frightened too ? frightened to what ?

I checked available sources on Internet, and it seems that it has been recopied everywhere...

Could a native confirm that this "to" makes sense here ? Otherwise, all these "sources" have to be taken with care...

I have seen many times that old books have been scanned and automatically digitized by robots with lots of mistakes that are subsequently taken for granted as references...

"frightened too", of course. I overlooked this error. Merci beaucoup!

but it seems that the error is well established in multiple sources on Internet...

That's probably because "too" and "to" have similar sounds; sadly, that happens a lot with languages with subtle sound differences among different letters or with uncertain pronunciation rules such as english.
It would surprise you knowing how much italian native speakers mistake "e" for "è" and vice-versa, even though they have completely different meanings.

But it's a sentence by Stevenson, it dates from the 19th century. It means that it has been copied and recopied wrongly all over Internet...

And the advocates of usage will probably defend that "to" is now a valid version of "too" because there are so many examples by Stevenson himself...

Oh, you meant the error is specific to this quote. Uhm, this makes it more complex. To know whether to consider it a proper usage or else could require some etymological research.
In the worst case, I tend to consider it a license rather than a correct usage, although I'm not native so my opinion has only a relative weight on the matter.

No, what I mean is: it is a clear mistake (and I doubt it was a mistake by Stevenson himself or his publisher), but we don't know when the error appeared in the transcriptions. The fact is that now, this mistake spread over the entire Internet...
I suspect it happened, like many errors of book transcriptions on the Internet, by a scanning/digitising error...

I never saw the original work itself, so that might be the case as well; I frankly hope so.

Unfortunately, languages will keep involving faster than they evolve, regardless.

It's probably a mistake in the OCR that was not corrected. At least, I can confirm that the project Gutenberg version does not contain the mistake: http://www.gutenberg.org/cache/epub/42/pg42.html

Nor everybody knows every abbreviation, so let me add:

OCR = Optical Character Recognition
deu => Texterkennung / Optische Zeichenerkennung
spa => Reconocimiento óptico de caracteres (ROC)
epo => Optika signorekono (OSR) / tekstorekono
fra => Reconnaissance optique de caractères (ROC)
ita => Riconoscimento ottico dei caratteri
jpn => 光学字符识别
por => Reconhecimento ótico de caracteres
rus => Оптическое распознавание символов
tur => Optik karakter tanıma

Another help
http://en.wikipedia.org/wiki/Op...er_recognition