One of the chief virtues of the anecdote is that it brings the reviewer down from his generalizations to the individual instances.  Generalizations mixed with instances make a fine sort of review, but to flow on for a column of generalizations without ever pausing to light them into life with instances, concrete examples, anecdotes, is to write not a book-review but a sermon.  Of the two, the sermon is much the easier to write:  it does not involve the trouble of constant reference to one’s authorities.  Perhaps, however, someone with practice in writing sermons will argue that the sermon without instances is as somniferous as the book-review with the same want.  Whether that it so or not, the book-review is not, as a rule, the place for abstract argument.  Not that one wants to shut out controversy.  There is no pleasanter review to read than a controversial review.  Even here, however, one demands portrait as well as argument.  It is, in nine cases out of ten, waste of time to assail a theory when you can portray a man.  It always seems to me to be hopelessly wrong for the reviewer of biographies, critical studies, or books of a similar kind, to allow his mind to wander from the main figure in the book to the discussion of some theory or other that has been incidentally put forward.  Thus, in a review of a book on Stevenson, the important thing is to reconstruct the figure of Stevenson, the man and the artist.  This is much more vitally interesting and relevant than theorizing on such questions as whether the writing of prose or of poetry is the more difficult art, or what are the essential characteristics of romance.  These and many other questions may arise, and it is the proper task of the reviewer to discuss them, so long as their discussion is kept subordinate to the portraiture of the central figure.  But they must not be allowed to push the leading character in the whole business right out of the review.  If they are brought in at all, they must be brought in, like moral sentiments, inoffensively by the way.

In pleading that a review should be a portrait of a book to a vastly greater degree than it is a direct comment on the book, I am not pleading that it should be a mere bald summary.  The summary kind of review is no more a portrait than is the Scotland Yard description of a man wanted by the police.  Portraiture implies selection and a new emphasis.  The synopsis of the plot of a novel is as far from being a good review as is a paragraph of general comment on it.  The review must justify itself, not as a reflection of dead bones, but by a new life of its own.