GEMOLOGY

MEASUREMENT
TOLERANCES
Accuracy and Precision in the
Gem Industry
BY GIA RESEARCH

F or many applications measurement is fairly

straightforward: we use the appropriate tool to
measure the object or substance of interest.
Whether it is the length of a board, the amount of
salt to add to a recipe or the dimensions of a gemstone,
them may be incorrect. In this article we will share
some of the measurement issues we deal with on a daily
basis in the Gemological Institute of America (GIA)
Gem Laboratory.
All measurements have some uncertainty associated
we often measure things with them. Uncertainty
without giving much means something not ex-
thought to the inherent actly known; in the case of
uncertainty of the process. measurements, it is often
Often this uncertainty is described as the error associ-
inconsequential, but in ated with the measurement.
some cases a high degree of Measurement uncertainties
precision and accuracy is can affect the accuracy of a
required.We need to know measurement and they are
that the measurement fundamental in determin-
reflects the true dimensions ing its precision. Accuracy
of the object measured and describes how closely a
that if we were to measure given measurement matches
the same item again we the “true” measurement.
would obtain the “same” Precision describes how
measurements for that finely the measurement has
item. Gemstones are just been made, but not whether
such a case. Since measurements are often used to the value is accurate. Precision tells us something about
identify a particular gemstone and in some cases — like
diamonds — to predict or assess features such as the Above: These three measuring instruments offer different levels of preci-
sion as shown on the scale of each one. The pearl gauge can be read to
overall cut quality, the reliability of these measurements ±0.05 mm, the slide micrometer to ±0.01 mm (or 0.0003 inches) and the
is critical. Otherwise, the conclusions we draw from digital screw micrometer is precise to ±0.001 mm.

Rapaport Diamond Report April 1, 2005 183

the scale of a micrometers and
measurement and table gauges are
whether a second used.Each of these
measurement measurement tools
matches the first has its own tested
one,while accuracy relates accuracy and preci-
to the correctness of it.Precision is important sion, but the levels of
because two measurements can only be determined to accuracy and precision
be the same or different with respect to some reported can degrade over time.The regular
precision. A generally accepted precision for a specific process of calibration allows the GIA Gem Laboratory to
measuring tool or process is often referred to as its toler- ensure that the measuring tools used are always operating
ance [e.g., a ruler marked every 1 millimeter (mm) has a at their highest levels.
tolerance of ±0.5 mm]. An example of this process is the calibration and regular
Even in the absence of operator error, measurement verification of optical noncontact measuring devices —
uncertainty originates from the precision of the tools Sarin or OGI devices.To carry out this process, one must
used to measure and the fact that measuring is not a obtain a set of calibration objects — gemstones — and
perfect process. measure them as accurately as possible on higher-precision
measuring instruments. Once these gems have accurate
GEMSTONES and precise measurements, we can use them to verify the
Accuracy, precision and the inherent uncertainty of accuracy of the optical measuring devices by assessing how
measurements affect the world of gemstones in many ways. close repeated measurements of the calibration gemstones
Consider these points: are to the known measurements. Discrepancies can be
•Measurement tools for gemstones such as scales, used to calculate adjustments to the measuring devices;
micrometers and optical measuring devices need to be after making such adjustments (as indicated by the man-
calibrated regularly to provide the most accurate ufacturer),the devices are tested again.When the repeated
measurements possible within their stated tolerances. measurements fall within the accepted tolerance of
•Reports of measure- the device, the device is
ments should include considered calibrated.
t h e i r precision. False These same calibration
levels of precision should stones are measured
be avoided. regularly to verify that
•The known tolerance the devices remain in
for each measuring tool good working order.
should be factored into
c o n s i d e r a t i o n o f a ny IDENTIFICATION
measurement taken. When a measur ing
In the GIA Gem Labo- tool is properly cali-
ratory, the accuracy and brated,the measurements
precision of measurements it produces can be used
affect the daily process of reliably to accurately
assessing gemstones.These effects can be broken into identify a particular gemstone. However, it is important to
roughly three separate, but related, areas: calibration, be aware of the precision and tolerance of the tool.
identification and quality assessment. Consider a report that shows measurements for a
gemstone of 6.02 – 6.06 x 3.66 mm. A jeweler wants
CALIBRATION to verify that the gemstone she has is the same as that
An integral part of the GIA Gem Laboratory’s process- listed in the report so she measures it using a calibrated
ing of goods is the accurate measuring and recording of
Above and opposite page: The precision and repeatability of optical
various properties of the gemstones it receives. High-pre- measuring systems, like these models from Sarin and OGI, are found by
cision gem scales, optical noncontact measuring devices, analyzing repeated measurements for several gemstones of different sizes.

184 Rapaport Diamond Report April 1, 2005

measuring tool and gets proportion value
6.04 – 6.07 x 3.64 mm. without remeasuring
For this example, the all proportion values.
uncertainty of these If a diamond is
diameter and depth remeasured and the
measurements — that is, new measurements
the tolerance of the meas- are found to be within the
uring device for these tolerance of the device used, those
dimensions — is ±0.01 mm. measurements are considered “the same” as the
Recall that two values that are first set and, therefore, there is no need to change any of
different, yet are still within the the values. If one or more values are found to be outside
stated tolerance of the measuring device,cannot reliably be the measuring tool’s tolerance, then the new measure-
distinguished from each other.The two measurements ments can be considered different and one or the other
are within tolerance (they match) if the difference complete set of values needs to be chosen.The diamond
between them is up to twice the uncertainty (here, should be measured additional times to help determine
0.02 mm).Therefore, in the example above, even though which complete set of values was the most accurate.
the measurements look different, they can represent the
same gemstone since each pair of values is no more than CONCLUSION
0.02 mm apart. However, if the jeweler measured the Accuracy and precision are both important parts
gemstone as 5.98 – 6.05 x 3.68 mm, then she could reli- of the measuring process. The accuracy of a meas-
ably suspect that it was not the one reported because the urement tells us how close the measured value is to
difference between one pair of values (5.98 versus 6.02) the true value. However, measuring is never exact.
was greater than 0.02 mm,the stated tolerance of the device. Each measuring tool has its own level of achievable
precision that tells how finely it is capable of measuring
QUALITY ASSESSMENT the particular quantity.
The measurements of a gemstone’s dimensions and A measured value is not just a number, but has some
proportions can also be used to assess other attributes of error or uncertainty associated with it.When two meas-
that gemstone. GIA’s ured values differ by less
research into diamond than their combined
cut has shown that the uncertainty, they are
proportion values of a said to be within toler-
round brilliant diamond ance of each other with
can be used to predict respect to that measur-
and assess the overall cut ing tool.Two measured
quality of that diamond. values are only reliably
Again,knowledge of the different from each
accuracy and precision other when their differ-
of the measuring tool is ence is greater than the
essential if we are to rely tolerance of the meas-
on these measurements uring device.
to predict cut quality. Measurements are
G I A re s e a rc h h a s used to identify gem-
found that all the facets in a round brilliant diamond stones and to assess other properties, including those
affect its cut quality;the complete set of proportions used leading to a cut quality assessment. Properly calibrated,
to predict cut quality must provide a realistic three-dimen- well-maintained measuring tools are necessary to achieve
sional model of a round brilliant diamond.This is one accurate measurements within the precision those tools
reason why, when requesting a measurement recheck provide. A complete, consistent set of measurements
of a single diamond proportion from the GIA Gem that reliably represents a given gemstone is vital for
Laboratory, it is no longer appropriate to change a single evaluating cut. ✦

Rapaport Diamond Report April 1, 2005 185