Close Window


Written, with help from students and clients,  by James D. Meadows
The Tolerancing Newsletter January, 2006

Subject: How Do We Monitor Quality When We Manufacture in China?

We are starting to do all of our manufacturing in China. To be honest, I don’t know how to monitor quality from long distance. There is talk that I will have to move to China, but since the cuisine tends to run toward fish heads and eel entrails (does an eel even have entrails?), I am looking for other solutions. How do we ensure that products produced in China (or India or Ethiopia or whatever country where workers will work for practically nothing while living in squalor) meet our functional requirements?

Doug (I’m not eating that!) from Detroit

Dear Not Eating,

Collecting variables data is a great way to monitor manufacturing capability in our culture and societies like ours. But I’m pretty sure that some of the low bidders from foreign lands are guys making things in an abandoned building once used as a prison for political prisoners (who are now buried out back somewhere).

I believe there is no more effective way to know that products will function than to send the manufacturers functional gages and fixtures that the parts must pass before they are sent here for assembly (assuming assembly is still being done here). Design, dimension and tolerance these gages and fixtures per the new standard on Dimensioning and Tolerancing Principles for Gages and Fixtures (ASME Y14.43-2003). Make certain that they reflect exactly how the parts function. If done correctly, anything that passes those measurement tools will, by definition, function.

I believe many of the tried and proven ways to assure quality practiced here in the United States will fall by the side of the road when we ship these jobs over to these stone-age societies, run by fanatical dictators. But judging from all we see, it appears to be inevitable. So, I think gages and fixtures designed to reflect assembly conditions and functional requirements will be among the most cost effective ways to make sure we don’t end up having to buy low cost, low quality products from our cuisine challenged new work forces.

Close your eyes, open your mouth and swallow real fast. We wouldn’t want to offend anyone. It was only a few years ago that I read a news story about a Chinese company making a refrigerator that wasn’t up to specs. The result was that the manufacturing manager and the quality manager were immediately taken outside and shot in the head.

Happy traveling,

P.S. I was at a company just last week that was in the same circumstances you now find yourself in. What I did was go over all of their drawings and correct them throughout the class. After that, I immediately designed, dimensioned and toleranced gages and fixtures for two of their products. These gages and fixtures, once completed here, were to be sent to China as the last word on what would and would not be accepted.

As the chairman of ASME Y14.43 standard on Dimensioning and Tolerancing Principles for Gages and Fixtures, which is the only national or international standard on this topic, I offer this service to any of our current or prospective client companies. I really believe that this is something that EVERY company manufacturing products overseas needs to do to protect themselves from the miscommunications inherent in doing business with cultures so different from our own. For details on this service or the topics contained in the course material, see the sections of our website under On-Site Training and Public Workshops.

Subject: Happy New Year!


Well, here it is 2006 and the wheels have really come off of American industry. The stockholders are getting richer, while the companies rid themselves of their American workers and close one manufacturing plant after another. GM lays off 30,000. Ford lays off 30,000. Delphi wants a 60% reduction in worker wages, which puts them at approximately the same level as a WalMart greeter. Etcetera, etcetera, etcetera.

I was at a company in Georgia last week and asked them where their product was manufactured. They said, “Right out there in the plant. Yes, sir. Not 30 feet from where we are standing right now. Made in the USA by American workers, by God!” I said, “That’s great.” Then I was silent for a little while and finally asked, “When is manufacturing moving to China?” They said, “In about 6 months.”

It doesn’t seem to be a matter of if, any more, just a matter of when. They said they had been to China and seen the production floor. It was made of dirt and the workers used wheelbarrows to lug the parts from place to place. They said it didn’t matter that they weren’t knowledgeable or capable, it only mattered what they charged. And they charged almost nothing.

If you go to the store and try to find goods made in America, you will be looking a very long time. So, say goodbye to the middle class. There will be the rich and the poor and a massive divide between the two. Crime will escalate and the inner cities will crumble. You won’t be safe in your own home. And it won’t be your home for much longer. We’ll all be salivating over owning our very own packing crate to live in.

So, Merry Christmas and Happy New Year, and put on this paper hat. No, not to celebrate, but to costume you in your new work outfit, so you won’t look out of place when you say, “Want fries with that?” And then you’ll become part of the touted new ‘200,000 strong workforce’ created in the last quarter of 2005.

Since this is a newsletter and you try to answer technical questions that people write in, here’s one: What job is safe from exportation?

Sinking fast in Philadelphia

Dear Sinking,
I’m just guessing here, but I would say that the answer to your question is Politician.

