Close Window


Written, with help from students and clients,  by James D. Meadows
The Tolerancing Newsletter April, 2005

As many of you know, for over twenty years now, I have used this newsletter as a way of allowing former students, future students and professionals with different levels of Geometric Dimensioning and Tolerancing knowledge to write in and ask technical questions.  I believe that those in and interacting with technical fields such as design, manufacturing, quality and inspection have the ability to learn to apply and read the language of GD&T.  Well, most of them anyway.  There was the time I was trying to teach one inspector what MMC is.  I said, “Look at the size specification on this shaft.  It is given as a limit dimension.  There is a big number and a small number.  The MMC of the shaft is the big number.”  He looked at me, then back at the page and said,  “The numbers look the same size to me, Jim.”  I thought about that for a moment, then said, “No, not the type style, the value.” 


Question on position:




I hope all is well with you---I am again trying to get you in here or at least get the engineers to you for your course.  I will persist.


I have a round part with thru holes placed every 60 degrees. The drawing gives a start angle and then a note specifying the size, with the position callout and the statement: "6 holes equally spaced". The start angle and B.C. are basic. They want the start angle to be a +/- tolerance and keep the "equally spaced" note as basic. I am having trouble with this.  Is this in keeping with the standard??  I have looked but can't seem to find any examples.


I would prefer to take the start angle off the drawing and put 6X 60 degrees basic with the note for the holes pointing to the horizontal axis of the part. The orientation is not important. 








You know what you're doing.  What they want to do is illegal and beyond that, it makes no sense.  The positional tolerance on these holes is what tolerances the basic angles of 60 degrees from hole to hole and the basic distance from hole to hole (and any angles and distances from the datum or datums).  If they give the angle a plus and minus tolerance and then give the holes a positional tolerance, you would have competing tolerances trying to do the same job.


So, as you say, keep the angles and the distances basic from hole to hole and from the datums, then rely on the position tolerance to tolerance all these basics.  If the relationship to the datums is less important than the relationship from hole to hole, use a composite positional tolerance.


I have shown a similar part example from one of my books below.





Diameters as Datum Features:


Dear Mr. Meadows,

My name is Joe and I work at a company in Iowa.  You gave our company some in-house training on geometric dimensioning and tolerancing - level 1 in December of 2001.

I just finished a heated discussion with two of my co-workers regarding diameters as datums.  The issue is when using a diameter as a datum feature, does the datum feature symbol have to be lined up with the dimension line?  One of my co-workers says yes, the other vehemently says no, it does not have to be lined up, the meaning is the same either way.  I'm confused.

Please see the attached sketch.  Do the two datum feature symbols mean the same thing?

On page 128 through 130 of your book Geometric Dimensioning and Tolerancing Applications and Techniques for use in Design, Manufacturing, and Inspection, you give explanations and examples of how the symbol is to be applied.

The two arguments go like this:

Argument #1 - since a diameter is a feature of size (and can only be a feature of size) whether the symbol is lined up with the dimension line or not is immaterial.  The datum feature is the diameter (surface) and the datum is the axis created by that feature.  Also, the first bullet point on page 129 states that, " the datum feature symbol extension line of the feature outline, separated from the size dimension.  This designates that the datum is the axis of the cylindrical feature." (emphasis added by me to “separated by”, not in the text)

Argument #2 - if the symbol is not lined up with the dimension line, the datum is the surface of the diameter,  not the axis.  Similar to the description given on pages 123 through 125 on centerplane datums for non-features of size.  Therefore to denote the axis as the diameter, the symbol must be lined up with the dimension line.

Can you shed some light on this?






There is no difference whatsoever.  The surface of a diameter can't become a datum.  Datums are axes or planes (regular planes or centerplanes).  Argument #1 wins.  Argument #2 has no merit.  Diameters always generate an axis.


When making a diameter a datum feature to generate a datum axis, we do try to line up the datum feature symbol with the dimension line as a matter of drafting preference, but to not do so doesn't change the meaning.


The argument made referencing page 123 and 125 does hold true for widths and centerplane datums, but not for diameters and the datum axes they generate.


Hope this helps.




Position with a Square Symbol?  Measuring Position on Square Features?



James - attached is a print that we have marked up with questions, but there
is an interesting call-out that I wanted to get your feedback on.

The part has a true position control on a square hole that is a portion of a star shaped hole and it has a square symbol after the position symbol and before the positional tolerance of 0.08.


1.  Is that correct as to how they drew it?

2.  How do you measure it? For example, if you had a CMM what would you do?

Just so you know - I took your gage design course at the UWM Continuing
Education which has been a very valuable tool as I do design lots of attribute gages for our product.  I look forward to your feedback.





