Close Window


Written, with help from students and clients,  by James D. Meadows
The Tolerancing Newsletter August, 2007

Subject: Re: Question on Measuring Projected Tolerance Zones

Mr. Meadows,
My Name Is Stacy. I’m from an aerospace company and I have a question.
Can you measure a projected tolerance accurately using a threaded hole?
The engineers @ Goddard Space flight center seems to think so...
Is there a method to check this?

Please advise.


Projected tolerance zones are most often used on threaded holes, and yes, they are easy to measure. I detail these methods in my text books. A short description would be to get threaded gage plugs that are split up the middle, so that they contract as you thread them into the holes. This contraction will assure that the threaded gage plug takes on the location and angle of the threaded hole's pitch diameter. A CMM probe can then be used to probe the plug at a circle where the plug is closest to the threaded hole outside of the hole. Then another circle is probed at the height of the projection. If both circles reap a center that is inside of the position tolerance zone, the hole has met its projected positional tolerance zone requirement. This variables data collection approach will tell you how far the projected axis of the hole's pitch cylinder has strayed from perfect (true) position.

If you don't have a threaded gage plug of sufficient height, then probe it at the base and over whatever height you have and ask the computer what it would have measured over the correct height of the projected zone.

Also, functional gages can be designed to accurately give attribute (good versus bad) data regarding compliance for positional tolerance zones that are projected.

In my textbooks, I have illustrations and page after page of explanations on this topic.

James Meadows

Subject: Datum Shift on a Restrained Part

Datum Shift on a Restrained Part - Can you legally use "CHECK IN A RESTRAINED CONDITION" note with a Profile or Position callout with MMC (shift)? Can you have Shift on a Restrained Part? Sounds like an Oxymoron or contradiction in terms at first look. Is there a way to restrain a part and shift at the same time with fixtures (like clamp on a movable plate) or in CCM?


One thing has nothing to do with the other. Pattern shift or datum shift as some call it, is an apparent displacement of the pattern of features from the datum or datums. For example, it means that if a pattern of holes is positioned to a datum axis, as the datum feature of size departs from its MMC or virtual condition (as applicable), the pattern of holes may shift as a group an additional distance from the datum feature axis. What is actually happening is that the datum feature axis (axis of the actual hole) has shifted away from the datum axis (imaginary axis in space that can be simulated by the axis of a gage pin).

I assume you already know that, but think of the part as not being inspected in a gage. Think of it being restrained on a plate with clamps. A Coordinate Measurement Machine can then collect variables data about the datum feature's hole size and the location of the hole pattern from the axis of the datum feature. If the hole pattern seems to be shifted off in one direction more than the individual position tolerances allow, the additional pattern shift tolerance is then allowed to see if that is enough to shift them back (as a group) into their individual tolerance zones. This data analysis can be done after the part has been measured in a restrained state. All the restraint does is try to secure the part, perhaps to keep it from warping during the collection of the measurement data. Analysis of that data ignores whether the data has been collected while the part has been restrained or in the free state.



Subject: Drawing Interpretation Issue

Hi Jim.

Perceptions are everything.

In the illustration above, my friend, Brad, claims the figure on the right is correct, given the info in the drawing. Brad says, “Here is an example of where perception goes wrong. The top hole is B the bottom hole is C, it has |POS|dia 1.5m |A|Bm| It is very common for people to believe the C hole is "not to print" in the right view.”

My claim is that because they share the same 'X' plane in the left view, and the bottom hole is labeled -C-, the right view is incorrect.

Please advise. Thank you.



In this case, Brad is correct. It doesn't matter that the hole becomes a datum feature after it is positioned to A and B. When C is being positioned, it has only to be perpendicular to datum A and located from datum axis B for distance. Provided the distance is maintained to within its positional tolerance, it can actually rotate infinitely around the axis of B. Subsequently, I assume datum C will be used to orient the planes that cross at (and create) datum axis B. That's when you would orient the rest of the part to this entire datum structure (A,B and C). In the procedure you have begun, the next step would be to profile the outside of the part back to A,B and C. In that way, the rest of the part would be oriented to the common plane that runs between B and C, and the rotation you show could only happen within the profile tolerance zone.


