(2) The gathering of data holds a key position in the scientific

method and in metrology.

In measurement, most data are numerical values

taken from a calibrated dial or scale of some sort, and usually by means of

some interpolation.

Certain scales are easier to read than others.

The

National Bureau of Standards recommends (for electrical meters) a knife edge

pointer with the thickness of the knife edge equal to that of the division

lines, and preferably of a size not more than one-tenth the distance between

two successive scale division marks. The scale division marks should be at

least 1 mm apart. Such a scale equipped with an antiparallax mirror can be

estimated to one-tenth of a division.

Certain types of scales can be

equipped with a vernier to make possible a finer reading.

The 10/1000

micrometer is an example of splitting the division into tenths.

Certain

clinometers have a vernier, which makes possible the splitting of degrees

into minutes, each of which is one-sixtieth of a division. An outstanding

example of the vernier principle is found in optical proving rings, where

divisions one one-thousandth of an inch apart can be read to one one-

hundredth part of a division.

(3) Observational errors usually result from misreading the scale,

failure to eliminate parallax, and failure to estimate fractional divisions

correctly.

f. Data.

(1) Raw data.

Raw data are of little value.

Before their full

meaning is apparent, the conditions under which they were obtained must be

known and they must be recorded, manipulated statistically, interpreted and

evaluated, and put into a suitable form for presentation.

They should be

recorded as soon as possible to avoid the danger of forgetting. Unless a

prepared form is available, the data should be clearly labeled to include

all pertinent circumstances.

Raw data are seldom of much value until all

systematic errors are allowed for and the random errors averaged out by some

statistical process. When it is finally determined to be the best guess of

the value in question, there is still the problem of how good is the guess.

In other words, what is the probability that it is completely correct.

(2) Evaluating data.

Data is nothing more or less than just plain

information. Data may be true or false, accurate or inaccurate. The job of

the calibration specialist is to obtain data and to evaluate its validity.

(3) Recording data.

(a) General.

Data should be recorded as soon as possible.

Objective information left to memory, even for a short while, starts to lose

its objectivity.

Integrated Publishing, Inc. |