quality compared to medieval scratch dials. See M. Bowen, ‘Saxon Sundial in the Parish Church of All Saints, Orpington’, and R.I. Page ‘Note on the Inscription’, Archaeologia Cantiana, LXXXII, 1967; also D. Scott, ‘Sundials in Anglo Saxon England, Part 4, The Late Period – Aldbrough and Orpington’, The British Sundial Society Bulletin, Vol. 12(i), Feb 2000.
   ¹⁹  T.W. Cole, Scratch-Dials on Churches. Interim List, 1934.
   ²⁰  Green, op. cit. Rosenberg taught at the King’s School Canterbury and read papers to the Canterbury Archaeological Society.
   ²¹  Welland states ‘a complete set of drawings in book form is under consideration’ and invited prepublication purchase. Did drawings exist for the entire list or was (intended) completion contingent on sufficient pre-publication interest?
   ²²  Welland cites no credits. Gerald Winzar informed him of his findings (Pat Winzar pers. comm.) and Welland’s listing includes almost all dials recorded by Winzar. It also includes all dials listed by Grove. That said, Welland’s compilation contains some 25 churches listed for the first time.
   ²³  For those concerned that three counties is too limited a sample, initial statistical analysis by the author of the (unpublished) British Sundial Society database for some dozen counties with complete to comprehensive surveys of eligible churches, indicates the three counties are not unrepresentative. See The British Sundial Society Bulletin forthcoming.
   ²⁴  Also supported by the correspondence of Welland’s 310 churches visited with 331 churches with medieval components. Whilst additional discoveries can never be ruled out, any such can be expected to be few and not materially affect the picture shown in Tables 2, 3 and 4.
   ²⁵  The other possibility – unknown survivals – does not arise for the Kent, Somerset and Rutland data under consideration (main text and note 24). If all churches have not been surveyed, relating listed scratch dials to the total eligible churches will result in an underestimate of survival. This is avoided if listed scratch dials are related to the number of churches surveyed (assuming the latter to be statistically representative). This is not always possible because of the natural tendency only to list positive findings. Statistically ‘non-finds’ are just as important! Non reporting of churches without dials is why the published partial surveys of Hampshire (Green op. cit.) and Herefordshire (Botzum op. cit.) are not included in Table 4. In contrast the Somerset and Rutland surveys are both complete and list churches without a scratch dial.
   Non reporting of churches without scratch dials also affects The British Sundial Society mass dial database. Fortunately such information can be retrieved and the database is in the process of being enhanced to permit statistical analysis (note 23).
   ²⁶  The appearance of scratch dials and the nature of incisions are not the work of professional stone masons. Most likely they were made by (or supervised by) the local priest, sextant or verger.
   ²⁷  Author’s analysis of Somerset (Horne 1917, op. cit.), Rutland (Ovens and Sleath op. cit.) Hampshire (Green op. cit.) and Herefordshire (Botzum op. cit.). Kent information has not been systematically recorded – but numerous moved dials are known. The cumulative impact of repairs, re-use and replacement is vividly detailed in J.F. Potter ‘Anglo-Saxon building techniques: Quoins of twelve Kentish Churches reviewed’, Archaeologia Cantiana, CXXVI, 2006. Two of these churches have scratch dials on the quoin stones – one obviously, the other possibly, moved.
   ²⁸  Horne (1917), op. cit., describes 19 per cent of Somerset dials as ‘doubtful’. Ovens and Sleath, op. cit., describe 8 per cent of Rutland’s as ‘very poor or faint’ and 42 per cent ‘poor’. Botzum, op. cit., describes 25 per cent of Herefordshire dials as variously ‘very weathered, eroded, remnant, flaking or indistinct’. Of the Kent dials recorded by Winzar 21 per cent are too indistinct to be categorised (Table 6).