Cecil Sharp in Somerset: some reflections on the work of David Harker

Harker's Methods

As Sharp was a different kind of collector to his predecessors, so Harker was a different kind of critic. Lloyd had described his own book as one for "beginners not specialists" and it was clearly intended for a general audience, but Harker's criticism had the appearance of rigour and was aimed at academics. Where Lloyd offered opinions based on examples, Harker provided statistics: indeed, much of his authority seems to rest on the statistical evidence he offers. Statistics, however, are notoriously open to abuse. For this reason, it is usual for statisticians to offer some evidence about their method, but Harker fails to do this. We are repeatedly given figures and percentages without knowing the base on which they are calculated, and without having any means of checking their accuracy.

Harker does not define what he means by his comparative terms. When it is alleged that "Sharp skewed his selection of texts for publication so as to favour songs taken from small villages rather than larger, and from both in preference to towns" (Harker 1985, 195), a great deal must depend on what is being defined as a "small village," or a "large village," or a "town." Because Harker does not do this, let us turn to the 1901 Census. [3] He calls Langport a "small town," but it had 813 people, just over a hundred more than its neighbour, Huish Episcopi, which to Harker is a "large village," and substantially less than the 1,021 of Cannington, another "large village." "Haselbury"--presumably, Haselbury Plucknett--is also called a "large village" despite having a population of 470, less than half Cannington's; and High Ham, which Harker calls "tiny," had 898 people--more than the "small town" of Langport. Not only are we not told what Harker means by his terms, it is doubtful whether he himself knows. Table 1 provides some comparative figures.

Much of Harker's argument depends on statistical balance: on counting the songs Sharp collected in particular areas and comparing these numbers with the numbers published. In particular, Harker claimed to have found a gross imbalance between songs published from rural areas and those published from towns. But Harker's methods create a statistical imbalance in themselves, because he did not analyse all five volumes (or, as Sharp called them, "Series") of Folk Songs from Somerset, and failed to take account of the fact that they were published piecemeal over a period of four and a half years while collecting was in progress. Harker studied Sharp's work up to the end of August 1907, which was the approximate date for the completion of English Folk Song: Some Conclusions and the compilation of Series Four of Folk Songs from Somerset. [4] This was convenient for him in more ways than one, since by missing out Series Five he ensured the maximum possible statistical imbalance between the material collected in the "rural" Hambridge area--where Sharp began his work--and that collected elsewhere; it thereby distorted Sharp's record of "urban" as compared with "rural" publication. When Series One was compiled--all the songs were collected before the end of August 1904--Sharp had barely begun to emerge from Hambridge where he began. He did not begin collecting in Bridgwater (pop. 15,209), for example, until August 1905. Out of the twenty-seven songs published in Series One, fifteen were from Hambridge, but in Series Three, only one song came from there and in Series Four and Five none at all. But Sharp was continuing to publish material collected in Bridgwater. Four songs from that town appeared in Series Five, together with three from other towns such as Minehead and Wells. The omission of these helps Harker's argument, but does nothing to illustrate the realities of Sharp's publication policy.