repeated, our attention to our numbers will be discovered, and alarm the suspicion of the hearer. On the other hand, if we properly blend and intermingle the several varieties above-mentioned, our design will not be so readily noticed; and we shall also prevent that satiety which would arise from an elaborate uniformity of cadence. But the harmony of language does not result entirely from the use of _numbers_, but from the _juncture_ and _composition_ of our words; and from that neatness and _concinnity_ of expression which I have already mentioned. By _composition_, I here mean when our words are so judiciously connected as to produce an agreeable sound (independent of _numbers_) which rather appears to be the effect of nature than of art; as in the following passage from Crassus, _Nam ubi lubido dominatur, innocentiae leve praesidium est_ [Footnote: In the sentence which is here quoted from Crassus, every word which ends with a consonant is immediately succeeded by another which begins with a vowel; and, _vice versa_, if the preceding word ends with a vowel, the next begins with a consonant.]: for here the mere order in which the words are connected, produces a harmony of sound, without any visible attention of the Speaker. When the ancients, therefore, (I mean _Herodotus_, and _Thucydides_, and all who flourished in the same age) composed a numerous and a musical period, it must rather be attributed to the casual order of their words, than to the labour and artifice of the writer. But there are likewise certain forms of expression, which have such a natural concinnity, as will necessarily have a similar effect to that of regular numbers. For when parallel circumstances are compared, or opposite ones contrasted, or words of the same termination are placed in a regular succesion, they seldom fail to produce a numerous cadence. But I have already treated of these, and subjoined a few examples; so that we are hereby furnished with an additional and a copious variety of means to avoid the uniformity of cadence above-mentioned; especially as these measured forms of expression may be occasionally relaxed and dilated. There is, however, a material difference between a style which is merely _numerous_, (or, in other words, which has a moderate resemblance to _metre_) and that which is entirely composed of _numbers_: the latter is an insufferable fault; but our language, without the former, would be absolutely vague, unpolished, and dissipated.