It was quickly recognised that published drawings were often representational and artwork. Further, a large proportion of textual information was descriptive relating to (superficial) artwork. This is useful for the evolution of the detail but of little use for regarding the architectural and structural datums.
Those texts which relate to the evolution of the build also tended to be confusing. For example, only a small percentage of 'original' facts are available. Writers seem to take a small quantity of facts and develop their hypotheses as to how they could have occurred (much like I am doing now).
The problem is that these theories have been adopted with little attempt at validation.
Unfortunately, these have then been referenced by other writers and eventually have become accepted as fact.
Acceptance of an hypothesis by another does not constitute validation.
The acceptability of a theory or hypothesis seems to be more dependent upon the understanding of the reader rather than upon its factual correctness.
Often, it seems that anecdotes have been rejected merely because they do not fit accepted hypotheses, rather than using them as an opportunity to test the hypotheses.
In my quality assurance work, I have always accepted that a feeling or an intuition, which can not be accurately defined, must be recognised and not rejected out of hand. The obligation being to investigate that feeling and to either prove, or disprove, it using a rational logic.
Invariably this means that:
For any information to be transferred from one person/place/time to another person/place/time, the recipient and user must be prepared to expend work in order to assure its quality.