Subject: Positional Tolerancing Question


I have a question on positional tolerancing for you. I've attached a document showing two versions of a few views on a print. The left view is the way the current print is dimensioned/toleranced, the right view is the way I originally dimensioned it (before our customers "GD&T expert" changed it) and the way I believe it should be.

He says my dimensioning scheme is invalid for a bunch of reasons. My questions for you are- Is either dimensioning scheme incorrect? If so, why?

Thanks for the help!


Current Print Not Correct With Basic Dimensions to Edges

Proposed Revision

Proposed Revision is Much Better


As far as the positional tolerancing is concerned, the proposed revision looks correct for the reasons you have listed on the drawing. The current print is not correct for the same reasons. Position tolerancing is considered as tolerancing the location of the center of a feature of size, therefore having basic dimensions to the centerplane or axis of a feature from the datums is the way to go.

Profile tolerancing, on the other hand, is a surface control. It tolerances the shape of the surface and is capable of controlling the size (all around or all over) and the angles and location of the surface. Therefore, for Profile of a Surface tolerancing, the basic dimensions are used to define the surface. This can be done by locating the center of a radius with basic dimensions from the datum or datums, and then using basic radii on the profile from these centers. If the feature lends itself to locating the surface from the datums, then that would be okay as well. From the view you sent me, it looks like neither has been done on these drawings. But, I have to assume, it is done in other views that I don't have access to.

Hope this helps,


Thanks for the help on this. I wanted your input before I challenge their "experts’" changes.

I'm not sure what you mean when you say "From the view you sent me, it looks like neither has been done on these drawings. But, I have to assume, it is done in other views that I don't have access to". I understand that it could be done with profile tolerancing, but I prefer the positional tolerancing with +/- size dimensions in this case. So, assuming we keep the positional tolerances- is everything correct?

Also- your comments bring up another question. We do have a profile tolerance on the two window openings, but we do not have the basic dimensions associated with those openings on the drawing. Is it acceptable to leave the basic dimensions off of the drawing and make the inspector utilize the CAD data for the true size/location of the features or do we need to add them? One more thing with those particular profile callouts- you will notice there is a flag directing you to note 9 next to the feature control frame. Note 9 directs you to use a gage to check the feature. If a gage is used, does this allow you to eliminate the basic dimensions?

Thanks again for your input.



My reference to the basic dimensions being missing was about the Profile controls. The position controls are fine.

You may reference the CAD model for basic dimensions that would complete the drawing. I did not have access to the notes that I assume you are saying do that. It would seem to me a simple thing for a few more dimensions to be added to the field of this drawing for these profile controls to be immediately clear. But it certainly is common for the CAD model to be referenced for this. I have to also mention the missing dimensions and the tolerances to define the periphery of this part. Again, I will assume they are elsewhere, perhaps defined at a previous drawing level.

It would be unusual to rely solely on an existing gage to replace a complete part definition. I would recommend completing the part with dimensions and tolerances that would make the gage the appropriate method of measuring it. One of the basic rules of part tolerancing in the Y14.5 standard is that all parts shall be completely dimensioned and toleranced.



The periphery of the part is dimensioned and toleranced in other views. I will probably recommend adding basic dimensions to the window openings.

I should be good to go on this now. Thanks!


Subject: Cylindricity versus Straightness and Concentricity

Hello James,

You probably do not remember me, but I took your GD&T course when I was with Siemens. I started working at a new place 6 months ago, and there are some questions that have arisen about GD&T.

I have attached a document file to show the stator core tolerance drawing that my company uses to measure some attributes on the stator.

Currently the drawing calls out concentricity of a max of .006 for 5800 frame and the OD size of lamination has a tolerance of .005 maximum. Yet, they have a very tight tolerance on the straightness of the stator that is less than .004 maximum.

I think the stator straightness is overkill since we are already allowing concentricity of 0.006 not to mention the tolerance from the OD of the lamination itself.

The current straightness is not process capable, so I am thinking about combining straightness and concentricity to a cylindricality call out with 0.006 max tolerance. Does that make more sense than having a concentricity and straightness call out separately?

Please let me know what you think about this.


Cylindricity is not the equivalent of combining straightness of a surface and concentricity. Since cylindricity can't reference a datum or datums, it can't center the diameter to the datum axis. Total runout controls cylindricity (roundness, straightness and taper) and concentricity. So, if you have a cylindrical surface that needs a good three-dimensional form control and also a coaxiality control, total runout would be the correct control to use. Total runout is defined as a geometric characteristic that controls 3-D form and centers the surface to the datum axis.

As far as the tolerances they are using being too tight for your manufacturing capability, that is something you probably know everything about and I can't comment on, since I don't know what manufacturing techniques you are employing.

