Parallelism is measured by comparing one surface or axis against a reference (datum) and recording the total variation across the feature. The most common shop-floor method uses a dial indicator swept across the surface while the part sits on a datum, and the difference between the highest and lowest readings is your parallelism error. More advanced methods involve coordinate measuring machines, optical flats, or specialized fixtures depending on the feature type and tolerance required.
What Parallelism Actually Means
Parallelism is an orientation tolerance. It controls how closely a surface or axis stays aligned to a reference datum at a constant distance. The tolerance zone is two parallel planes spaced apart by the value called out on the drawing. As long as every point on the controlled surface falls within that zone, the part passes. The zone can shift up or down (translate), but it cannot tilt relative to the datum. This makes parallelism a specific case of angularity, just locked at 0 degrees.
Because parallelism is always referenced to a datum, you cannot measure it in isolation. You first need to establish the datum surface, then measure the controlled feature relative to it. Choosing the right datum matters: it should be a functional surface that the part actually mates against or locates from in assembly. The datum acts as an anchor, restricting degrees of freedom so the measurement is repeatable.
Measuring Surface Parallelism With a Dial Indicator
This is the most common method in machine shops and inspection rooms. You need a surface plate, a dial indicator (or lever-type dial gauge), an indicator holder or stand, and the part itself. If the part doesn’t sit stably on its datum surface, you may also need adjustable jacks or a fixture to level it.
Place the part datum-side down on the surface plate. The surface plate becomes your reference plane since it’s flat to within a few millionths of an inch. Mount the dial indicator in its holder so the contact tip touches the surface you’re checking for parallelism. Zero the indicator at one end of the surface, then slowly sweep it across the entire feature, tracking the needle as it moves. Record the maximum and minimum readings. The difference between those two values is the parallelism error.
For example, if your indicator reads +0.002″ at one spot and -0.001″ at another, the total variation is 0.003″. If the drawing calls out a parallelism tolerance of 0.005″, the part passes. This approach works the same way a flatness check does, with one key difference: the part must be fixtured against the datum, not just placed arbitrarily. If you skip that step, you’re measuring flatness, not parallelism.
Getting a Clean Reading
Before measuring, wipe down the surface plate and the datum face of the part. Even small debris between the part and the plate will tilt the part and skew your results. Run the indicator slowly and at consistent pressure. If you’re using a lever-type gauge, keep the contact angle steady as you sweep. Take readings at multiple paths across the surface, not just one line, especially on wider parts where twist could hide along a single track.
Measuring Parallelism Between Two Axes
When a drawing calls out parallelism between two holes or shafts, the measurement gets more involved because you’re comparing centerlines, not flat surfaces. The general approach uses precision pins, mandrels, or expanding arbors inserted into the holes to physically represent each axis.
Start by establishing the datum axis. Insert a precision mandrel into the datum hole and fixture the part so that mandrel sits in V-blocks or rests on a surface plate. Then insert a second mandrel into the controlled hole. Using a dial indicator, take height readings at two points along the length of the second mandrel, as far apart as possible. The difference in those readings, divided by the distance between them, gives you the angular deviation. The parallelism error over the full feature length is that angular rate multiplied by the feature’s actual length.
For higher precision, some shops use expanding collets or adjustable ring assemblies that center themselves inside the bore. Each adjusting element is set so it contacts the bore wall uniformly (typically held within 0.01 mm). A test bar then passes through both assemblies, and indicator readings at each end of the bar reveal the misalignment between the two axes.
Using a Coordinate Measuring Machine
A CMM automates and dramatically improves the accuracy of parallelism measurement. The machine probes multiple points on the datum surface to establish a reference plane, then probes the controlled surface and fits a plane through those points. The software calculates how far the measured plane deviates from perfect parallelism to the datum.