Subject: GD&T Profile Question

Hi Jim

I took a course that you gave several years ago here in the Toledo area. It was sponsored by the local SME chapter if I recall correctly. I enjoyed the course but have become a bit rusty over some of the finer points.

The question I have involves a surface profile. Is it possible to remove elements from the profile while still retaining the influence of the control frame over the remaining elements? This is illustrated in the attached JPEG. We wish to gain tolerance for the radii of this profile - having them controlled by the drawing's standard tolerance and have the control frame only apply to the flat surfaces of the profile. If this is possible, is it necessary to do anything more than remove the basic block from the radii callout?

Thanks in advance for your analysis and reply.



Removing the basic dimension box from all radii wouldn't be clear enough to get you what you want. Why not keep what you have and either 1) Write a local note beneath the profile feature control frame that says "EXCLUDES ALL RADII" and then make the radii non-basic dimensions, or 2) Write a local note beneath the profile feature control frame that says "EXCLUDES ALL RADII" and then call out a separate profile control with a larger tolerance that says "APPLIES TO ALL RADII" and keep the radii as basic dimensions?

Hope this helps.

Subject: Question Regarding Datums

Hello Jim.

I have a statement regarding cylindrical datums that I'd like you to agree with or refute. There might be some honor involved.

I have a 12.0-12.6 cylindrical shaft that is datum feature A. I have a cup in one end, 8.4-8.8, and position tolerance of a diameter of 1 at MMC to datum A at MMC. My gage will represent datum feature A at a diameter of 12.6. On a similar part, I have the same cylinder and cup sizes, but I can live with a straightness of 0.2 applied to the diameter on datum feature A. Now my gage is a diameter of 12.8 (virtual condition).

I say that the position of the cup is measured to the axis of -A-, not to the axis of the actual part. A colleague says "No. You measure to the axis of the O.D. of the part."

I told him that you always take tolerance measurements from the geometric counterpart of the piece, the Datum Reference Frame. In this case (and every case) the x/y planes whose intersection makes up the axis of the gage to inspect the part is where you take the measurements from. He asked, "Who do you think you are, Jim Meadows?"

On another note, I've been reviewing your old newsletters from the late 90's. Some of those dudes who wrote to you are definitely OUT there! I wonder if they still are. But the tips are invaluable.

Please respond if you can. Thanks, Jim.



You are correct. The reason you are correct is that the datum feature is referenced at MMC. That means that the datum feature axis (axis of the real diameter) may shift away from the datum axis (imaginary axis) as the datum feature departs from its virtual condition or MMC (depending on which example you described is being discussed). The center of the gage represents the physical embodiment of the imaginary datum axis and is the origin of measurement. It is quite possible that datum feature A would not be in the center of the gage.

If the datum feature had been referenced at RFS, you and your colleague both would have been correct, since the gage would consist of a contracting gaging element and there would have been no difference between the axis of the datum feature and the datum axis. Both would be at the center of the gage.

I can understand why your colleague thought what he did. Without a gage to measure the part, one would commonly chuck-up on the actual datum feature and measure from its axis. But, with A referenced at MMC, he would then have to determine if the part checks bad using that technique. And if it does check bad, he would have to determine if the datum feature's departure from its MMC (or virtual condition as applicable) would allow the part to shift into its tolerance zone, making it a good part. A gage would automatically do this shift for the inspector.

Hope this helps.

Subject: Reporting Profile

Hi James,

I've had the honor of taking a couple of your courses offered through Chrysler corporation. I'm currently struggling with the method of reporting the deviation for profile of a surface with many of our suppliers. The method used at our facility is to double the maximum deviation form the basic. To keep it simple 1.00 basic with a profile of .01", part measures 1.004 / .999 reported profile equals .008.

Is there any "written in stone" method of reporting this value? I've been to numerous websites and GD&T forums that all express varying methods of reporting this value. I value your insight and knowledge, and would appreciate any guidance in resolving this issue once and for all.