No, there is no provision in the Y14.5 standard for using a square symbol inside of the feature control frame.  However, you can create a square positional tolerance zone for a feature by giving it two feature control frames.  Each position tolerance would point in a different direction (one in the X direction and one in the Y direction) and both position controls would reference the same datums.  Still, it is pretty easy to figure out what they wanted to do, and in tolerancing, clarity is No. 1.  In other words, even if they circumvent the rules, why not give them what they want? They can always go back and correct the drawing to abide by the Y14.5 rules ASAP.   As to how to measure it, technically with a position control, one should not probe the hole, but instead probe a substitute for the hole, such as a gage pin.  But since that is so difficult to do, most CMM operators would simply probe the hole and treat the location tolerance zone as though is was plus or minus 0.04 in both the X and Y directions.


I have shown a similar example from one of my books below (not a square zone, but the concept is close).  It creates a rectangular tolerance zone for a round hole.  Notice the perpendicularity control prevents the axis of the hole from leaning across the hypotenuse of the rectangular positional tolerance zone.







Thanks greatly for the quick response.  One other question.  I have a drawing that uses perpendicularity on a surface to a datum axis constructed by two diameters (A-B). Does the call-out for perpendicularity relating this surface (which later becomes datum feature C) to this compound datum axis make sense?  Let me clarify what I am asking.  If I needed to measure this perpendicularity control, I would try to hold datum A by some means and then basically check face runout. The problem I have is they say to measure from A-B.  I have an engineer who says that we need to hold both A and B simultaneously, and I do not see how that would make sense.


A related issue is if you look at the right top detail the face has an angle to it making it very difficult to check.


Also, the opposite face from C has a callout to A-B, but would it make more sense to have it be parallel to C instead?







That's a lot of questions.  One at a time.  Measuring perpendicularity to a compound datum axis (A-B) is thought of as a little different than checking total runout.  But the result is that the same geometries are accepted.  Total runout is implied to be measured while chucked up on the datum diameters and rotating about the datum axis they generate.  Perpendicularity is implied to be measured in a stationary state while centered on the datum axis generated by both diameters.  Perpendicularity can be measured on a coordinate measurement machine (CMM).  Probing both diameters and asking the computer to generate a single axis (hopefully from the highest points probed), then probing the surface or surfaces being controlled by the perpendicularity characteristic is actually quite doable.  Whether or not it makes sense should be considered from a functional aspect.  If both diameters are used in the assembly to stabilize the part, then it would seem to make sense.


The question about the surface with an angle callout that is difficult to check seems to be a different issue.  But as long as that angle is needed to complete the part definition, it only makes sense to dimension and tolerance it. Whether it is easy or difficult to measure should not be the deciding factor.  Our ability to measure things is only limited by our creativity and our budget, but parts must be completely defined.


As to the last question (Should one face have a parallelism control to the other instead of a perpendicularity control to the same datums?), one is almost as good as the other.  If you go with the parallelism control on the second face, you accumulate error to the original compound datum axis (A-B), but have a more direct control to the first face.  If you go with the perpendicularity control, there is no accumulation of error to the original compound datum axis (A-B), but there is an accumulation of error to the first face in that the two faces would be related to one another to within the sum of their perpendicularity tolerances to the compound datum axis.


Good luck.




Composite Positional Tolerancing

 Good Morning Jim---

   A question about composite tolerancing:

        I have 3 coaxial holes 8.3-8.5, positioned within a diameter of 3mm to the main datums B & C).  Then I have them set up as a set (composite) with no datum reference within a diameter of 0.3 (to hold the alignment of the 3 holes’ axes to each other).

When I run my numbers to figure the inner & outer boundaries, do I take the size and add the 0.3 only or do I add the 3 as well.

I know that the 0.3 must remain within the 3.





The holes will have two inner and two outer boundaries.  One set of inner and outer boundaries will be based on the collective effects of the size limits and the 3mm of position tolerance.  Completely separate from that the other set of inner and outer boundaries will be based on the collective effects of the size limits and the 0.3mm of position tolerance.


You didn't tell me why you are calculating the inner and outer boundaries, but whatever the reason (tolerance stack-up analysis, determining worst mating conditions, calculating minimum wall thickness, etc.), you must determine which set of boundaries is applicable and then use the appropriate geometric tolerance to find the unknown you are seeking.  The first rule of tolerance stack-up analysis is to know where you are going to be able to choose the correct path to get you there. 