I hope this helps.

Subject: Question Regarding Profiles

Hello Mr. Meadows,

I once took your class at the Chrysler Tech Center in Michigan and we are having a discussion on how a feature is toleranced versus the way we know to measure it and I was hoping you could clarify something for me.

We are in the process of measuring profiles on an HVAC unit and an example of the call out is. From what we know and reading through your book on MMC and profiles we should not add MMC to the profile as it does not have any benefit.

Our engineer here claims that another expert (name withheld) states in his book to treat it as a True Position and it does give you more tolerance. I talked with two other measuring houses and they both refer to Lowell Foster’s book which says there are no bonus tolerances applied to profiles. Also, another fellow said that ASME Y14.5M-1994 does not recommend bonus tolerances on profiles.

Can you please advise me? This call out applies to the surface of an outlet duct. Thanks in advance.



Where should I begin? Profile is not allowed (as any surface control isn't) to use a MMC symbol after the geometric tolerance. It is allowed to reference datum features of size at MMC. That would not increase the size of the profile tolerance zone, but it would allow that zone to move (datum shift) as the datum feature departs from its virtual condition or MMC as appropriate. Unless position appears on the design drawing, you can't treat profile as though it is position. Mainly, because it isn't position. They are vastly different controls. There is no greater authority on GD&T than Lowell Foster. Lowell is long retired now, but was the chairman of the current Y14.5M-1994 standard on Dimensioning and Tolerancing.

The expert you named is very knowledgeable and I can't imagine him making a statement in his book that advises readers to "Just treat your Profile control as though it is Position." I have to assume they just read it wrong.

Good luck. In the future, please consider me for any on-site work you may have. I promise not to lead you astray.


Subject: Controlling Multiple Angles With One Orientation Symbol


Good afternoon, it’s been a few years since we last talked. I had your training in the 90’s while employed at a disc drive maker. As you can see I am no long with them, but have moved to the Aerospace industry. I know you are a VERY busy guy, but I have two pressing questions. I’ve attached both drawings for your review. There is only one question on both drawings and they are related (I think). I can find examples in your text, but not directly related my situation. What I want to do is control a feature for both angularity and perpendicularity. I want to reference both datums in the same geometric control with one geometric characteristic symbol, so that I can set the part up on a primary datum feature to control angularity and then onto the secondary datum feature to control perpendicularity. Can one symbol be used to control multiple angles?

Any help you may provide would be greatly appreciated.



The answer to both questions is yes. What you have done is not only correct, but commonly done. It is a question that comes up so much that the Y14.5 committee was considering combining perpendicularity, parallelism and angularity into one symbol so that anyone could control multiple angles in the same control without this question arising. In the end, they decided not to do it, in that they were worried about losing the specificity that each symbol brings to the language. When this comes up, most of us do exactly what you have done. Just choose one of the geometric characteristic symbols for orientation (I usually choose the one that is the relationship to be held to the primary datum feature) and reference both datums in the control.

It's good to know you ended up with another good job in the same beautiful state of Colorado.

Keep in touch.

Subject: Cones


How are you? Question: In dimensioning a conical surface, if I have a linear dimension with a +/- tolerance and a gage diameter with a concentricity tolerance attached to it, and then a note that states the angled surface is to be concentric to the datum within a value---is this not over-dimensioned? I think there should be the linear +/-, the gage diameter and then a profile tolerance directed at the surface in question and that's it………but that's just me. What do you think?




If you add a basic angle for the conical surface and reference the datum axis in the profile of a surface control, I would agree that is a much better approach than they have used. See the following Conicity control illustration. It is similar to what you are talking about.



Subject: Angle dimensions


Paragraph 2.3.3 and FIG 2-2(b) in the Y14.5 standard do not seem consistent. In 2.3.3 it says, “Where angle dimensions are used, both the plus and minus values and the angle have the same number of decimal places.” Following those rules I’d expect to see the dimension expressed as 25.6 degrees followed by 0.0 over -0.2. Instead FIG 2-2 (b) shows 25.6 degrees followed by 0 over-0,2. Am I missing something or is there an error in the picture?



First of all, the rule shown in 2.3.3 is a stupid rule. But, I do believe you have discovered an inconsistency. I think the mistake is in not showing a zero in front of the .2 tolerance in the 2.3.3 paragraph example. That is inconsistent with illustration 2-2b. The reason seems to be that 2.3.3 is trying to comply with the inch rules of having the tolerance and the dimension it tolerances shown with the same number of decimal places. Inches also uses no leading zero in front of the decimal point, while millimeters requires it. The question is, what does an angular dimension and tolerance shown in degrees require?

