Some examples of character-level transcription errors

(Also see the statement of principles which describes how EEBO texts are sampled and on what basis their error rates are calculated.)

A. CLEAR ("INEXCUSABLE") ERRORS

IMAGE	CAPTURED AS	SHOULD BE	ERRORS COUNTED
One or more letters added to text.
	sickness	sicknes	1
	my louers.	my loues.	1
	moistened	moistned	1
	Timne,	Time,	1
One word substituted for another
	all the vnderstand	all that vnderstand	2
One or more words omitted from text
	God's shall	God's 't shall	2
	A Street.	The SCENE, A Street.	9
One or more letters omitted from text
	to be the	to bee the	1
	Children,	Children,)	1
	presribe	prescribe	1
	aise	arise	1
	Discurse	Discourse	1
Two letters transposed
	Parmendies.	Parmenides.	1
	...mel|ioration...	...me|lioration...	1
One character mistaken for another...
	P.58	p.58	1
NOTE: Case confusion: "p" in "p.58" should be captured as lower-case. Compare the clearly upper case "P"s in the same image.)
	...lease...	...leafe...	1
NOTE: Clear "f" mistaken for long "s"
	stockt	flockt	2
NOTE: compare the form of the true "st" ligature in "Ministers":
	vis.	viz.	1
	Rea	Rex	1
	age.	ago.	1
NOTE: type is very small here, but word is clearly "ago" when magnified, not "age".
	attiq;	atti&abque;	1
NOTE: Failure to identify special abbreviation symbol (-q;) for "-que".

B. DOUBTFUL CASES

Beside the clearly inexcusable and the clearly exusable errors, there are always a few occupying the middle ground. Here are a few examples with notes on how they were resolved.

IMAGE	CAPTURED AS	SHOULD BE	EXPLANATION
	inquiry	iniquity	NOTE: edge clutter has almost drowned this word, but "iniq$$$y" can be read with fair confidence. The remainder should really have been left as "$" rather than guessed at. Two errors could have been assessed here (the missing "i" before the "q" and the mistake of "r" for "t" (or "$"). IN fact, both errors were treated as excusable because the shadow touching all the letters left some doubt as to their value.
	askedly	nakedly	NOTE: The "n" is broken and the "a" is slightly blotted but readable. $akedly would have been acceptable, but there is no justification for the reading "askedly"; compare the very different "as" that appears on the same line: . Treated as inexcusable.
	sull	full	NOTE: cross-stroke on "f" is somewhat shortened, almost to the point of resembling an "s". Treated as excusable.
	multie aliis	multis aliis	NOTE: the first "s" is only slightly less distinct than the second one, and really doesn't resemble an "e"; but it is blotted enough, and the type is small enough, that we let this pass as excusable, but doubtfully.

C. EXCUSABLE ERRORS

In most of the following cases, careful examination of the letter forms is enough to suggest that the capture was erroneous, but in most cases it is only the context that allows one to confirm that without question. Considered in isolation, these words cannot, we think, be read with certainty, and mistakes are therefore to be expected, and to be excused. E.g., in the first example, the bowl of the "e" is filled in and the back of it is curiously straight, so as to make it resemble a broken "t". The "m" was then read as "no" in order to make an intelligible word: in context, however, the word clearly must be "me," not "not. Similarly, defects of various other kinds make the other misreadings excusable, at least in our estimation.

IMAGE	CAPTURED AS	SHOULD BE
	not	me
	live	love
	hand	land
	fall	full
	to resti|	to testi|
	or will	as will
	auferrut.	aufertur.
	be speaks	bespeaks
NOTE: an example of "hypercorrection" of the spacing in the original. I.e., the print was spaced correctly, but was incorrectly "corrected" in capture. Such errors, since they depend on interpretation of the text, are always treated as excusable.
	bed kindred	had hindred
	euen	men

A NOTE ON GOOD AND LUCKY GUESSES

The following are examples of what seems to us to be accurate but interpretive transcription: cases in which the data is almost certainly correct in the sense that it probably captures what was printed in the original book--but in which the page image is defective enough that accurate capture can only have been accomplished by resort to the sense of the text, or so it seems. These are certainly not errors; in fact, we appreciate the boldness of this way of proceeding, and recognize that in many cases in which letter forms are indistinct, recognition of individual letters is (subconsciously or otherwise) dependent on prior recognition of the word in which it is placed. But we must point out that this method of proceeding also has risks: the more often one "guesses," the likelier it is that one will sometimes guess wrong and (more importantly) the likelier it is that the same kind of interpretative guesswork will be mistakenly applied even to perfectly clear words. In the latter case, of course, the result would be errors.

IMAGE	CAPTURED (CORRECTLY) AS
	Northampton-shire
	not
	servile