The difference between an omission and a statement
Two cells can look identical and rest on completely different evidence: one holds a quotation, the other holds a conclusion somebody drew from a gap. Once they are in a table nothing distinguishes them. As of 2026-09-12.
| How the cell was filled | What a reader can do with it |
|---|---|
| A quotation from the page | Cite it, and check it against the address |
| A read and recorded silence | Rely on the gap being real on that date |
| An inference from an omission | Treat it as probable, not as published |
| An unread point | Nothing; it is not a finding |
Inclusion rule. The states a cell can be in where the source is a vendor's own published page. Order. Strongest evidence first.
1The two middle rows are the ones that get confused
A silence that somebody went looking for is evidence. An omission from a summary is an observation about a summary. Both produce an empty-feeling cell and only one of them supports a claim about behaviour.
Keeping them apart costs a phrase per cell. It buys the ability to say which parts of a register would survive somebody checking, which is the only property that makes a register worth consulting.
2Why the error is quiet
Nobody notices an overstated cell, because it reads like every other cell. The cost arrives much later, when somebody relies on it and finds the page never said anything of the kind.
At that point the register has spent its credibility on a sentence it could have written more carefully for free. There is no mechanism that catches this except writing the weaker cells as weak.
3What a careful cell looks like
By omission from the plan table. Not documented on the pages read. Not addressed in the reference. Each of those is longer than the confident version and each says exactly how much is known.
A reader scanning such a column will notice that some cells are hedged and some are not, and will weight them accordingly. That is the intended effect, and it only works if the hedges are used honestly.
4The same discipline applies to counting
Saying that a number of entries describe a mechanism requires deciding what counts as describing it. Counting inferences alongside quotations produces a bigger number and a worse one.
So a count is only as good as its inclusion rule, and an inclusion rule that nobody published is not a rule. Every table that carries a count should carry the rule beside it.
Background on mechanism and practice. Nothing here is attributed to a product, and nothing here is a reading of any instrument. The sourced material is on the generator table. Related: Recording silence, Why rows carry dates.