The term memorial reconstruction refers to the hypothesis that the scripts of some 17th century plays were written down from memory by actors who had played parts in them, and that those transcriptions were published. The theory is suggested as an explanation for the so-called "bad quarto" versions of plays, in which the text differs dramatically from a later published version, or appears to be corrupted or confused.
The theory however is facing growing criticism by a number of scholars, for being overly applied, and for being an elaborate theory, yet with little evidence to support it.
In 1623, the preface to the First Folio of Shakespeare's works specifically marketed its content as correct, in contrast to the garbled texts of "stolen and surreptitious copies" published previously. Memorial reconstruction has been supposed to be one of the ways in which texts were "stolen". Examples of possible memorial reconstructions are early editions of Shakespeare, including the second quarto (1598) of Richard III and the 1603 first quarto of Hamlet. It has been theorized that the only version to survive of Christopher Marlowe's The Massacre at Paris is a text obtained in this way, although there is no concrete evidence to support this assertion.
The theory has been used to explain the content of some quartos, and even to identify the actors responsible - on the assumption that they would get their own parts right, along with cue-lines and possibly other lines performed when they were onstage, but would most likely make more errors when reconstructing scenes in which their character was not present. The cast members of an Elizabethan dramatic production had their own parts written out for them, with relevant entrances and cues — but they did not have their own individual copies of the play text as a whole.
The theory has, however, been criticised on various grounds; that it is not based on serious research into the way actors actually remember or misremember lines; that texts may have been "stolen" by other means; or that the so-called "bad" quartos are early or alternative versions of plays that were later revised.
The theory emerged in embryo during the nineteenth century, but was only clearly defined by W. W. Greg in 1909 when he analysed the quarto text of The Merry Wives of Windsor, systematically comparing the divergences from the Folio version. He concluded that scenes containing the character of the Host are much closer to the Folio version than are other scenes. He therefore deduced that the actor playing the Host had played a significant role in creating the text of the quarto publication. In 1915 Henry David Gray analysed the first quarto of Hamlet using the same method. He concluded that the actor who played Marcellus was responsible for the reconstruction. He explained the fact that the "mousetrap" scene, in which Marcellus does not appear, was also accurate by suggesting that the same actor must have also played one of the roles in that scene. Both Gray and Greg argued that hired actors playing minor roles would be more susceptible to bribery than established actors in the company, as they had much less to lose. They also suggested that an anonymous writer filled out the missing verses.