Thanks in advance,


It's nice to hear from you. There is no "written in stone" method of reporting the value of the deviation from the basic profile. I get this question a lot, so maybe there should be one uniform way of reporting it. Since there is no standard specifically about measuring and reporting profile tolerance deviations, and since the Y14.5 standard is not a measurement standard, there is no final word guiding us to report this in any one way.

So, as with many situations, you will have to decide as a corporation the best way to report it and then go about trying to get all your people and suppliers to comply.

In May of 2007 an ASME Y14 subcommittee was approved to begin work on a standard that would “standardize” methods of reporting measured values such as profile. So, this problem is being addressed by ASME.

Good luck.

Subject: Question Regarding Perpendicularity, Projected Tolerance Zones and Total Runout

Hi Jim,

My name is Derek and I had a couple of GD & T and stackup courses from you MI a couple of years ago. I'm currently working for in Whitehall, MI and have run into a problem I'm hoping you can shed some light on.

Below I have a simplified version of an armature with a press-fit pin which rides in a tight-clearance housing. Functionally, I care about maintaining a consistent gap between the bottom face of A and the top of C. I need to know the best way to control the armature print (A) so that the perpendicularity of my press fit hole is held to the bottom face. My first instinct is to make the bottom surface datum A, and call out the hole perpendicular to it. However, if I gage it as close as possible to how it functions, I'd put a best fit gage pin equal to the length of B in the hole, place the pin in a V-block, and measure the total runout of the bottom surface of A as I spin the part.

Can I do this with a projected tolerance zone, or wouldn't this approach work?

Thanks for any insight you can provide.


Either method should achieve what you want, given the proper tolerances and notes.

I like the idea of inspecting total runout of the bottom face of part A (while the part spins). The problem that method faces is that you can't simply put a projected tolerance symbol after the datum feature you reference in the perpendicularity control on the bottom of part A. What you could do is, first position the hole on the part to other datums and give it a projected tolerance zone. Then assign it a datum feature symbol of its own. Then, when you reference it as a datum in the perpendicularity control on the bottom of A, you should probably write a short local or general note to explain what you explained to me about inspection (To be inspected by putting a best fit pin into datum feature B that is equal to the length of the depth of the hole plus the projected tolerance zone height of datum feature B, then placing the pin into a V-block, and measuring the total runout of the bottom surface of part A while rotating about the axis of the gage pin). This type of note isn't absolutely necessary, but since this is an unusual control, it might expedite the measurement procedure and assure they are inspecting the part in exactly the way you feel duplicates the way it functions. Another option is to show parts A and B already assembled and position pin B to other datums, then reference B in a total runout control used on the bottom of part A.

The other option of simply positioning the hole on the part to the bottom of A and other appropriate datums, then giving it a refinement of perpendicularity (to the bottom of A) with a projected tolerance zone is easy and also good. They wouldn't inspect the part while it spins, but the results should be virtually the same. This method is more conventional.

Hope this helps.

Subject: Query regarding Wall thickness calculations

Dear Sir,

We are part of Mahindra & Mahindra ltd. At present we are offering Engineering services in Commercial Vehicles to Global Customers. Keeping in view of importance of GD & T we have purchased a set of 12 DVD's on GD & T prepared by you 5 months back. While going through these DVD's & text book we came across following difficulties related to Minimum Wall thickness calculations (page number 198 to 200) related to Fig. 9-24.
Here the Allowed movement of feature D at LMC to C is .130 & again the same value is used for displacement (pattern shift). As Datum C is non feature of size, how this effect of allowed movement of feature D at LMC to C comes in calculation for minimum wall thickness?
We request you to clarify these points. Waiting for your early reply.



In both examples datums have been switched. When datums are changed, error is accumulated. In both these examples, the holes within the pattern are measured from D. D has a position tolerance of its own. So, the first .130 is because D moves out of position (to the datums it is related) the allowed position tolerance (and the holes measured from it follow D).

The second or .130 is because the pattern of holes positioned to D are referenced to D at MMC. This MMC symbol after D means D is represented at its virtual condition to the datum that precedes it in the hole pattern's position control. Therefore as D departs from its virtual condition (toward its LMC) the holes as a group can shift away from the center of D another.130.