I have shown a similar situation from one of my books below. 






Rule #1 and Inspection


Hi James:

Given; inspect a part with only size dimensions, no geometric tolerance.  So, checking to Rule #1, you make a two point inspection to check that the part is within MMC and LMC.  Finally, you check the envelope at MMC.  That check is my problem.  The geometric characteristics are indirectly checked by Rule #1, but how do you check, for example, the outer envelope of a round shaft without directly checking directly for roundness.  Another example, is cylindricity.  Other indirect geometric controls, or all, may fall into my problem.  Help!


I have the impression that inspecting to Rule #1 should be less costly than inspecting to geometric controls.  In general, how do you do this without directly checking geometrics?  Inspecting the envelope at MMC for Rule #1: if high volume, with hard tooling? Costly?  Or, if at low volume, with open setup?   But then, how would you check the envelope at MMC of a "straight" part without directly checking the geometric control for straightness? 








The purpose of Rule #1 is to have the size tolerance control the form tolerance.  The way size is inspected is to ensure that it controls form.  In fact, size tolerance controlling form tolerance is most often more restrictive than using a geometric control on a part. With a geometric control such as cylindricity, one can allow a shaft or hole to grow or shrink a lot with a large size tolerance, but be better in its 3-dimensional form by adding a tighter cylindricity tolerance.  Just calling out a tight size tolerance would control both size and form to within the tight tolerance.


Measuring a part would require cross-sectional checks for LMC (like a micrometer check for shafts). Then, an envelope check for MMC would require (per Rule #1) a GO gage to either be used or simulated.  So, the answer to your question is that geometric controls don't make the part more expensive.  Their goal is to make the part less expensive, but to still function well.  And, you can't check the envelope of perfect form at MMC (created by Rule #1) without also checking (on a cylinder) the cylindricity which controls roundness, straightness and taper.


You could always eliminate Rule #1 by a variety of methods including just writing a note that states, "Perfect form at MMC is not required."  But then, unless you add a geometric control, you will have no form control at all and the part could look like a banana or even worse, it could curl completely around on itself. 


You can do pretty much anything you want to do, but remember that Rule #1 is in place to allow you to control the size and form to within the same tolerance.  Geometric controls such as cylindricity are available if you would like to have a smaller form tolerance than size tolerance.  Straightness of the derived median line (axis) is available to apply to a shaft or hole to eliminate Rule #1 and allow the feature to be out of straight even at MMC.  And as I mentioned, notes such as "PERFECT FORM AT MMC IS NOT REQUIRED." are available to say that size and form are entirely independent of one another.  So, just decide what is best for the product, and then set about creating that situation by what you put on the design drawing and/or CAD model.




Mistaken Identity



I saw a guy with a bumper sticker that said, “Ask me about GD&T.”  Thinking he might be an expert or at least another interested practitioner of (G)eometric (D)imensioning and (T)olerancing, I followed him.  Every time I figured he was about to stop, he would take another dusty side road or go down a dark, garbage strewn alleyway.  Before I knew it, it was midnight and I was 50 miles outside of civilization.  He finally pulled into the parking lot of a really sleazy-looking joint called “(G)irls, (D)rinks and (T)rouble.”


Come to think of it, I was right after all---we did have a lot in common.  Still, finding this out took us on a 3-day escapade, most of which I don’t recall (although I’m fairly certain some firearms and sirens were involved).  Anyway, I am left with the faint recollections and physical evidence of being (G)ross, (D)isorderly and (T)attooed.  Know any way to have the words “U.S. Amy” removed from one’s chest?  The tattoo artist apparently never won a spelling bee. 


My problem is that I’ve never known anyone named Amy and, besides, even if he had gotten that word right, what I seemed unable to remember at the time was that I was in the Navy---not the Army.



Tattooed and Blue


Dear Tattoo,

(G)ive (D)iscipline and (T)herapy a try.  Then, (a) date only women named Amy or (b) Consider a battleship as a cover-up tattoo.






Tolerance Stack-Up Analysis & Accumulated Tolerance Error When Multiple Datum Reference Frames are Used


Hello Jim

I recently purchased your book 'Tolerance Stack-UP Analysis" and have a question related to exercise 8-5, problem #1 "answer page.

In step 1 you indicate the following:


      .250 =LMC hole   

    +.050 = Geo. tol @ LMC


      .310 dia = outer boundary of hole

    +.130 = pos. tol. of datum D




Then you add .130 for pattern shift of datum D. This is where I become confused. Can you please explain to me why you add the .130 as pattern shift after adding it as positional tolerance of D at LMC.  Wouldn't the .130 shown above account for the pattern shift?