The most widely used fitting method is total least-squares, also called orthogonal distance regression. It finds the best-fit plane by minimizing the squared distances from all probed points to that plane. This is the default in most CMM software. A second approach, called minimum zone, finds the two closest parallel planes that contain all the measured points. The minimum zone result is typically a smaller number and more closely matches the tolerance zone definition on the drawing, but it’s computationally more complex and not always the default setting.
If you’re running a CMM, pay attention to which algorithm your software is using. A total least-squares fit can sometimes pass a part that a minimum zone evaluation would fail, or vice versa. For formal inspection reports, check whether your customer or drawing standard specifies one method over the other.
Probing Strategy
The number and distribution of probe points directly affects your result. Too few points and you can miss high or low spots. Spread points across the full surface, concentrating slightly more near edges where form errors tend to be largest. For axis parallelism on a CMM, probe each bore at multiple cross-sections along its length so the software can construct a reliable cylinder axis.
Optical Methods for Precision Work
For extremely tight tolerances, optical flats and monochromatic light provide parallelism measurements down to fractions of a wavelength of light (on the order of millionths of an inch). When two very flat surfaces are placed close together with a slight wedge gap between them, light reflecting between them creates visible interference fringes, called Fizeau fringes.
The spacing and straightness of these fringes tell you about the surfaces’ relationship. Evenly spaced, straight fringes mean the surfaces form a uniform wedge. If the fringes curve or vary in spacing, one or both surfaces have form error. The fringe spacing relates directly to the wedge angle: each fringe represents a half-wavelength change in the gap. By counting fringes across a known distance, you can calculate the angle between the surfaces and determine how far from parallel they are.
To set up, clean both surfaces with compressed gas to remove dust particles. Press the optical flat against the test surface and adjust until the fringe density is fairly low (wider fringes mean a smaller wedge angle, meaning the surfaces are closer to parallel). Photograph the pattern with manual exposure settings for analysis. This method is practical mainly for small, lapped components like gauge blocks, sealing surfaces, and optical elements.
ISO vs. ASME: How Standards Differ
If you work with international drawings, be aware that ISO and ASME standards interpret geometric tolerances differently in some situations. The core concept of parallelism is the same, but the default rules around it are not.
ASME Y14.5 applies stricter default provisions. Rule #1 (the envelope principle) and the simultaneous requirement mean that features are controlled more tightly without needing extra callouts. ISO GPS uses the independency principle by default, meaning each feature is considered separately unless the designer adds modifiers like “CZ” (common zone) to link them. In practice, this means two coplanar surfaces on an ASME drawing are automatically treated as one plane for tolerance purposes, while an ISO drawing treats them independently unless marked otherwise.
These differences affect how you set up your measurement. On an ASME drawing, you may need to verify parallelism across multiple surfaces simultaneously. On an ISO drawing for the same part geometry, you might measure each surface on its own. When in doubt, follow the standard called out in the drawing’s title block.
Common Error Sources
Parallelism measurements are sensitive to several factors that can inflate or mask the true error. Temperature is the most significant for precision work. Metals expand and contract with temperature changes, and if the part and the measuring equipment are at different temperatures, the results will drift. Standard metrology practice calls for a controlled environment at 20°C (68°F), typically within plus or minus 2°C. Let parts soak at room temperature before measuring, especially if they just came off a machine or out of a shipping container.
Vibration is another factor. Floor vibrations from nearby machinery can cause the indicator needle to bounce, making it hard to read true peaks and valleys. Granite surface plates help dampen this, but in noisy shop environments, you may need to isolate the measurement station or take readings during quieter periods. Probe errors on CMMs, including stylus deflection, worn tips, and inconsistent contact force, also contribute uncertainty. Recalibrate the probe regularly and replace worn styli before critical inspections.
Finally, fixturing errors are easy to overlook. If the datum surface isn’t seated firmly and cleanly against the reference (surface plate, chuck, or CMM fixture), every measurement taken from it will carry that initial tilt as a built-in error. Always verify the datum setup before moving on to the feature measurement.