Second, they are showing an illustration in FIG 2-2b that is actually complying with the metric system rule of when a tolerance on any dimension has either a plus or a minus tolerance that is zero, no plus or minus sign is shown on that zero tolerance, no decimal points are shown and only one zero is used. See 2.3.1. The credo of this the Figure is, "When in doubt, treat it just like any other metric dimension and tolerance." The problem is that the 2.3.3 section treats the angular tolerances as though it is in inches.

Maybe the problem arose because the text in 2.3.3 came right after they wrote the rules for inches, but the illustrations are separately created.

Now, the solution to the dilemma. In section 1.1.4 Figures, it says, "The figures in this Standard are intended only as illustrations to aid the user in understanding the principles and methods of dimensioning and tolerancing described in the text." In the ASME committees this passage is understood to mean that if the illustrations ever disagree with the text in the sections, then the words win and the illustrations lose.

I hope everything is going well with you. Keep in touch.


Thanks, I was aware that text supersedes the figures, but I wanted to clarify the cause of the inconsistency. I do have one last question on this issue, what about the use of trailing zeros when the tolerance is not explicitly specified? I don’t see them used in the standard, but I don’t find any supporting information either? If I’m not allowed to use trailing zeros when I specify a metric dimension, I can’t easily use the general tolerance notes in the title block.



You've certainly hit on the other problem. In inches, you may use trailing zeros, thereby invoking the correct title block general tolerance. With millimeters, you may not use trailing zeros, which makes using the title block tolerance very difficult. You may use a basic dimension and have a geometric tolerance (Profile or Position, or whatever is appropriate) in the general tolerance block for it (I mention this because Y14.5 is trying to eventually get away from all plus and minus tolerances). Or you may simply ignore the rule about no trailing zeros and thereby use the title block tolerance with the appropriate number of decimal places. Another option is simply saying in a general note something like, "UNLESS OTHERWISE SPECIFIED, ALL SIZE TOLERANCES ARE PLUS AND MINUS 0.1."

As for text in the standard to support this stuff, it is found in 1.6.1 and 1.6.2. There is no easy solution to that problem.

I'm not certain that is what you were asking about, so I answered the most common question asked about no trailing zeros and getting the title block tolerance to kick in with millimeters. If I have misunderstood the question, please clarify and write again.


Subject: Calculating Allowed versus Actual Deviations from True Position

Hello Jim,

Have you had a chance to review this? I continue to get conflicting opinions on this and cannot get PPAP approval on this. I am caught between our tool building source and our customer. Again, the crux of my question can be summed up in the formula used for the conversion of coordinate measurements to TP location. Is it correct to include in the calculation formula the deviations between actual location and theoretical true position for all three coordinate planes rather than using two planes for hole location validation? All of my other customers accept two plane calculations without issue. Normally we would use the formula: 2 x the sq. root of X squared + Y squared where X and Y are equal to the differential from actual location to the true position in the X and Y axis.

Our customer is using the formula: 2 x the sq. root of X squared + Y squared + Z squared where X and Y are as noted above and Z is the differential from the actual location of the surface the hole is in to the theoretical exact location of the surface in the Z axis.




The only time the formula 2 X the square root of X squared + Y squared + Z squared is used for the calculation of the actual deviation from true position is on spheres where a point is being located (the axis of a sphere is a point). It shows where the 3 planes of the sphere intersect in 3 dimensional space.

For cylindrical holes and shafts, the formula 2X the square root of X squared + Y squared is used because a cylindrical hole or shaft generates an axis which is only the intersection of two planes in space. This formula is just a derivation of the Pythagorean Theorem. The Pythagorean Theorem is two-dimensional and finds only a point on the axis of the hole or shaft. The correct way to employ it for cylinders is to find two points on the axis, one at the top of the hole and one at the bottom. The points are theoretically to be connected to show the axis of the maximum inscribed cylinder for a hole or the axis of the minimum circumscribed cylinder for a shaft. Probing a substitute for the hole or shaft, such as a gage pin that has been mounted into the hole, to establish the axis should show you the axis we should establish (if this is done to correctly comply with the theory). It should also show you the complexity of complying with this theory and the problems of probing the circle of the gage pin at the very top and bottom of the hole. This is why so many people just probe the hole.

So, if you establish that each of those circles generates a point that is within the positional tolerance zone, you have correctly proven the hole or shaft is in compliance with its positional requirement.

To summarize, instead of treating the hole as though it is a sphere, we treat it as though it is the two most extreme circles on the maximum inscribed cylinder axis. We don't use 2 times the square root of X squared plus Y squared plus Z squared to do this. We use 2 times the square root of X squared plus Y squared only, and we do it separately for each of the two circles described.

