evidences

The following point to the fact that the plural of 'evidence' is 'evidences'. The Collins Cobuild Dictionary for Advanced Learners also states that 'evidences' is the plural noun. I think it is a British English usage.

1. Includes (fols. 1-2) a draft of his preface to A defence of the Christian revelation on two... points... contained in... Observations on the history and evidences of the Resurrection of Jesus Christ, by G. West; and... Observations on the conversion of St. Paul, by G. Lyttleton (London, 1748), with miscellaneous papers and transcripts (c.1700?) of material of antiquarian interest, c.1620-1658, n.d., guarded onto the stubs of the pages of A Sermon [on Psalm cxliv] preached before the Incorporated Society for the Propagation of the Gospel in Foreign Parts... on February 24, 1758; by James  [Johnson] Lord Bishop of Gloucester (London, 1758). 

2. For just as the archæologist, when he excavates the site of some ancient city, finds the various forms of its civilization arranged in chronological strata, so we find evidences of each past generation and its activities in the superimposed strata of our vocabulary.[1]

My conclusion is it is used both by in American and British English. I don't think Thomas Sowell has used the word wrongly.

Yes, both are fine; you might want to make it absolutely clear that a person had not simply edited a book, for instance.

To return to the earlier question, "evidences" might be used where you wanted to refer to individual items of evidence. Many seemingly non-countable abstract nouns can be used in this way; thus "felicity" is the state of happiness, but "felicities" might be individual instances of happiness. However, it might appear mannered.

MrP