I greatly appreciate your time in explaining this.


Thank you




See the following illustration:








What you have to ask yourself is, “Would there be a difference if the 4-hole pattern’s position control did not use the MMC symbol after D?”.  When you switch datum reference frames in mid-part definition, you accumulate additional tolerances.  This analysis recognizes that the datum feature generating the axis from which other things may be measured may first experience a movement of its own (from the first datum reference frame).  Then, wherever that axis lands, it then becomes a new datum axis (D) from which the pattern of holes is measured and from which that pattern may shift away due to the MMC symbol next to the datum feature D in the hole pattern's position control.


Think of it as two gages.  One gage measures the location of the first hole (and gets the .130), then the second gage measures the location of the hole pattern (and it gets the .130 also).  Each gage must represent the hole from which the .130 comes with a gage pin that is .130 smaller than the hole may be produced (LMC of .600 minus Virtual Condition of .470=.130).  For a more complete answer with pictures, see the last newsletter for July/August 2004.


I hope this helps,




Datums for Symmetrical Parts


Dear Mr. James D. Meadows,

I have recently purchased your book GEOMETRIC DIMENSIONING AND TOLERANCING
Applications and Techniques for Use in Design, Manufacturing, and Inspection
(per ASME Y14.5M-1994).  Thank you very much for such a good book.

Could you please answer the question (please see the attachment):

When you establish datums on symmetrical part (for instance such as Fig. 15-44 and 15-58 in your book (please see the attachment also), is it clear for a part inspector which datum surfaces should be simulated?  Since the part is symmetrical, the inspector may choose any of the four surfaces as a secondary (B) or tertiary (C) datum feature.  Also, the inspector may choose a surface opposite to A shown on the drawing as a primary datum A.  It looks ambiguous to me.  How does the inspector determine the correct datums when all of them look identical? Is my concern valid?

Thank you very much for your time and consideration.  I look forward to hearing from you.

Best regards,





Your concern is a valid one.  The Y14.5 standard addresses this by recommending that when this is a concern, the datum feature may be made distinguishable by physical identification.  One way to do this is to design in a feature such as a notch to make it possible to tell one surface or width from another.  Other ways are for the manufacturers and inspectors to physically identify the surfaces by stamping or etching letters into the surface.  When I was a die maker (many, many years ago) I used to carry a crayon or piece of chalk with which to mark surfaces, so that I would be consistent in setting up parts to machine or measure.


I hope this helps,




MMC on Datum Features


Hi Jim, attached is a linear guideway drawing that is dimensioned to
communicate to the supplier which holes we will be using to attach their product.  We assemble this part to a machined extrusion with tapped holes (with the same dimensioning scheme).  The clearance holes are referenced to a datum feature hole at MMC).  My question is regarding the MMC modifier on the Datum A hole.  A question came up in our Quality department regarding how to apply the bonus tolerance as the Datum A hole departs from MMC.

Do we check the datum A feature size and the holes to be inspected and add the amount both features depart to get our total position tolerance zone?  OR Does it only allow the hole pattern as a group to depart from the datum feature axis?  I believe that if the holes were dimensioned to each other as opposed to individually dimensioned from Datum A, then the bonus tolerance gained from Datum A hole departure from MMC would only apply to the hole pattern?

(Note:  I think I need to change Datum A to be the mounting surface opposite the counterbore side, Datum B as is - centerline of the rail width and Datum C to be the hole now noted as Datum A.  We will be making gages for these parts soon.)

Thanks for your help.






The short answer to your question is that the additional tolerance gained from datum feature A (by virtue of referencing the pattern of holes to A at MMC) only allows the hole pattern as a group to depart from the datum feature axis.


The way the basic dimensions are depicted on the drawing, whether they are from A or from one hole to another, has no bearing on this drawing interpretation.  It is the same either way. 


The only way the drawing could be interpreted differently is if each hole was a datum feature and the next hole was positioned only from the previous hole (in other words, referenced to the previous hole as its datum).  This would mean each hole would reference a different hole as a datum feature and the accumulated error would be huge.  But if all holes are referenced from the same datums, located with basic dimensions, there is no accumulation of error.  That is the big difference between basic dimensions and plus and minus dimensions for location.  When plus and minus tolerance was used (hopefully in the distant past) for the location of holes dimensioned from each other, it made for the possibility of horrific accumulated error.  Basic dimensions can be used as vehicles to add or subtract from one another without accumulated error (as long as the datums aren’t switched).


