What Machinists Actually Do When Engineering Drawings Are Missing Dimensions

May 10, 2026

The Print Says What It Says, Which Is Nothing

Somewhere between the engineer's CAD seat and your toolbox, a dimension went on vacation. You've got a part that needs to ship Thursday, a drawing dated 2014, a revision letter nobody remembers approving, and a hole pattern that's referenced to a datum that doesn't appear anywhere on the page. The supervisor wants to know why the first article isn't done. The engineer is in a meeting. You are, as always, the last adult in the building.

This is the actual job. Not cutting metal. Cutting metal is the easy part. The job is figuring out what the drawing meant to say before someone with a clean shirt decided it was done.

The Standard Procedure for Missing Dimensions on Engineering Drawings

There is a procedure. It's not in the ISO handbook. It's not in your apprenticeship binder. It evolved on its own, the way calluses do, and every machinist over forty performs it without thinking.

Step one: Stare at the print. Not casually. The full stare. The one where your eyes go slightly out of focus and you start to believe, against all evidence, that the dimension might appear if you wait long enough. It won't. But the staring is important because it's also where you do the math.

Step two: Check the other views. Sometimes the dimension you need is on the section view, hiding behind a leader line that crosses three other features. Sometimes it's on sheet 2 of 1. Sometimes it's on a detail balloon that was supposed to be enlarged but is currently the size of a tick mark. You find it about thirty percent of the time. That's not a great average, but it's better than the alternatives.

Step three: Add it up from what's there. Overall length minus the two known features minus the chamfer call-out equals the missing dimension. Congratulations, you are now doing the engineer's job, except you're doing it on a stool next to a coolant puddle instead of in an ergonomic chair.

Step four: Scale the drawing. Everyone says don't scale the drawing. The note literally says DO NOT SCALE DRAWING in the title block, usually in bold. Everyone scales the drawing. Anyone who claims they don't is lying or new.

When the Title Block Lies

The title block is supposed to be the one part of the print you can trust. Material, finish, tolerances, revision, scale, units. It is, in theory, the contract.

In practice, the title block is where assumptions go to die. The default tolerance block says ±0.005 on three-place decimals, but the part has a slip-fit bore that obviously needs to be tighter, and there's no callout. You check the assembly print. The assembly print references a part number that was superseded two revisions ago. You check the BOM. The BOM is a PDF from someone's email. The someone no longer works here.

So you make a decision. You hold the bore to a press fit because the part next to it is a bearing and bearings don't care about your tolerance block. This is called "engineering judgment" when the engineer does it and "going off the print" when you do it. The terminology depends entirely on whose desk the QC report lands on.

The Phone Call You Don't Want to Make

Eventually, you give up and call the engineer. This is a defeat, and you both know it.

The conversation has a script. You say, "Hey, quick question on print 47-something-dash-B." They say, "Which feature?" You say, "The counterbore depth on the back face." There is a pause. You can hear them clicking through their model. They say, "It should be right there." You say, "It's not." There is a longer pause. They say, "Huh." Then they say one of three things:

1. "Just make it the same as the other one." (Which other one? The drawing has four counterbores. None of them are dimensioned. This is what brought you here.)

2. "Let me get back to you." (They will not get back to you. The part will ship anyway.)

3. "Use your best judgment." (This is the magic phrase. It means: I don't know either, and if it's wrong, it's now your fault.)

You hang up. You go back to the machine. You make a decision based on what makes mechanical sense, what fits the mating part you already cut last week, and what won't get you yelled at if QC measures it. You write the number on the print in pencil. You initial it. The next time this part runs, the next guy will see your pencil mark, mutter something, and use it. That's how tribal knowledge gets made. One pencil mark at a time.

What Veterans Do That New Guys Don't

A new machinist sees a missing dimension and stops. He's been trained that the print is the law, and if the law is silent, work stops until the law speaks. This is correct in theory. It is also why he produces twelve parts a shift and you produce forty.

What you do, after twenty years of this, is read the part instead of the print. You look at the feature and ask what it's for. A hole near an edge with no positional tolerance is probably a clearance hole — make it loose, you're fine. A hole in a bearing journal with no callout is definitely a press fit — hold it tight. A surface with a 32 finish symbol next to a surface with no symbol is telling you the second surface doesn't matter; turn it once and move on.

You're not guessing. You're interpreting. There's a difference, and the difference is twenty years.

You also know which engineers to trust and which to verify. The senior guy who used to run a Bridgeport before he got his degree — his prints are clean, and if something's missing, it's missing on purpose because it doesn't matter. The new guy three months out of school — his prints are catastrophe. Over-dimensioned, contradictory, datums that float, GD&T callouts that don't close. You spend more time deciphering his work than cutting it.

The Tolerance Stack Nobody Asked You to Do

Here's a fun one. The print gives you four dimensions in a chain. The overall is also dimensioned. The chain adds up to a number that does not match the overall. The difference is 0.015. The tolerances are ±0.005.

You are now the proud owner of a tolerance stack problem that the engineer either didn't notice or noticed and didn't fix. You have to pick which dimension to honor. You pick the one that mates with something. The other ones are "reference," whether the print says so or not. You make a note. You don't tell anyone. The part ships. The world continues to spin.

A Brief Q&A With Yourself at 2 AM on Second Shift

Q: What do you do when a dimension is missing and the engineer is unreachable?

You make the part the way it obviously wants to be made, and you write down what you did so when someone asks in six months, you have an answer. The answer doesn't have to be right. It has to be defensible.

Q: What about GD&T callouts that don't close?

Treat them as suggestions. A position callout to a datum that isn't established is not a callout, it's a wish. Hold the feature to the tightest interpretation that makes the part functional. If QC complains, point at the print and shrug. The shrug is half the job.

Q: What if the missing dimension is critical?

Then it's not missing. It's somewhere. Check sheet 2. Check the detail view. Check the section. Check the model if you have access. Call the engineer. Call the engineer's boss. Stop the job in writing. There's a difference between a missing chamfer callout and a missing bore diameter, and you know which is which.

Q: What if you make the wrong call?

You make the wrong call sometimes. That's the trade. You document, you learn, you don't make that call again. The guys who never make wrong calls are the guys who never make any calls, and those guys are slow, and slow guys get laid off first.

The Part Ships Anyway

That's the thing nobody tells the new hires. The part ships anyway. The drawing was incomplete, the engineer was in a meeting, the tolerance stack didn't close, the revision was wrong, and the part shipped on time and worked fine. It worked fine because someone on the floor — probably you, probably at 11 PM, probably while everyone else was at home — looked at it and figured out what it was supposed to be.

This is not in your job description. It's not on your performance review. It will not be mentioned in the quarterly meeting where the engineering manager talks about "our integrated design-to-manufacture process." But it is, in fact, the process. You are the process. The CAD seat is decoration.

The drawing is a starting point. The part is the finish. Everything between those two things is muttering, pencil marks, and the kind of judgment that takes a couple of decades to build and about four seconds to apply. The print says what it says, which is sometimes nothing, and you keep cutting anyway because that's what the job has always been.