Hope this helps,


Subject: Tolerancing Vs. Measuring


I took your GD&T course several years ago. I am having difficulties with our manufacturing plant and the way I have dimensioned a print. Attached you find a side view of the component in question.
Focus on the 30 mm basic dimension. This dimension is coming off of Datum A (with a flatness callout). The opposite side has a profile of 0.3mm. I have stated to them that the 30mm essentially has a tolerance of +/-0.15. Their question is that if they measure with "calipers" in certain areas without defining datum A that they can rob from the flatness callout on datum feature A. To sum up they are saying that they can have in their incoming and receiving inspection a tolerance of 30+/-0.20.
I have told them that this is not possible and that they need to modify the way that they measure during incoming inspection. This would entail defining Datum A and then take the 30mm dimension with a profile of 0.3 (plus or minus 0.15).
Can you help me in telling them why they can not do what they are doing? If you need to call or can I call you to get your thoughts?




So, their point is that if they ignore the primary datum referenced in the profile control and measure the part incorrectly, it will allow the part to pass inspection? That's an interesting argument. Still, there are those pesky things known as "rules". You may not ignore the datum plane to be formed by the three highest points on datum feature A. The datum plane is established correctly, then the 30 millimeters is measured from that plane with all of the tolerance going to the profiled surface. The profiled surface has a tolerance zone consisting of two parallel planes 0.3 apart, or plus and minus 0.15 which is centered at the 30 millimeter dimension. The flatness control on datum feature A is a separately verified tolerance.

Measuring things wrong to get a compliant answer is not one of the alternatives spelled out in any of the standards.

James Meadows

Subject: Runout


How are you? I trust you are keeping busy?

A question about runout: My drawing states a runout tolerance on a diameter that is referenced to a datum axis. The diameter (feature of size) happens to be .070 long. The engineer is telling me to ignore the runout tolerance. I just can't seem to do that. I believe the length of the geometry is irrelevant. Size and any geometric characteristic, in this case, runout, should be maintained.

Am I wrong? If I'm right, is there a simple way to explain it other than whipping out the standard?




I don't know what to tell you, except that you are right. All tolerances must be respected and complied with. Otherwise, why bother fully defining the part? If the engineer is saying don't bother measuring that, because it either doesn't matter to the functionality or that he is certain it would always be within the tolerance then I guess he has that right. Not all tolerances must be measured. If one is certain (because of past experience or because of the accuracy of the manufacturing process chosen) that a requirement will met, it is the right of any company to decide it is not worthy of measurement. I would want it in writing though. If, because of that decision, the part fails to function and/or someone is injured by the product, I wouldn't want to be the one held responsible. It is all about risk.


Subject: Pin Hole GD&T

Hello Jim,

I have a GD&T question that I hope you can help me with. Attached is a drawing showing a part with 2 alignment pin holes on each face, and an angled thru hole. The angled thru hole has to align with corresponding passages in the components that mate with the 2 faces (see figure 1). I would like to minimize the positional movement of the angled passage as it relates to the pin holes. In other words, I don't want the position of the angled passage to be greatly influenced by the position of the pin holes. I would also like to make the gauging as painless as possible.
The attached print shows one scheme that we have come up with for controlling the position of the pin holes as well as the position of the angled passage. Does this scheme make sense in regards to my goal above, or is there a better scheme to achieve the same goal. If you could give me your thoughts on this, I would appreciate it.




The control on the datum pattern C holes looks good. With the seating surface B as primary and the outside diameter A as secondary, you have a viable datum scheme. You should consider putting a MMC symbol after datum feature A in that control to aid in making it easier to inspect. That would allow a gage hole to be built that represents datum feature A at its virtual condition, instead of using a contracting gauging element. Consider using the A at MMC concept every time datum A is referenced on this part.

Using the B/A/C at MMC datum scheme for the second two pin holes is OK. Using A/C at MMC on the angled hole is not advisable. Switching primary datums is where the problem begins. It doesn't assume everything has to work at the same time while seated on datum B.

Consider this approach. Position all five of the holes to A/D at MMC. If the pin holes are the same size you can even use one position control for all four of them and a second control for the angled hole. By positioning the five holes to exactly the same datums, it makes them one pattern of five holes. This is a very powerful rule in the Y14.5 standard called the "Simultaneous Requirement" rule. Even better, in the Y14.43 standard on gauging, it is called the "Simultaneous Gauging Requirement" rule. It says that when multiple patterns of features are located from the same datums used in exactly the same order and with the same material condition symbols after datum features of size, they become one pattern and must be gauged with the same gage. If there is any concern that readers of the drawing may not be aware of that rule, you can write a clarifying, redundant local note beneath each position control that says, SIM REQT.

