be done, can it? Whereas the number of open lines which occur on average per hundred lines of verse in a play can quite easily be added to the number of feminine endings per hundred lines of verse in the same play. Of course I would have preferred to have expressed this in terms of the percentage of lines in which the endings of the regular end-stopped iambic pentameter line were broken by one or other of these techniques. But the raw data didn't permit this, as a single line of verse can and does occasionally have both. Nevertheless, the combined figure is certainly a good enough approximation, and I wasn't prepared to do a completely new count of lines of verse, open lines and feminine endings for every one of the 77 plays covered in my paper, even for you.

Nor does this have anything to do with the dating of individual plays. If the dates of individual plays could be ascertained in this way it is how they would do it, but they don't. In fact the most sophisticated approach so far was very similar to your "super-super-super-variable" and was still found to be insufficient for individual dating on its own.

But I'm not talking about dating individual plays am I? What I am talking about is what can be learned when huge numbers of individual pieces of data combine to produce an otherwise unexpected pattern, and what the explanation for that pattern might be. In this case it is quite obvious that the pattern thrown up is mortal to the Oxfordian theory. Obvious, that is, to all except those who refuse to countenance anything which runs counter to the way things must be, since their own particular theory is right, and mere facts can never be allowed to affect that belief in any way.

Peter Farey

Paul Crowley said...

This rubbish software objects
to everything I try to post.

Forget it.

Paul.

Peter Farey said...

Paul Crowley said...
This rubbish software objects
to everything I try to post.
Forget it.

Oh no! At last the chance to nail my "mathematical monstrosity" once and for all and the damned software won't let you. How very disappointing.

Meanwhile, I guess I'll just have content myself with a few comments I left out earlier. I think I dealt quite adequately with your first "statistical" point, but what about the others?

Paul said:
Secondly, (and not too different) he forgets that Shake-speare (whoever he was) was both human and a supreme artist. That artist consciously decided whether or not to end-stop or to employ a feminine ending on each occasion. He was not (as the Stratfordians would have us believe) an automaton, obliged to follow some pre-set pattern.

Peter says:
"Shakespeare" wrote precisely what sounded or felt right to him at the time. The idea that while he was writing he might "consciously decide" to use either of the two techniques is indeed ludicrous. That what sounded or felt right to him (and others) changed as the years went by, however, is certainly not.

Paul said:
Thirdly, Peter adopts another (exceedingly bad) Stratfordian assumption that "Shake-speare" was a follower and not a leader.

Peter says:
Had you got around to reading what I said you would have seen "in the first graph the contemporaries' trendline is only two points away from his level in 1598. By 1604, however, he is ahead by some ten points on average, and by 1613 that difference between them has grown to twenty per hundred lines of verse." Does this sound like a follower to you?

Paul said:
Fourthly, he ignores the Oxfordian theory that the canonical plays were written primarily for the royal court, and Oxford (as "Shake-speare") would have been setting trends in works not usually available to fellow-playwrights for a decade or more.

Peter says:
Not available! What sort of logic is that? Unfortunately, the Elliott and Valenza data cover only four Marlowe plays. Using my own counts, however - which I have shared with you in the past - we see that Marlowe's translation of Lucan's Pharsalia experimented with a level of open endings which "Shakespeare" would not typically reach until half way through his career. Luckily Marlowe had no idea that such an utterly irresponsible use of open lines wasn't "available" yet!

Peter Farey

George Dillon said...

As a wavering Oxfordian, this certainly does give food for thought. I don't believe in fighting a lost cause to the death (which is why I'm not a Stratfordian), and I don't want to get drawn into a lengthy argument about Meres, or the validity of including in this kind of stylometric analysis plays which were co-written (particularly in Shakespeare's declining years) or the obvious warping effect the final two plays have on the steepness of the Shakespeare trend.

What I do wish to note is your less than equal presentation of the graphs and the way you have shaped them to suit your argument, rather than presenting them straight and making objective observations including uncomfortable ones for your own theory.

I recently attended a meeting of the de Vere Society where the 'Dating Project' was being discussed and Professor William Rubinstein (co-author of the Neville-ite "The Truth Will Out") criticised the randomness of previous Oxfordian redatings of individual later works on the grounds that it lacked what he called (I think) an "evolutionary career trajectory" the kind of which exists in the conventional dating scheme. In other words, Shakespeare (the writer) developed and that the sequence of the plays is pretty much established according to this development, so any redating needs to move the whole cannon, in order.

Your article does seem to make the Oxfordians' task that much harder...

However, the graphs reveal something else, which would have been much more noticeable had you used equal scales and ranges.

Unfortunately I cannot post images here, or I would show you the Shakespeare results resized and superimposed on the Contemporaries results (cut and paste the images from this webpage into a graphic editor, resize the Shakespeare layer width to 48% and height to 103% and it will fit nicely).

