The second term in the above model equation stems directly from the definition of P (see ISO 10360-1:2000,9.3 and ISO 10360-2:2001, 5.2.4).The third and fourth terms highlight the form deviations of the test sphere. If the test were run with a perfecttest sphere, the form error, ε form, would be zero, and the value of the probing error obtained would be Ppt,which is indeed the value sought.No other uncertainty contributors are expected for the probing error, P, in normal circumstances.Due to the simple structure of the error model, the standard uncertainty of the probing error is simply thestandard uncertainty of the contributor ε formu(P) = u(ε form) (A.4)The recommended expression for the uncertainty caused by the form error is( ) ( )22form 2u ε = ⎛ F ⎞ + u F ⎜ ⎟⎝ ⎠(A.5)whereF is the form error reported in the calibration certificate;u(F) is the standard uncertainty of the form error stated in the calibration certificate.NOTE 1 The form error is always positive and hence cannot have a mean of zero. A rigorous application of the GUMwould require a correction of the systematic error (see GUM:2003, F.2.4.5), which is not indicated in ISO 10360-2 and isusually impractical. The above equation assumes no correction, but overestimates the first term slightly. However, thevalue of F is usually small in practice, and the overestimation is unlikely to affect the test significantly.NOTE 2 The highest (peak) and lowest (valley) point pairs, which dictate the probing error, P, and the form error, F, donot necessarily coincide because of probing effects, as well as the limited number of sampling points. As a consequence,the effect of the form error on the probing error, P, can be smaller than the calibration value F/2.A.2.2 List of other possible contributorsThe recommended model in A.2.1 is likely to be adequate in most circumstances. However, the tester is urgedto investigate whether more uncertainty contributors are relevant in his specific application.The following contributors may be relevant in some applications.⎯ Fixturing of the test sphere. If the test sphere is fixtured loosely or vibrations are present, the testsphere may shift during the measurements, e.g. due to probing forces, vibrations and inertial forces. Firmfixturing of the test sphere usually makes this contributor negligible.⎯ Bending of the test sphere stem. If the test and reference sphere stems exhibit different rigidities, thebending due to probing forces may cause errors. Commercially available test spheres usually make thiscontributor negligible.