I like your idea of making the mounting surface opposite the counterbore a primary datum feature, then B as secondary and the hole as tertiary.


But let me offer another suggestion.  What if you did not make the hole a datum feature at all.  You could position all 17 holes as a pattern to the mounting surface as primary, the width (now called B) as secondary and the left edge as tertiary (making the .295 from that edge a basic dimension) within a diameter of .100 at MMC.  Then you could refine that with another segment of position (below) that referenced all 17 holes (to each other-which is always toleranced by the tightest position tolerance anyway) to within a diameter of .0065 at MMC and reference the mounting surface as primary and the width B as secondary.  This would allow all 17 holes to move left or right (in the view you sent me) plus or minus .050 (a diameter of .100), but keep their position tolerance to each other and the mounting surface and the width B all to within a diameter of .0065 at MMC.  Both levels of position would use their own position symbol.  This is called a Two-Single-Segment Position control. Just a suggestion. 


Also, you should think about changing the entire drawing to either inches or millimeters.  Having both on the same drawing is kind of confusing.


Hope this helps,




Calculating a boundary for RFS material condition modifier


Good afternoon Mr. Meadows,

Thank you for your time.  I have been working at this company for 6 ˝ years and have come to a basic understanding of GDT.  One of my goals at work is to better my skills in this field.  Recently the company library obtained a copy of your workbook "Tolerance Stack-Up Analysis".  Over the past two weeks I have been reading and trying some of the exercises in the book.  However I have come to a concept that I do not understand and was hoping you would be able to help me.  I am having difficulty understanding how an RFS call out works in a tolerance stack up.  On pages 1-10 and 1-16 there are two diagrams showing the worst case outer boundary and the worst case inner boundary, respectively, in an RFS call out.  For example in the diagram on page 1-10, I perceive the center points of the 51 mm diameter LMC dimension of the hole to be outside the 1 mm diameter positional tolerance zone.  Does not the median line composed of all the actual local size centers need to lie within the 1 mm diameter positional tolerance zone?  I would appreciate it very much if you could help me understand how the worst case is 49+2+2+1=54 for axially out-of-straight holes.

Thank you very much,



See the illustrations below.






Your question is a good one.  Most people misunderstand the concept of position tolerancing.  It falls under a principle that states that the axis of the maximum inscribed cylinder for a hole and the axis of the minimum circumscribed cylinder for a shaft must reside within the positional tolerance zone.  So, the answer to your question is, “No, the median line composed of all the actual local size centers does not need to lie within the 1 mm diameter positional tolerance zone.”  This concept not only applies to position tolerancing but also to perpendicularity, parallelism and angularity when used on holes, shafts, slot widths and tab widths.


You will find that in the rest of the textbook I have not used these additional (size tolerance times two) tolerances to calculate the boundaries for holes and shafts, but instead, in RFS conditions, I have just used the size tolerance one time.  As I explain in the book, this is because that although pages 1-10 and 1-16 are correct, I consider them a special case (just for banana shaped holes and shafts) and don't feel it represents the most common occurrences for parts.  Still, you have to be the judge of when this may or may not be a consideration.  I just try to show everything that could go wrong. 


I gave a brief presentation on this oddity recently at a ASME Y14.5 meeting (the ANSI committee that writes the Dimensioning and Tolerancing standard) and many of those present were shocked.  One member even came up afterward and tore off the flip chart pad sheets where I had drawn illustrations similar to what I have on pages 1-10 and 1-16 in your book.  He said he wanted to take them with him because it was such an eye opening experience.  No one disagreed with the illustrations. 


Just remember, if you work on very thin parts, such as holes in sheet metal, you probably don't have to worry about this phenomena (banana-shaped holes).


Hope this helps,





Good morning Jim,


I wanted to thank you for your help.  Your insight on the subject matter is truly appreciated.  As of this morning I have completed all the exercises in Chapter 1 of the workbook.  It is encouraging in that I am understanding the concepts so far. 


Thank you again,






I hope you will all feel free to write in and share any insights you have and ask whatever technical questions have been plaguing you and your colleagues.  In doing so, you make it possible for all readers to learn about real problems you face daily in the workplace.  And hopefully, together we can find some answers to many of them.


I know that most of you will use good judgment in how you utilize this forum.  Not like the gentleman who faxed me 63 pages of questions in one day or the one who called from a foreign country asking for bail money.  He said, “I attended one of your public seminars and you said if I ever needed any help, I should call you.  Believe me, if I ever needed help, I need it now!”  As much as I’d like to help, my commitment to do so only goes so far.


Jim Meadows

Close Window