In metrology, the reading error is that what happens when evaluating the relative position of the index of the measuring instrument with respect to the scale; this error is generally due to four causes:
- resolving power of the human eye: it is defined as the angle of minimum separation between two points that the eye is able to discern as two separate and distinct objects (it is about 0.1 mm which corresponds to 100 μm, however with many physiological variables);
- parallax error: due to the fact that the index and the scale of the measuring instrument are located on different planes (the operator’s gaze should always be perpendicular to the scale for a correct measurement). Parallax error is primarily caused by viewing the object at an oblique angle with respect to the scale, which makes the object appear to be at a different position on the scale. For example, if measuring the distance between two ticks on a line with a ruler marked on its top surface, the thickness of the ruler will separate its markings from the ticks. If viewed from a position not exactly perpendicular to the ruler, the apparent position will shift, and the reading will be less accurate than the ruler is capable of. In the context of reading a piece of volumetric glassware, such as a measuring cylinder, burette, or volumetric flask, the meniscus should be at eye level otherwise there will be an error in the reading. If the meniscus is above eye level an increased volume measurement will be made, conversely, if the eye is above the meniscus then a lower volume reading will be made. A similar error occurs when reading the position of a pointer against a scale in an instrument such as an analog multimeter. To help the user avoid this problem, the scale is sometimes printed above a narrow strip of mirror, and the user’s eye is positioned so that the pointer obscures its own reflection, guaranteeing that the user’s line of sight is perpendicular to the mirror and therefore to the scale. The same effect alters the speed read on a car’s speedometer by a driver in front of it and a passenger off to the side, values read from a graticule not in actual contact with the display on an oscilloscope, etc.
- interpolation uncertainty: when the scale of the measuring instrument is linear, it is of the order of ±10% of the distance between two successive divisions. If the scale is not linear, this uncertainty value can also increase considerably. To reduce the interpolation error, reading systems with nonii, micrometric screws and silicone scales can be used;
- background noise: it is the set of all those causes that impose index movements overlapping the displacement produced by the measurand. Regarding the estimation of the error value produced by the background noise, when it comes to appreciating the average value of the observed or recorded signal, it is admitted equal to ±10% calculated on the double amplitude of the oscillation.