What they show is that Shakespeare started below his contemporaries, only catching up in 1595 and then developed at such a rate that he reached the 'finishing line' of 60 per 100 in 1610. Indeed his rate of development was so steep that had he lived another 20 years he would have had more line endings than there were lines.

But if you fiddle with Shakespeare's graph another way - resizing Shakespeare's career to have it starting earlier, so 1590 becomes 1585 (while still ending in 1615), suddenly All's Well: Shakespeare starts at the same point as the others and all but the last four plays fall within more or less the same development trend.

And Bang! goes Stratford (and Canterbury too!).

I suppose Shakespeare being a slow starter could, of course, be explained by him really being an uneducated glover's son, and of course he was a genius when he started, which was how he was able to do it at all and catch up so quickly, but the plays don't really exceed his rivals in terms of knowledge (or line endings) because by the end he wasn't really a genius but a really ordinary geezer just like us.

So maybe it was Stratford Will after all.

Which just shows if you set out to prove something with statistics, as long as you select the right statistics and make some pretty graphs, and dont pay too much attention to mathematical rigour you can support almost any argument...

I dare say a mishmash of computers and selective data could even be used to support the idea that Marlowe was Shakespeare... which of course he wasn't because if he had been he wouldn't have written so much rubbish... Merry Wives... honestly!

Peter Farey said...

George Dillon said...
What I do wish to note is your less than equal presentation of the graphs and the way you have shaped them to suit your argument, rather than presenting them straight and making objective observations including uncomfortable ones for your own theory.

What a suspicious mind you have George! I can assure you that the graphs are exactly how Excel created them once I had included all the data that Elliott and Valenza provided. The only thing I confess to changing (in fact according to what you say I should have done) was to make the vertical scale of the contemporaries figures match that of Shakespeare's. Removal of the last two plays would have meant the trend line intersecting the 70 level in 1614 rather than 1613. So what?

The fact is, however, that the only thing I intended to illustrate by the first graph was that there was a general movement away from the regular end-stopped iambic pentameter throughout this period, and that Shakespeare was by no means unique in this respect. Nothing in my argument required any sort of matching of the two graphs to be done.

Nor have I any idea why you should think that the superimposition of the two sets of data onto one graph should give rise to observations which are uncomfortable for my theory. As you know, I did in fact point out how Shakespeare's figures increased at a much faster rate than his contemporaries. That his trend line starts below theirs is of course mainly to do with the steepness of the line itself rather than being a clear statement of how his earlier plays actually compared with those of his contemporaries. Incidentally, your statement that "indeed his rate of development was so steep that had he lived another 20 years he would have had more line endings than there were lines" ignores the fact that, since each line can both be open and have a feminine ending, the theoretical maximum per 100 lines is 200.

Your suggestion of stretching the plays out from 1585 does of course mean that a whole lot of new plays must have been in the Lord Chamberlain's Men's repertoire by 1598, and in the verse form as measured by E & V if the trend is to be maintained. Since these would include Much Ado and Twelfth Night one could not but deprecate Meres's choosing instead Comedy of Errors and Two Gentlemen of Verona as examples of Shakespeare's being best in comedy for the stage, and his going for Titus Andronicus and King John as his best tragedy, rather than Hamlet, Othello, Lear and Macbeth would be quite ridiculous. On the other hand, the quite uncanny knack he had for picking out all except one of those with a rate below 35, and rejecting all except two with a rate above that would display a discriminatory power verging on the impossible. And impossible I fear it is.

Peter Farey

PatB said...

I followed this question on the Oxfordian site. It was a long thread, and I may have missed something, but I couldn't figure out who had done the work of counting up the F/O endings in the canon, and even after reading this, it still isn't clear. Is it you or is it Elliot and Valenza?

Peter Farey said...

The figures are those of Elliott and Valenza, taken from their "And Then There Were None: Winnowing the Shakespeare Claimants", details of which are given in notes 1 and 2 of the article. I'm sorry you found it confusing.

Peter Farey

PatB said...

Okay, thanks. But you're saying E&V didn't use line-ending analysis as a method of dating the plays. What did they use it for, then? It would be perverse to collect data unless you thought it might tell you something. Has it, for example, got any use in eliminating authorship candidates who can't be eliminated by dating, as apparently Oxford can?

Peter Farey said...

Yes, they were using it (as the title of their paper suggests) as part of their procedure for the elimination of "Shakespeare Claimants", a procedure which is far too complicated to go into here!

Peter Farey

Anonymous said...

I seldom leave comments on blog, but I have been to this post which was recommend by my friend, lots of valuable details, thanks again.

Pyotr Grozny said...

I haven't read all the comments but for me there has been a problem with E of O as Shakespeare because I broadly subscribe to the orthodox chronology. However if you consider the possibility that either E of O didn't die when stated, or that someone else (such as his son-in-law William Stanley) carried on writing after his death, or a combination of these, then the problem is not necessarily insuperable.