²⁹  Sufficient perhaps to prompt some to question their validity. For Table 5’s original prevalence of scratch dials to be considered suspect, argumentation and evidence (outweighing that presented) is required to both, overturn the presumed universality of scratch dial diffusion, and indicate scratch dial loss to be overstated. This dual requirement is a logical consequence of the methodology employed. It is interesting to note Ovens and Sleath, op. cit., in the introductory discussion to their recording of Rutland’s scratch dials, favour the presumption they were originally present on all churches, and intuitively speculate less than 30 per cent have survived.
   ³⁰  Winzar’s photographs, line drawings, measurements of dimension, line angles and position constitute a benchmark unsurpassed nationally. His thoroughness is exemplified by the fact that on a quarter of the churches surveyed he has recorded the most dials (Annex).
   ³¹  Church limewashing and dial painting renders the estimated original prevalence of multiple dials less inexplicable. Only one or two would be in active use, the rest being limewashed out.
   ³²  More complex categorisation (based for example on gnomon hole – between stones or separately drilled, lines or pock marks, presence of a circle) has appeared in the literature. Horne 1929, op. cit., has twelve. There are severe practical and methodological difficulties with such taxonomies. The more multivariate the greater the challenge of defining a taxonomy that is both exhaustive and mutually exclusive i.e. capable of statistical analysis. Moreover the taxonomies developed in the literature are biased towards surviving, rather than original, dial appearance. A simpler categorisation orientated toward broad original, rather than the detailed vagaries of surviving, appearance, offers the prospect of more meaningful insight.
   ³³  Additional counties, especially those near to Kent, need to be examined before the/any regional context can be determined. Kent also appears to have far fewer dials with scratched enumeration around their edge compared to other counties.
   ³⁴  Half and circle dials share a common maximum dimension – the diameter of a circle.
   ³⁵  Ovens and Sleath, op. cit., report similar results for Rutland: 76 per cent with mass, and 84 per cent with noon lines. Compared with this paper their figures contain offsetting effects – the inclusion of quarter dials and the adoption of ±10º tolerance.
   ³⁶  Missing mass and noon lines can be attributed to the weathering out of the scratched line or it being painted only, not also scratched, on the original dial. None hour lines can be attributed to badly set out dials, walls badly off east-west alignment or the time/event indicator for various activities undertaken by the priest.
   ³⁷  Although the radial lines of individual scratch dials have been examined in the literature, systematic statistical consideration is rare. An exception is C.M. Lowne, ‘An Analysis of Some Mass Dials of Sussex and Kent’ The British Sundial Society Bulletin, Vol. 97(ii), April 1997, analysing 70 scratch dials from East Sussex and south-west Kent. Comparison is hindered by differing methodologies (a single stage discriminatory analysis versus this paper’s staged segmented approach) and sample size (only half Winzar’s) both contributing to over a quarter of Lowne’s dials being statistically indistinguishable by time system. As however two-thirds of Lowne’s distinguishable dials conform to this paper’s Fig. 2, the broad thrust of his results is not inconsistent with this paper.
   It is appropriate to note L.G. Welland ‘Scratch Dials and a Suggestion of their Working’ Bygone Kent, 1981 principally because of Welland’s contribution to listing Kent scratch dials. However the paper is neither definitive nor conclusive, believing ‘… the whole puzzle may be insoluble’.
   ³⁸  See C.H.K. Williams ‘Charing Clocks, Clockmakers and Clockkeepers’ Archaeologia Cantiana, CXXV, 2005, and C.H.K. Williams, ‘Seventeenth & Eighteenth Century Clock Demand, Production and Survival: An Economic and Statistical Analysis’, Antiquarian Horology, Vol. 28, March 2005.