University of Virginia Library

III

What other reasons are there for believing Peele to have been the author of Titus Andronicus, Act 1? Dover Wilson summarized the work of earlier scholars and added some observations of his own. He listed parallels between Titus Andronicus and Shakespeare's plays and poems, and compiled an inventory of a dozen "common Shakespearean turns of speech," familiar to him from his editorial labours on the New Shakespeare series.[19] These afforded evidence of Shakespeare's presence in every scene but the first, which is full of verbal parallels to Peele's plays and to his poem The Honour of the Garter, written in the middle of 1593. Wilson, recording these parallels in his introduction and commentary, showed also that Peele's diction ("diadem," "gratulate," "re-salute," "gramercy," "panther," "remunerate," "gratify," "consecrate" for "consecrated," and so on) and his "clichés and tricks" of composition were prominent in Act 1 of Titus Andronicus. The tendency for speech after speech to begin with a vocative and continue with an imperative verb is especially marked in the first half of Titus Andronicus, Act 1, as in the opening scenes of Edward I and The Battle of Alcazar. And the same mechanical repetition of words and phrases occurs.

Maxwell added an argument from syntax. The construction in lines 5 — 6 of Saturninus' opening speech — "I am his first-born son that was the last / That ware the imperial diadem of Rome," where "his first-born son that" means "the first-born son of him who" — is unusually frequent in Act 1 of

Titus Adronicus, appearing six or seven times as often as in the rest of the play, and Maxwell shows that in Peele's non-dramatic poetry it is also about six times as common as in Venus and Adonis and The Rape of Lucrece. His search of a fair number of contemporary plays revealed that the construction — a possessive adjective or pronoun as antecedent of a relative clause — "is common in Peele's later work, fairly common in Kyd and Marlowe, rare in Shakespeare, Greene and Lodge."[20] The incidence was also low in most of the anonymous plays that Maxwell examined. David and Bethsabe is the play by Peele with the largest number of examples of this syntactical mannerism. As Maxwell noted, the merits of his particular test as an indication of authorship are twofold: use of the device would seem to bear little or no relation to subject matter, and similar rates of usage are not likely to be due to conscious, or even unconscious, imitation by one author of another.

R. F. Hill surveyed the use of the rhetorical devices in Shakespeare's early plays, and found Titus Andronicus uncharacteristically sparing in its use of some figures (such as antimetabole, epanodis, symploche, epanalepsis, asyndeton, and brachiologia) and uncharacteristically prodigal in its use of others (such as certain forms of epizeuxis, chiasmus, and pleonasm).[21] He pointed out that alliteration, more frequent in Titus Andronicus than in other Shakespeare plays, was employed to excess in Act 1. Several of the oddities appear to be particularly prevalent in Act 1, and although comparative data are not available for Peele, he is easily seen to be partial to alliteration and to several of the most prominent of the tricks that Hill categorizes. Hill concluded that Titus Andronicus was either Shakespeare's first play or the work of more than one author.

Drawing on Spevack's Concordance, statistician Baron Brainerd sought lexical items whose frequencies in Shakespeare's plays covaried with chronology, in an attempt to calculate an "omnibus predictor" of date of composition.[22] Beginning with plays for which the dating is relatively uncontentious, he was able to combine variables into a fairly good predictor. A few plays, including Titus Andronicus, were "deviant" with respect to the variables tested. The plays found to fall into this category with Titus Andronicus were ones suspected to be of dual or multiple authorship or to have been subject to authorial revision at a later stage of Shakespeare's stylistic development.[23]

In Studies in Attribution: Middleton and Shakespeare I investigated the rare-word vocabulary of Titus Andronicus in relation to a division of the play based objectively on the percentage of feminine endings per scene as these had been determined by earlier scholars.[24] A treatise by Philip W. Timberlake covering use of the feminine endings in all Elizabethan drama up to the year 1595 revealed that even in his earliest works Shakespeare tended to employ feminine endings at a higher rate than ever attained in the known plays of Greene, Peele, Nashe, Lyly, Lodge, or Marlowe.[25] The percentage of feminine endings in Titus Andronicus associates most scenes with Shakespeare rather than his early contemporaries, but for 1.1, 2.1, and 4.1 the figures are low. I noted that the scenes selected as doubtful on this metrical evidence and labelled "Part A" are deficient in other features that characterize the young Shakespeare's verse: for instance they have fewer compound adjectives and Shakespearean images than the rest of the play ("Part B"). The division into two "parts" was not intended to be hard-and-fast or to have any absolute validity; it was simply a means of testing the hypothesis that two "strata" existed. These might be authorial or chronological.

