The relationship between pressure and voltage output (calibration curve) of the gauge tube is not linear. The scale of the x axis ( pressure) is logarithmic and the scale of the y axis (voltage) is linear

The DV-6 thermocouple gauge tube's accuracy is +/- 0.2 mvolts (the DV-4 's is +/- 0.3 mvolts) But when converting that to pressure,the x-axis's scale is logarithmic. Therefore you can interpolate individual accuracies based on the +/- 0.2 mvolts for each point on the calibration curve. As the pressure increases the accuracy decreases because the 2 mV is ""stretched"" numerically on the log scale

For example, +/- 0.2 mV at 10 mTorr is +/- 1 mTorr (10% reading) but at a pressure of 500 mTorr the +/- 0.2 mV is 500 +/- 100 mTorr (20%) of reading.

If you look at the analog dial face of the VT-Meters, you can see that at the lower pressures the resolution is much better than the higher pressures. The rule of thumb, is if you make a pressure reading on the dialface, the accuracy of that reading is about the width of a pencil, centered over the reading