The normal probability paper test is rough but very convenient, which can be directly on the sample to give an approximate evaluation.

Figure 3 is the rise (fall) time samples were taken from the normal probability paper test chart Log function transform, known from test results, as is the logarithmic transformation can be broadly distributed in the fitting line near, can be considered approximately the normal requirements, a further test value, the other two parameters are also obtained the same results.

Assume that G represents the function which transforms the probabilities to the probability paper; that is, all points ([x.sub.m];G([p.sub.m])) fall on the same straight line which represents the cumulative distribution function (CDF).

Plot next the points ([x.sub.m];G([p'.sub.m])) on the same probability paper. Here, values are the same as those above but the probabilities [p'.sub.m] represent Blom's [6] plotting position (m - 0.375)/(n + 0.25), which have been developed for normal distribution.

Geometrically, this simply means that, to improve the fit, the straight line on the probability paper, resulting from linear regression, is replaced by another straight line.

The medians were plotted against the measured pores on log-normal probability paper. This graph will produce a straight line on log-normal probability paper if pore sizes have a log normal distribution.

The 30 measured pore sizes were drawn against the median ranks on log-normal probability paper (Fig.

Probability paper was used, according to the method described by Harding (6), to identify the young-of-the-year cohort from the length-frequency histograms.

(1) Using the Weibull probability paper, we plot the estimated values of F(t) = F([t.sub.i]) = i/(N + 1) against [t.sub.i] and determine [Beta] and [Eta] from these plots.

(3) In the present paper we will show how a spreadsheet-based analysis is so straightforward and clear that there is no need to plot on probability paper, etc.

This can easily be accomplished in a spreadsheet format, where this trial-and-error procedure is quite convenient compared to when the data are manually plotted on a Weibull probability paper.

The scatter in the mechanical properties of GFRP rods was investigated by two-parameter Weibull function because the two-parameter Weibull distribution function represents a straight line in the Weibull

probability paper with slope ([Alpha]), which is the inverse measure of dispersion in the experimental data.