[13] presented a paper for the design of experiment to analyze gauge capability. The same designed experiment was used to test and compare the variation for two automatic measurement systems. Three metrology factors were used as a starting point and the factors were compared at two levels. Test results were analyzed using the expanded ANOVA method [13] in Minitab. Results discussion ensued and the conclusion showed that even automatic measuring systems can significantly vary in variability especially when subjected to different factors. [13] also focused only on automatic measurement systems but here, the same designed experiment is used for the GR & R study on manual measurement systems such as a vernier caliper shown in Figure 1 .

Minitab the application used here for the analysis as statistical tool has a built-in process for randomization. Therefore, the was no need for doing it manually.

Replication means repetitions of an entire experiment or a portion of it, under more than one condition. This is demonstrated by running the experiment ten times by three operators at two levels.

The idea is to arrange similar experimental runs into blocks or groups. This was achieved by first creating two large blocks of two manual measuring systems. These blocks were then divided into smaller blocks representing the three operators. The smaller blocks were further sub-divided on two levels.

The experiment investigated the effect of three parameters on measurement uncertainty of the automatic equipment at two levels. It was designed to investigate the effect of three factors on two levels of measurement uncertainty using the full factorial design with 10 replicates. The designed experiment in [13] was adapted to suit the manual measurement system. Therefore, the is no design of experiment in this paper.

Parts/Measurands:

1) Diameter of the large circle

2) Diameter of the medium circle

3) Diameter of the small circle

4) Length of the connecting rod

Factors:

1) Manual and

2) Digital

Manual Instrument:

1) Vernier Caliper

Number of Runs

The number of runs would be:

$\left(4\text{parts}\right)\times \left(3\text{operators}\right)\times \left(10\text{repeats}\right)\times \left(2\text{gauges}\right)=240$ (1)

(measurements):

Three operators measured the three circle diameters (Ø) and the length of the connecting rod (Con-rod) with manual Vernier Caliper and digital Vernier Caliper. Where X is mean and R is range (difference between max and min data values) ( Tables 1-4 ).

Analyses were done with the ANOVA method for GR & R studies in Minitab application where X is the mean and R is the difference between max and min data values.

The result for digital and manual Vernier Caliper for part 1 are analyzed and presented in Figure 4 and Figure 5 respectively.

The result for digital and manual Vernier Caliper for part 2 are analyzed and presented in Figure 6 and Figure 7 respectively.

The result for digital and manual Vernier Caliper for part 3 are analyzed and presented in Figure 8 and Figure 9 respectively.

The result for digital and manual Vernier Caliper for part 4 are analyzed and presented in Figure 10 and Figure 11 respectively.

The 3 operators measured a select part using a manual and digital Vernier caliper with 10 replicates. In a typical nested gage study, 10 parts, 3 operators and 2 replicates with a total of 60 measurements are usually the standard [5] [15] and the degree of freedom for error equals 30, which allows one to have about 90% confidence in estimating the repeatability within 20% of the true value [16] . Under this typical setting, the estimate for repeatability is reasonably precise.

In this paper, four parts were chosen. 10 replicates were then divided into 2 groups of 5 parts using the principle of blocking. The number of replicates was increased to 10 instead of the standard 2 replicates using the principle of replication to cater for the shortfall in the number of parts. Details of this experimental set-up are given by [13] . Therefore, to mimic a typical gauge study, the true representation of the runs was:

5 similar parts × 2 replicates × 3 operators × 2 gauges = 60 (2)

This was done for the 4 different parts; therefore, the total number of runs was:

60 × 4 = 240 (3)

A total of 240 measurements were done using blocking and replication principles compared to the usual 60 measurements.

A summary of these results is thus ( Table 5 ):