This should accomplish exactly what you want in relating all five holes to one another, with no accumulation of error and be easily gauged.

I hope this helps.

Subject: GD&T Question Regarding Position Using only a Perpendicularity Datum


I have a quick question for you. I am going through some prints for one of our customers to clean up any errors prior to release. I have marked up a couple of items on the attached file and would like you to review what I've written and confirm that my interpretation of a datum callout on a two hole pattern is correct (or tell me I don't know what I'm talking about).

There may be other errors on the page, but I haven't finished looking through it yet.... if anything else jumps out at you, let me know.

One other thing that I don't like is having all of these dimensions on an assembly print (much of the info is replicated from the individual part prints). Our customer wants this print (with all of the dimensions) for the assembly plant to use for any test fixtures, etc. they may need to make for the final assembly.

Thanks for the help!


A Small Portion of Chris’ Markup With His Note


Everything in your note was correct except where you said that the position control should be a perpendicularity control. Position is the correct geometric characteristic symbol in that they are trying to position the two holes to each other as well as make both perpendicular to A.

Hope this helps and didn't reach you too late.


Great, thanks!

If there was only one hole on the part, then it would have to be a perpendicularity call out right? The positional call out is only necessary to relate the two holes to each other (and provide perpendicularity control).........


That is exactly correct.


Question: Positioning Dowel Holes

Good morning, Jim.

I have a quick question regarding dowel positioning.

In the attached adobe file, we show dowel hole -D- positioned within a diameter of 0.7 to datums |A|Y|Z| (y & z are casting datums)

I maintain that for dowel -E-, position within a diameter of 0.1 with a projected tolerance of 8 mm to datums |A|D|Z| (Z must be added) is sufficient to locate the dowel hole. The engineer says we need the a position tolerance of 0.7 so manufacturing can "hold the chordal distance from -D-."

What are your thoughts on this subject?

Thank you and best regards.


Dear Jim,

It's been a while since I sent you the question above.

Since I haven't heard from you, as yet, regarding my question, I have three thoughts: (1) you're either on vacation or too busy to answer, (2) I'm correct and you don't feel you have to tell me I'm correct or (3) I'm out of my question quota.

My problem is that the engineer still says, "What does Jim think?" Should I tell him that Jim thinks I ask too many questions? (lots of luck)

As they say in West Virginia, 'I gotta git me one a them there ASME stamps.'




Sorry about the delay. I’ve been busier than a one-armed paper hanger. Have you considered using a composite positional tolerance for the 11.92-12.03 holes to A/Y/Z to within 0.7 and then to each other (and A) to within 0.1? Then, you can still name one D and the other E. Also, refining the perpendicularity on each hole implies they work separately. It is done, but only if they each bolt to a different mating part.

The projected tolerance zone is good on the threaded holes and the dowel holes. It essentially acts as a refinement of perpendicularity of the dowel holes in the area where the dowel pins get inserted into the mating part.

But, to answer your question, if D is positioned to A/Y/Z to within 0.7 and E is positioned to D to within 0.1, then E is positioned to A/Y/Z to within the sum of those tolerances (0.7 + 0.1=0.8) automatically. Only if you need E positioned to A/Y/Z to within a tighter tolerance (such as 0.7), should you have both controls.

Your point is that you also want to add a Z after the A/D in the position control on E. Unless you reference another datum such as Z after A/D for angular orientation (to stop rotation in that control), the E feature will be able to rotate around the axis of D infinitely (if that is the only position control it has). If you don't add the angular orientation datum after A/D in that control and you keep the position to A/Y/Z to within 0.7, then that will control the rotation of the E hole on the part.

I don't see Y or Z on this drawing, so I assume they are your casting datums. But to reference them, they must also appear on this drawing level as well as the casting drawing.

To elaborate on the answer to your other question, yes Ted, I have been on the road so long that I've begun to call the hotel I'm staying in that night "home". Sorry, I got behind on my emails. Yours wasn't the only technical question that has gone unanswered recently. But I’m catching up!


Subject: Profile of a Surface

Mr. Meadows:

I have been trying to find an authority on surface profile tolerancing.
Well! After reading your news letter and examples, I have to look no further.

