44
was also increased. This second curve is the upper line in the figure and the resultant R2 for this fit is also better than 0.999. Both of these curves are within the 95% confidence intervals for the individual data points. A family of curves each with different values of k and KM can be fit to the data using Equation 4.2, each of which will closely approximate the data and result in a low error sum of squares. The curve which the non-linear regression predicts as a best fit may, therefore, be more a factor of random or systematic error in the data than an actual indication of the true value for k. This represents a serious limitation to the use of Equation 4.2 for prediction of nodule conductance even if Vmax is known.
Validation of Vmax Estimates for the 1984 Field Data
In Chapter II and VI estimates of Vmax from field studies using the reduced model with an assumed KM are reported. It has been demonstrated that this model can not be used to predict accurately the nodule conductance, k, if the KM is unknown or changing over the experimental treatments. Also the simulations reported in Table 4.1 indicate that depending on the degree of error in the assumed KM and the true value of the nodule conductance, the resultant estimation of Vmax from the reduced model might also be incorrect.
To test the validity of the Vmax estimates from the field studies the data from the 27 assays of intact hydroponically grown plants assayed at 0.1, 0.2 and 0.4 mm3 mm3 oxygen described above were used. The analysis procedure used with the field plants (see Chapter II) was repeated with these data such that Equation 4.2 was used to estimate a