The investigation of vocabulary was confined to words that appeared in Titus Andronicus and once or twice in other Shakespeare plays. Nearly a century ago the German scholar Gregor Sarrazin had shown that such words most strongly linked plays composed at approximately the same time.[26] The rare-word links of Titus Andronicus Part B to Shakespeare plays of four successive chronological groups of about the same total size fell as follows: 37:26:19:23. As we should expect, links with the earliest group predominate. For Part A the figures were 33:9:8:8. They thus exhibit a far more extreme concentration of links with the earliest group. The difference between Parts A and B in the degree of concentration of links with the first group (33:25 compared with 37:68) is statistically significant, with odds of about a hundred to one that it is due to chance.[27] The indications are that Part A was written either some years before Part B or by a different author. The second alternative may seem the less likely. Would writing by other dramatists of the 1580s and early 1590s share with Shakespeare's own writing a tendency to be most strongly linked in its rare-word vocabulary to his first group of plays?

The answer is that it might well do so. M. W. A. Smith and Hugh Calvert have tested Peele's Edward I and Greene's James IV and discovered that each exhibits just such a pattern.[28]

Marina Tarlinskaja's recent research into Shakespeare's verse would appear to confirm the presence of two strata in Titus Andronicus. Her main concern is with the extent to which each of the ten syllabic positions in the standard blank verse line is occupied by a stressed or unstressed syllable in accordance with the iambic paradigm. She finds that for different "ictic" or "non-ictic" positions the degree to which theoretical expectations are realized changes throughout Shakespeare's career so as to create consistent chronological trends. Most of the metrical details with which she is concerned would associate Titus Andronicus with I Henry VI as the earliest of Shakespeare's plays. But certain features of Part A are unmatched in the canon before Much Ado About Nothing and Henry V. So while on the theory of Shakespeare's sole authorship the vocabulary evidence would place Part A appreciably earlier than Part B, some of Tarlinskaja's metrical evidence would place Part A appreciably later than Part B. The contradiction might be resolved on a theory of dual authorship. Tarlinskaja herself speculates that Shakespeare may have written Part A at a time when his metrical practices had not yet stabilized, but she "is really tempted to attribute 'Titus' to two different authors."[29]

Tarlinskaja's findings were published too late to be taken into account by Eugene Waith, who did, however, consider and dismiss the implications of my vocabulary data. Quoting in his edition of Titus Andronicus a letter from Gary Taylor in 1981, he objected that the three scenes comprising my Part A "are linked by no narrative or formal logic, and that dramatic collaboration almost always involved a division of the plot along some obvious logical lines" (p. 17). But, as Taylor recognized by 1987 in his essay on "Canon and Chronology" in the Oxford Textual Companion, in the Quarto of 1594 Act 1 and the first scene of Act 2 (as they are in most modern editions) form a single uninterrupted scene, which initiates the action, while 4.1 initiates the counter action. "The division suggested by feminine endings is thus compatible with patterns of collaboration in the drama of the period" (p. 114).

At any rate, a significant disparity, in vocabulary and metre, would remain if we were to redefine Part A, reducing it (by about one-third) to the first Act alone, and my remarks on further "unShakespearean" or "Peelean"