Very simply, I work for a manufacturing company of turbine blades and vanes, where the surface profile callout is common, but the interpretation in some instances is not being agreed upon.
I hope you could help us, so I would like to just give you two examples:

We have a bilateral surface profile tolerance of 0.016” (inch), let’s say, to datum A, B, C.
If the actual measurement falls in a band of + 0.012” and - 0.002”, can we say that the feature is in tolerance? Since the total is 0.014” (± 0.007”)

Or is it, not in tolerance? Since one of the bilateral bands of ± 0.008” falls out of range (at + 0.012”).
The other example is in regards of a bilateral 0.2 mm surface profile of a feature formed in a “U” shape.
Well is the tolerance applied to ± 0.1 mm all along its surface?

Or is it, for example any combination of plus and minus totaling 0.2mm?

If the latter is the case, then the width of the “U”, which is a basic dimension of 7.3 mm, could not be controlled.

To be completely honest I have not been able to find anyone that can shed some light on this, a coworker of mine went on a web side that went as far as saying that it should be up to our own interpretation and that we should make guidelines according to what we want it to be. A little far fetched if you ask me!

Could you please clarify this? I would be very grateful

Thank you,



If the Profile tolerance is .016 equal bilateral and has a specific location by basic dimensions from the datum reference frame, any probed points that are outside of the plus or minus .008 tolerance zone means the part is bad. Unless you change the drawing to say the middle of the zone is elsewhere, the produced part has violated the profile tolerance zone.

The answer to the U shaped part is the same. A profile tolerance of 0.2 that is also equal bilateral, is plus or minus 0.1 per surface.

It takes a special designation to say the profile zone is unilateral, unequal bilateral or is a series of zones all with different tolerances applying between different points. You don't have that.

Oh, and tell your coworker to stay away from that other website. Telling you that you should make your own guidelines to decide what a profile control is supposed to mean is the silliest answer I've ever heard. We have a national standard that tells us what it means.

Hope this helps.


Subject: SPC on GD&T questions

Dear Mr. Meadows,

I was referred to you by a colleague of mine who attended your training classes recently.
I have a question on implementing/ gathering SPC on a geometric dimension. How do I mention this on the print?

Because some dimensions on our prints that we give to our suppliers require that SPC (histograms, control charts, process capability studies etc.,) be applied to them. Is there a method on how I could apply SPC to GD&T? If I mention "ST" in the feature control frame, does that mean they (suppliers) can gather statistical data (histograms, control charts etc.,) on that geometric dimension? Does the same rule apply for normal dimensions other than geometric dimensions?

I would very much appreciate your help on this issue. Thanks in advance.




Specifying an "ST" on a dimension just means that its tolerance was calculated using a statistical method. It means that this tolerance has been increased over the arithmetically available tolerance giving up 100% interchangeability in mating part assemblies. It has been done to allow only those who have proven they are holding statistical repeatability and accuracy rates that are conducive to producing parts that are within an acceptable range to your functional needs to have a larger feature tolerance. This is based on the unlikely possibility that if they use that larger tolerance, it is such an aberration that it will still likely mate with the mating part or that another, better-made part may take its place in the assembly. It is done to reduce production costs.

As of this date, the ASME Y14.5 standard on dimensioning and tolerancing and other related standards have no standard way of requiring histograms, control charts, process capability studies, etc. be applied to a particular dimension. All that is stated is that an ST tolerance is to be used only by a company that has already proven that they do these things to your satisfaction.

A section for the next revision of the Y14.5 standard does include a "statistical feature control frame" that would follow a geometric tolerancing feature control frame. The "statistical feature control frame" would state some things like acceptable repeatability rates, accuracy rates, how the statistical tolerance was calculated and perhaps other pertinent statistical requirements. But that has yet to be finalized and approved. At this point in our history, all I can recommend is that you write a set of drawing notes that include the SPC requirements. If, for some reason, you don't want to include these detailed requirements in the drawing notes, perhaps you can reference a separate document that contains these requirements.

Hope this helps.

Subject: Using Parallelism on a Datum Feature


I took a very informative 5 day course that you taught at Beckman Instruments in Palo Alto about 14 years ago. I had taken several one day classes previously that barely scratched the surface of Geometric Tolerancing. One basic misunderstanding that I run into all the time is the concept that a datum is a perfect representation of the referenced feature and not the feature itself. For instance a lot of people believe that parallelism between two surfaces means that those two surfaces mirror each other exactly (dimple on one side corresponding to bump on the other). I have even had a machine shop tell me that once they machine two surfaces parallel to one another they will always be parallel regardless of what happens to that part from then on. I have also heard over and over again that you cannot ask for parallelism without a flatness callout on the referenced surface. I understand that it's bad practice because depending on how the part is set-up for inspection you can get different results, but it's perfectly legal to call out a parallelism to a datum surface without a flatness callout isn't it? I guess my question is more theoretical than practical, do I make sense or am I all wet?





