In our next illustrative sentence we get away from details either formally or almost formally introduced; yet the relation to be shown by the punctuation is just the same and quite as evident:

34. She had a face altogether of the sunny south,—a pure skin, black hair, and blue eyes.

Professor Wendell uses this punctuation a great deal and, we think, very effectively. Other good writers do not use it at all; but they seem to have no satisfactory substitute for it. Professor Wendell frequently uses it twice in one sentence, as it is used in No. 33. The following sentence is from his “Literary History of America” (page 2):

35. These records [of things seen and felt by men] are often set forth in terms which may be used only by those of rarely special gift and training,—the terms of architecture and sculpture, of painting and music; but oftener and more freely they are phrased in the terms which all men learn somehow to use,—the terms of language.

We may perhaps turn aside from our discussion of the dash to consider a point in the above sentence which illustrates our general principle of grouping, especially as exhibited in Sentences 10 to 10-3.

Many punctuators would set off by commas the group “oftener and more freely” in No. 35 on the ground that it is “an intermediate parenthetical group.” This would be thoughtless punctuation based upon a rule of questionable meaning. The but relation in this sentence is between two groups of words restricted, respectively, in meaning by “often” and “oftener”; and commas should not be permitted to destroy a grouping that shows the contrast. Somewhat shortened, the sentence would read thus, readily exhibiting the point under consideration:

35-1. These records are often set forth in terms of architecture; but oftener they are phrased in terms of language.

The italicized words here emphasize the restriction of the thoughts that are in the but relation in this sentence. In No. 10-2 commas perform this office by holding the proper words in the but relation. We may not assert that commas in the second clause of No. 35 would change the meaning to the extent that it is changed by the absence of commas in No. 10-1; but the point of emphasis would be somewhat changed by their use in No. 35.