features in Titus Andronicus will focus on this, the most suspect portion of the play. Rates of occurrence of high-frequency function words have been proven to be useful indicators of authorship.[30] Spevack's concordance furnishes information about these. He gives rates for every word in every play, these being expressed as percentages of the total number of words (of "tokens," that is, not "types"). For all thirty-eight plays, including The Two Noble Kinsmen, the rates for and, which is the second most common word in the canon after the, vary from 2.398 for The Two Gentlemen of Verona to 3.844 for Titus Andronicus. The mean rate per play is 2.998 and the standard deviation is 0.387. Titus Andronicus is the only play for which the rate is more than two standard deviations from the mean.[31] Act 1, in which the rate for and rises to 4.809, is chiefly responsible for the anomalously high rate of the play as a whole. If it is excluded, the rate for Titus Andronicus becomes 3.556, well within the normal range of two standard deviations from the mean for a Shakespeare play. The disparity between Act 1 and Acts 2 — 5 is highly significant, statistically speaking. The odds are less than one in a thousand that it is a chance phenomenon — which is not to say that dual authorship is the only possible explanation.[32] In its rate of use of with, Titus Andronicus again falls at the end of the range of Shakespeare's plays, and again Act 1 is chiefly responsible. The average for the Shakespeare plays is 0.878, the standard deviation 0.113. Titus Andronicus, with a rate of 1.136 is the sole play to fall outside two standard deviations from the mean.[33] For Act 1 the rate is 1.282, while the rest of the play, at 1.103, would just fall within the normal range.[34] This time the difference between Act 1 and Acts 2 — 5 is not statistically significant, but Act 1 is especially anomalous.

Would the rates for and and with in Titus Andronicus, Act 1, be inconsistent with Peele's practices? Computerized counts of the opening scene or scenes of Edward I, The Battle of Alcazar, and David and Bethsabe — amounting to roughly two thousand words from each play — give some basis for an answer.[35] For these three samples the rates for and are 3.872, 5.314, and 4.329;

for with they are 1.518, 1.213, and 1.644. All six of these rates are, like those for Act 1 of Titus Andronicus, outside the normal Shakespeare range. In combination the three Peele samples yield a rate for and (4.517) a little lower than that of Titus Andronicus, Act 1, and a rate for with (1.453) a little higher. Peele therefore shows the same partiality for and and with that distinguishes Act 1 of Titus Andronicus from the rest of the Shakespeare canon.

Less arid than these statistics are some details concerning a trick of style that recurs conspicuously in Act 1 of Titus Andronicus and is found in the very first line: "Noble patricians, patrons of my right." This is the ending of a blank verse line with a preposition or conjunction, followed by a possessive pronoun plus a monosyllabic noun, as in "of my right," "with your swords," "to our foes," "of his name," "and his sons." The formula, usually preceded by a two-syllable word stressed on its first syllable, produces a pyrrhic foot followed by a foot that is some way between an iamb and a spondee; alternatively, one might say that each successive syllable carries marginally more stress than the one before it, but only the last of the four is strongly stressed. The rate of occurrence in Titus Andronicus, Act 1, is one in every 12.7 lines.[36] In the rest of the play it is one in every 24.7 lines. The odds are less than one in a thousand that this disparity is a matter of chance.[37] Counts for the opening Acts of Shakespeare's eight other earliest plays yield the following rates: one in 22.8 for The Two Gentlemen of Verona, one in 37.3 for The Taming of the Shrew (ignoring the Induction), one in 20.6 for 2 Henry VI, one in 19.3 for 3 Henry VI, one in 21.9 for I Henry VI, one in 24.0 for Richard III, one in 17.6 for The Comedy of Errors, and one in 34.8 for Love's Labour's Lost (where the first Act is very short). The total for all eight plays furnishes a rate of one in 22.8, which is close to that for Titus Andronicus, Acts 2 — 5. Each of the four Acts of Peele's David and Bethsabe provides a similar match to the first Act of Titus Andronicus: one in 13.6, 10.8, 12.2, and 8.3 of the full pentameter lines end in the "of my right" kind of formula; the overall rate for the play being one in 11.2. It is more difficult to calculate figures for Edward I, in which many scenes consist mainly of prose or rhymed verse, much of it doggerel, but for blank verse speeches the rate is about one in 13.2. In addition, many lines end in phrases such as "on the way" and "at the name," where the definite article substitutes for the possessive pronoun. Although both the "from his flesh" and "on the ground" sorts of line ending are quite common in The Battle of Alcazar, the percentages are within the normal Shakespearean range.