Sorry for the delay. Your email got buried under a pile of others. It's interesting what misconceptions people will come up with. I have seen training classes on videotape that make the same mistake as you describe concerning parallelism. Parallelism is definitely not when two surfaces mirror each other error for error. In the context you describe (a surface controlled to a datum plane), it is, as you know, when one surface is flat, straight and parallel to a datum plane all to within the specified parallelism tolerance. The datum plane is formed by a minimum of three high points of contact on the planar datum surface.

It is acceptable to have a parallelism control of a surface to a datum plane, when the surface that the datum plane is created from does not use the geometric characteristic symbol of flatness. Flatness is often controlled by the size tolerance between two surfaces under our Rule #1. But, unless that size tolerance is sufficiently tight to do a good job of flatness, the datum feature may be out of flat so much as to allow it to experience a significant rock. If this rock is large enough, it could make any parallelism tolerance held to it have a problem with repeatability of measurement data. If you shim up the datum feature to prevent the rock and equalize the angle, it may measure parallel within acceptable limits, but if it is rocked to one extreme or another, it may not measure within the parallelism tolerance.

One of the few true things I learned about tolerancing when I was studying design in college was that the thing you are measuring from should have the tightest tolerance. This is for measurement repeatability. So, as you say, although there is no Y14.5 standard rule that says the primary datum feature must have a flatness geometric characteristic symbol used to control its flatness, it is a good idea if the primary datum feature is controlled sufficiently for flatness to make any measurements oriented to it repeatable in the data that is collected.

There is a section in the Y14.5.1 standard on the Mathematical Principles of Dimensioning and Tolerancing that deals with the allowed ways and extremes a datum feature may be rocked. I expanded on it some in my textbook on Geometric Dimensioning and Tolerancing. But, you seem to be on the mark with your knowledge of the rules. I share your shock at the lack of GD&T knowledge of others and can imagine your frustration at their adamant ignorance in not believing what you tell them when you try to set them straight. It isn't what they know that gets them in trouble, but rather what they know that doesn't happen to be true.

It was nice to hear from you after all these years.



Subject: Question on implied 90 degree angles and size tolerances


Would a part be discrepant if the general tolerance for angles is +/-1 degree and it violates this tolerance, but the part is within its size limits? See illustration.




The answer is no, the part is not within the specification limits. The passage in Y14.5 is 2.7.3. It says, "The limits of size do not control the orientation or location relationship between individual features. Features shown perpendicular, coaxial or symmetrical to each other must be controlled for location or orientation to avoid incomplete drawing requirements." These tolerances are often shown as geometric tolerances. In the instances where they are not, angular tolerances shown in the "UNLESS OTHERWISE SPECIFIED" tolerance section of the drawing apply and must also be met. As for the dimension of 90 degrees, as you know, the standard also says in, "By convention, where center lines and surfaces of features of a part are depicted on engineering drawings intersecting at right angles, a 90 degree angle is not specified. Implied 90 degree angles are understood to apply. The tolerance on these implied 90 degree angles is the same as for all other angular features shown on the field of the drawing governed by general angular tolerance notes or general tolerance block values." It goes on to say that geometric tolerances can also be used to tolerance 90 degree basic angles.


Thanks for the response…. your answer is what I expected, but I wanted to make sure.

I asked the question because we are looking into supporting form and orientation constraints in the tolerance analysis product. As I probably mentioned to you before we currently only vary size and position. More specifically, Hitachi gave us examples where they consider orientation to play an important role, and in several examples features of size were manipulated without regard to the implied 90 degree rule. Would you happen to know if ASME and ISO are the same in this regard?



Yes, ISO does orientation the same way. It does size differently for form. ISO 8015 which deals with size, says; "A linear tolerance controls only the actual local sizes (two-point measurement) of a feature, but not its form deviations (for example circularity and straightness deviations of a cylindrical feature or flatness deviations of two parallel plane surfaces). Note: For the purposes of this International Standard, a single feature consists of a cylindrical surface or two parallel plane surfaces. There is no control of the geometrical interrelationship of individual features by the linear tolerances. For example, the perpendicularity of the sides of a cube is not controlled and, therefore, it requires a perpendicularity tolerance dictated by the design requirement." Under Maximum Material Principle, it says, "If for functional reasons there is a requirement for the mutual dependency of the size and orientation or location of the features, then the maximum material principle (circled M) may be applied (see ISO 2692)." This, of course, deals with features of size that use geometric controls such as perpendicularity or position of features of size of zero at MMC.


Close Window