The purpose of this study is threefold: (1) to determine a set of variables lexical, metrical and structural that covary significantly with the date of composition of Shakespeare's plays, (2) to attempt, using some of these variables to date those plays with disputed dates, and (3) in the process to indicate some of the plays in which the writing style differs significantly from that of attested Shakespeare plays. Like my earlier essay on genre (Brainerd, 1979), this study is based on data gathered from A Complete and Systematic Concordance to the Works of Shakespeare (Spevack, 1968) which is itself based on the Riverside Shakespeare (Evans et al., 1974). The statistical procedures used are handled for the most part by means of the statistical packages SPSS (Nie et al., 1975) and BMDP (Dixon and Brown, 1979). I will indicate the names of these procedures when they are required. To determine a set of variables that covary with time, the percentage of occurrence of 120 lemmata, comprising approximately 450 word-forms, were chosen as possible covariates with the data of composition of the 38 plays in the canon. The 'dates of composition' of these plays are taken to be those given in the introduction to the Riverside Shakespeare. 1 Only twenty of these lemmata proved to be sensitive to date of composition (section 2). In addition to lexical variates, other variates such as the average verse line length in words, the percentage of split lines, 2 and certain type-token relationships were also considered.