Abstract
Calculations based on a theory for a square guarded hot plate (Donaldson 1961) in which the metal plates on either side of the heater, the gap, the side insulation, etc., are taken into account show that the errors due to the form of construction and the operation of the apparatus fall into three categories:
errors due to the actual existence of the gap, i.e. the error that occurs when the side insulation is perfect and there is no temperature difference across the gap;
errors due to imperfect side insulation;
errors due to any temperature difference across the gap.
These errors have been computed for a range of square guarded hot plates and the results of these computations are discussed. It is found that the errors are reduced if the guard plate is reduced in width and also if the gap width is increased. It is shown that by choosing the edge insulation best suited to the sample being tested and the apparatus being used and by operating the apparatus within narrower limits than stipulated by the Standards, the overall error in thermal conductivity measurements could be reduced to less than 1%.