You get it twice because of the change in datum reference frames and the fact that 1) as D moves the hole pattern moves with it and then 2) because of the MMC reference after D, the holes as a pattern can shift away from D as the D hole grows from its virtual condition to its LMC.

See the calculation of wall thickness to datum feature C below.

Hope this helps.

Subject: Question About Converting Inches to Millimeters

Hello James,

I have a question related to design tolerancing. I am doing the project of converting a product from Inch to Metric. When we convert inch tol. to Metric, can we convert directly into metric or do we have any other procedure to convert? For example the hole dim is 3.507/3.515 (inch) and the shaft dim is 3.497/3.500 (inch). Thanks in advance.



When we convert, we convert directly from one unit of measurement into the other. When we do this, we must be cautious in that there will be some error in the conversion. If we don't carry it out to enough decimal places, this error could be a source of problems.

I don't know why people convert at all. It seems like it would be more accurate (and less work) to simply work in the unit of measurement that the part was designed in.


Subject: Tolerancing questions

I was in your course in New Jersey on 22-24 Jan. I sat in the front. We talked about Feynman, etc.
I have some tolerancing questions.
A major vendor does something repeatedly on their drawings and I want to know for sure what it means and if they are doing it correctly.
In the sleeve. (I will fax it to you) they callout an inside diameter 1.815/1.813 with a total runout of .002 to A, the centerline. Does the 1.815/1.813 infer it is centered on the centerline within the .002 or do they need the .002 runout callout to state this?
They sent me a waiver saying they exceeded the total runout to .0028 but did not say they exceeded the 1.815/1.813. Is this possible the way this part is dimensioned? If so, how.
I was under the impression that the runout callout usually specifies a tighter tolerance than given by 1.815/1.813 but this company frequently calls out a diameter and either circular or total runout with the same tolerance as the diameter callout. Does this add anything? (They use circular runout for surfaces without much length.) Sometimes they use diameter callouts without a geometric tolerancing box. What is the difference?
The tee end screws onto a bar that sees considerable tension. I'm concerned that the 1.500/1.495 legs are perpendicular to B, the center line.
There are 2 position tolerances on the 1.500/1.495 diameter, .005 to B and .002 without a datum callout. The part exceeded the .005 to B but not the .002. What does this mean?
The drawings are proprietary, so I can't send the entire drawing.
Thanks for your help,


It's nice to hear from you so soon. I hope everything is sliding along in greased grooves in your world.

I'll answer question #2 first. The position tolerance of .005, as you describe it, has not been met for the two diameters in their positional tolerance (angle and location) to datum axis B. But it has met the positional tolerance relationship (.002) of the two diameters to each other. Apparently, the two diameters of 1.495-1.500 are coaxial enough (aligned) to each other, but not well enough positioned to datum axis B.

The answer to question #1 is that the size tolerance does not control the centering of the diameter. It only controls how big the diameter is, and in this case, because it is a tighter form tolerance than the runout tolerance, the size tolerance controls how well shaped the diameter is. The size tolerance does not control the coaxiality to the datum axis A at all. Therefore, any control of how well centered the diameter is to datum A is toleranced by the total runout control. Total runout normally control concentricity and cylindricity. In this case, with the size tolerance so tight, the total runout control only controls the concentricity (centering) of the diameter (but it is still inspected as though it is total runout). This is covered in the Y14.5 standard on Dimensioning and Tolerancing. In section 2.7, it says the size tolerance controls the form of a feature of size. For a cylinder, it controls the roundness, straightness and taper. But it goes on to say in 2.7.3 that the limits of size do not control the orientation or location relationship between individual features. It says that features shown perpendicular, coaxial or symmetrical to each other must be controlled with a tolerance to avoid an incomplete drawing requirement. Specifically, to your situation, that means the 1.813-1.815 diameter needs a concentricity, circular runout, position or total runout tolerance (as it has) to show how far off center of datum axis A the diameter may be.

Hope this helps.


Subject: Measuring Parts While Applying Restraints

Hi Jim!

I’m from an Army facility.

I was asked about measuring flatness on a part that can be restrained
for inspection.

