In "Symbolic Logic, Part i." (London: Macmillan, 1896) he employed
both methods. The Introduction is specially addressed "to Learners,"
whom Lewis Carroll advises to read the book straight through, without
_dipping_.

This Rule [he says] is very desirable with other kinds of
books--such as novels, for instance, where you may easily
spoil much of the enjoyment you would otherwise get from the
story by dipping into it further on, so that what the author
meant to be a pleasant surprise comes to you as a matter of
course. Some people, I know, make a practice of looking into
vol. iii. first, just to see how the story ends; and perhaps
it _is_ as well just to know that all ends
_happily_--that the much persecuted lovers _do_
marry after all, that he is proved to be quite innocent of
the murder, that the wicked cousin is completely foiled in
his plot, and gets the punishment he deserves, and that the
rich uncle in India (_Qu._ Why in _India? Ans._
Because, somehow, uncles never _can_ get rich anywhere
else) dies at exactly the right moment--before taking the
trouble to read vol i. This, I say, is _just_
permissible with a _novel_, where vol. iii. has a
_meaning_, even for those who have not read the earlier
part of the story; but with a _scientific_ book, it is
sheer insanity. You will find the latter part
_hopelessly_ unintelligible, if you read it before
reaching it in regular course.