KEY WORDS : Australia, Adelaide, Glaisher stand, Stevenson screen, Climate Change.

 This short note presents seasonal time series for the 1888-1946 period of an Adelaide experiment comparing Glaisher stand and Stevenson screen temperature data examined by Nicholls et al (1996). Site changes and a station move for the Adelaide experiment may account for discontinuities in the seasonal trends which make the data unrepresentative as a test of comparative thermometer exposure. Research directions are suggested that might cast light on these problems in the data.

 Nicholls et al (1996) have brought to light records pertaining to meteorological equipment from colonial times for Western Australia, Queensland and South Australia.

 Nicholls et al give a brief account of the 1887-1948 experiment comparing temperatures measured in a Glaisher stand against those from a Stevenson screen and conclude that, "Over the year, the mean temperatures were about 0.2°C warmer in the Glaisher stand, relative to the Stevenson screen." For their source data they rely on two unpublished reports, Richards et al (1992 and 1993), which are essentially records of numerous statistical exercises carried out on the Glaisher / Stevenson data by students at Swinburne University of Technology in Melbourne. These reports do not contain an adequate analysis of the possible causes of significant discontinuities in the time series despite Richards et al (1993) containing a summary detailing at least six screen changes and a station move, events which in most cases correspond with discontinuities in the temperature data.

 It is this writer’s view that the Richards et al (1992 and 1993) reports are not adequate supporting references under the circumstances and that Nicholls et al (1996) should have submitted their data for review as a new project.

 The experiment commenced in 1887 and records exist for the dual exposures until 1948, which leads Nicholls et al (1996) to make the point that "This period of comparison far exceeds other comparisons between stands and screens, as reported in Parker (1994)." Perhaps the long period of this comparison is less of an advantage given the numerous site and equipment changes. Could it be more likely that interest in the experiment only lasted a decade or so and then it simply "ran on" because nobody took a decision to cease the dual readings? It is thus possible that factors such as differential equipment deterioration could have set in causing some of the trends seen in the difference time series. Nicholls et al (1996) present no time series showing trends over the 61 year period and make no mention of equipment changes or site variations.

 Richards et al (1993), in Appendix 5 records a four page summary of station history supplied by Dr. N. Nicholls and compiled from BoM records and archives, which details as best as the old records will permit, at least six screen changes over the period with at times more than two sets of instruments recording concurrently and a site change to new offices in 1944. In this summary the old style thermometer enclosure is referred to as a "Greenwich stand". Richards (1993), on page 14 refers to the four page summary as "...vague and ambiguous, and did not make clear when and where equipment was moved." Climatologists who have researched century old records would not be surprised at this. However, readers of Nicholls et al (1996) get no hint of the inhomogeneities related to these site changes that can be seen in both the maxima and minima difference seasonal time series shown below.

 The maxima graph shown in Fig. 1, reveals discontinuities in the data at the time of four of the instrument changes, which might have been enough to impress the Richards et al (1993) team that external non-climatic forces were impacting on their data at specific times. In particular the 1898, 1910, 1925, 1938-39 and 1944 site changes appear to be reflected in abrupt changes in the course of the some of the seasonal traces. However Richards et al (1993) concluded that the variations in monthly maximum differences could "...be seen as randomly scattered."

 In the case of the minima differences, Richards et al (1993) are able to draw attention to the gross aberration in the traces circa 1937-47 which as they point out is due to a drastic decline in the Glaisher readings from August 1938 and relate this to "...some unknown influence...".

 Nicholls et al (1996) have not presented the data and associated historical records that would be required to support valid conclusions as to the Glaisher / Stevenson difference revealed by the Adelaide experiment. The onus is on Nicholls et al to do more with this data and related records if their findings are to be considered along with those of Parker (1994). At the very least there is a need to compare these data to other temperature records from the Observatory as well as records from other sites.

Hughes,W.S., 1995. Comment on D.E.Parker, ‘Effects of Changing Exposure of Thermometers at Land Stations", Int. J. Climatology, 15, 231-234.

Nicholls,N., R.Tapp, K.Burrows and D.Richards, 1996. Historical Thermometer Exposures in Australia. Int. J. Climatology, In Press.

Parker,D.E. 1994. ‘Effects of Changing Exposure of Thermometers at Land Stations’, Int. J. Climtology, 14, 1-31.

Richards, D., G.Wilson, K.Sheng How, S.Kang, A.Tan and S.Chueng, 1992. Comparison of Temperature Measures. Report for Bureau of Meteorology Research Centre, Swinburne University of Technology, November 1992, 176 pp.

Richards,D., G.Nind, K.Ramchand and P.Keyhoe, 1993. Differences in Temperature Recordings between the Glaisher Stand and Stevenson Screen. Report for Bureau of Meteorology Research Centre, Swinburne University of Technology, October 1993, 123 pp.

 The extra detail Warwick Hughes provides on the Adelaide comparisons of thermometer exposure confirms the conclusions of Nicholls et al. (1996). Warwick's figures show that mean temperatures in the Glaisher stand are biased relative to the modern Stevenson Screen. So, as noted in Nicholls et al. (1996), 19th century temperatures, which were often measured with open exposures such as Glaisher stand, are biased warm relative to 20th century temperatures.