One of our inspectors is asking where in the ANSI Y 14.5 does it say we
can inspect for flatness while we have a part restrained. I could not
find any specific quote, or paragraph in any literature. Apparently our
inspector does not have a problem measuring the dimensional features,
but has a problem measuring the flatness feature in a restrained

What's your opinion; I don't see anything wrong with restraining the
part and measuring the flatness.



The most pertinent standard on measurement is ASME Y14.43-2003. It does, however, reference Y14.5 in the passages that deal with restraint. These two standards were coordinated to make the rule as clear as possible in Y14.43. The key is that unless a note is written specifying that restraint will be used during measurement, all parts are measured in the free state. This is due to the fact that if enough restraint is used, many parts that would otherwise fail inspection can be distorted into passing. Distortion of parts to obtain compliant measurement results is forbidden. If the restraint used does not distort the part in a manner that significantly changes the measurement results, then restraint may be used. So, restraining a very rigid part may be okay, while restraining a flexible part would not be okay.

See the attached page from Y14.43 for the rules in 2.8.1, 2.8.2 and 2.8.3.

Hope this helps.


Subject: A question Regarding Hexagons.

Hi Jim,

I hope all is well with you. I am a former student of yours and I’m looking forward to your Comprehensive Advanced GD&T course in Nashville in December. I have all of your books and I appreciate your immense knowledge on all things GD&T. I named my first child Jim Meadows. I think it's a nice name, but my wife is afraid she'll get picked on by mean kids. Anyway...
A colleague and I are having a difference of opinion and I'm hoping you could help us out. What we have is a part with an internal hexagon bore (think socket head) cut into it. It is dimensioned as ".314 +.010/-.010 Hex" across the flats and it is given a depth and positioned on a centerline. Is the dimensioning of this feature complete? Do not consider design intent and assume there is no general note regarding corner breaks.
One opinion is that the feature is fully dimensioned, and that an acceptable part feature would be an internal hex - no part of which violated the boundaries of two perfect hexagons at .304 and .324. For instance a hex bore that was .324 across the flats would no be allowed any straightness error, but rounds on the points would be permissible so long as they never violated the perfect hex at .304 boundary. The thought behind this is that size limits control form and they do so no differently on a hex bore than they would on a cylindrical bore.
The other opinion is that insufficient details were provided by the print - that a hexagonal bore is not one single feature and that both the distance across flats and the allowable radius or break in the corners must always be dimensioned. A print that only shows a dimension across flats and a depth of hex must be rejected as incomplete.
What do you think?


In ASME Y14.5M-1994, in section 2.7.3 there is a rule that applies here. It comes under the heading Relationship Between Individual Features and goes something like, The limits of size do not control the orientation or location relationship between individual features. Feature shown perpendicular, coaxial or symmetrical to each other must be controlled with a tolerance to avoid an incomplete drawing requirement.

You can look it up to get the exact quote, but that's it within a couple of words. The essence of the statement is that a feature of size is cylindrical, spherical, two parallel line elements or two parallel planar surfaces. And that your feature of size which is several sets of "two parallel planar surfaces" are each controlled for form (flatness, straightness and parallelism), but they aren't related to each other, unless you say to within how much in a control other than just a size limit. If there is a general tolerance note on the drawing that says, Unless Otherwise Specified, All Angles Are +1 degree (or some number of degrees), that would relate the sets of parallel surfaces to one another to within that tolerance. But even then, we would have to make certain that was an acceptable amount of tolerance and then what about the location of the hex on the rest of the part. Is it supposed to be centered to some diameter, like the hex head on a screw is supposed to be centered to the diameter of the screw? Or, are there other diameters or widths or edges it is to be located to?

Usually, if such a feature is important enough to ask a question about, as you have, it is important enough to completely dimension and tolerance. I don't think a size tolerance is a complete definition of the hex. I think you could position each set of parallel surfaces to each other and to a datum reference frame by simply applying a position control to the 3X .314+.010 dimension.

If you wanted to get really crazy, you could make the .314 dimensions basic then profile the hex all-around to a datum reference frame (a complete set of datums).

That's my opinion. I hope it helps. Tell your daughter that I'm sorry.


Close Window