Three Kinds of Analysis

About This Page

A recurring objection to this site's method runs as follows: the Shakespeare Authorship Question is a literary and historical matter, and applying criteria drawn from scientific reasoning to it is a category error. The criteria should come from the discipline that owns the question, literary criticism, not from a framework that resembles experimental methodology.

The objection identifies a real tension. The answer to it clarifies what kind of question the SAQ actually is, which turns out to be the most important methodological issue in the debate.

What separates the three kinds

Scientific analysis, in its proper sense, requires controlled conditions. A hypothesis is formulated, an experiment is designed that could falsify it, the experiment is repeated by independent parties, and results are expressed with quantified uncertainty. The strength of scientific inference comes from two features that are very hard to fake: replication (the same result obtained by different people under the same conditions) and randomisation (variation that cannot be attributed to the researcher's choices). Science is the right tool when you can set up controlled conditions and when the phenomenon you are investigating is repeatable.

Historical analysis works differently. The historian faces a world of singular events, things that happened once, in specific places, at specific times, and left traces in documents that may be incomplete, damaged, forged, or lost. No historian can replicate the Battle of Agincourt. What they can do is gather all available evidence, test each piece for authenticity and provenance, check whether independent sources corroborate each other without having been in contact, and see whether a coherent account emerges that resists alternative explanations. The strength of historical inference comes from convergence, multiple independent threads pointing to the same conclusion, and from the elimination of competing hypotheses.

Literary analysis is different again. Its primary object is the text itself, not the person who produced it. Close reading identifies patterns of language, imagery, and structure. Source studies trace what the author read and how they transformed it. Stylistic analysis characterises an author's distinctive habits. Literary analysis is the right tool for questions of meaning and form, what a text does and how it does it. It is a less reliable tool for questions of historical identity, because the same text can often be interpreted as consistent with more than one biographical account.

Which tool fits which question

"Who wrote the works of Shakespeare?" is a historical question. It asks about a specific person, a real individual who was alive during a specific period, whose life left traces in parish records, legal documents, letters, and account books. That person either visited Italy or did not; either had access to a particular library or did not; either corresponded with specific courtiers or did not. These are factual questions about a historical individual, and the evidence that can settle them is the evidence that historians use: primary sources, biographical records, documentary corroboration.

Literary analysis contributed enormously to identifying what we should be looking for. Scholars working entirely on textual questions, without any authorship agenda, noticed the unusual precision of the Italian topography, the depth of the legal vocabulary, the accuracy of the falconry terminology, the specific relationship to Montaigne's unpublished manuscript. These literary observations define the profile of the author. But identifying the profile is not the same as identifying the person. Once the profile is established, the question of who fits it is answered by historical evidence, not by further textual analysis.

Scientific tools have genuine applications at the edges of the problem. Forensic document examination is the right method for evaluating the Hand D attribution in Sir Thomas More. Computational stylometry can, in principle, test whether the writing habits of a known author match those of the canon, though the reliability of stylometric methods under Early Modern drama conditions is contested, and this site excludes them on those grounds rather than on any principled anti-scientific position. If a primary source document ever surfaced connecting a specific individual to the authorship directly, dating and provenance analysis would be the relevant scientific tools for evaluating it.

What this site actually does

The framework used here is historical in method. The evidence is drawn from primary sources: biographical records, correspondence, dedications, archival documents. The standard for a positive score is documentation, something that survives in a form that can be examined, sourced, and challenged.

What the site borrows from scientific thinking, not from science itself, is the logic of pre-registration. Criteria are fixed before any candidate is examined. This eliminates the most common failure mode in advocacy-driven inquiry, which is selecting evidence that fits a preferred conclusion and calling that a method. Pre-registration is not unique to science; it is the structural equivalent of what a court does when it agrees on the rules of evidence before any testimony is heard. It is equally at home in historical methodology.

The convergence framework borrows a second idea from scientific epistemology: that independent threads of evidence pointing to the same conclusion are stronger than any single thread, in proportion to their methodological diversity. This is not a scientific import either; historians call it consilience of evidence, and it has been the standard epistemological framework of historical inquiry since the nineteenth century. What would be scientifically unusual is treating it as proof. The site does not. A high convergence score means a candidate fits the available evidence better than others. It does not establish authorship.

The objection and what it usually means

When someone argues that scientific analysis is inappropriate for the SAQ, they are sometimes making a genuine methodological point about the limits of quantification applied to humanistic questions. That concern is addressed directly: the scores on this site are marked approximate, scoring judgments are explained in tooltips so they can be challenged, and the framework explicitly disclaims the kind of precision that would only be available if the evidence were quantitative.

More often, the objection is doing something else. It is arguing that the question should be settled by literary-critical consensus rather than by evidence assessment. The implied standard is: "Shakespeare scholars say Shakspere wrote the plays, and that is what literary scholarship has established." The difficulty is that this is not a method; it is a conclusion. Literary criticism did not establish the Stratford attribution by examining the primary sources; it inherited the attribution from a tradition that long predates any systematic scrutiny of the biographical record. When that scrutiny was eventually applied (by Diana Price, most rigorously) it found that the evidentiary basis for the attribution was weaker than for any other major author of the period. Consensus is not evidence. Consensus can crystallise around an assumption as easily as around a finding.

This objection can come from any direction, including from advocates whose candidate scores well under the very framework they are questioning. Objecting to systematic analysis while retaining its favourable conclusions is not a methodological position; it is a preference for the result without the method.

The appropriate response to "which method works here?" is to ask what kind of question you are trying to answer. If the question is what the works mean, literary analysis is the right tool. If the question is who wrote them, historical analysis is required. If specific evidence warrants it, scientific tools (forensic examination, computational analysis) can be brought in for particular sub-questions where they are reliable. This site applies historical method to a historical question, uses literary scholarship to characterise what that question requires of any answer, and borrows the epistemological habits of scientific inquiry (pre-registration, convergence, approximate scoring) to make the analysis as transparent and challengeable as possible. That is matching the tool to the question, not a category error.

A note on stylometrics

Stylometrics deserves its own treatment because it is the one genuinely scientific tool that has been seriously applied to the SAQ, and because this site excludes it. The exclusion is not an objection to quantitative methods in principle. It is a specific objection to the reliability of the Zeta-test methodology under the conditions of Early Modern drama, where collaboration between authors was routine, where manuscript transmission introduced scribal variation, and where genre constraints produce systematic linguistic differences that stylometric models struggle to separate from authorial signal. A method is only as good as its signal-to-noise ratio under the actual conditions of the data. The exclusion of stylometrics is itself a scientific judgment, not a retreat from scientific standards. The full case is made on the Why Not Stylometrics page.