What this checks
Paste the segments of a dimension chain and state the target length. Each segment contributes its nominal value and its tolerance, and the checker adds them in the worst-case sense: every segment long at once for the high limit, every segment short at once for the low limit. The result is the range the assembled length can actually take, compared against the target.
This is the arithmetic behind a question that comes up on every assembly: will these parts fit, and how much room is left when they do. Doing it by hand means adding the nominals, adding the tolerances separately, and then not being sure whether the answer is the sum of the tolerances or the root sum of squares. This page does the worst-case form, which is the one that never lets a bad assembly through.
How to read the verdict
CLEARANCE means the assembled length is inside the target range even at the worst combination, so there is always room. INTERFERENCE means it is outside even at the best combination, so the parts cannot fit as drawn. MARGINAL means the answer depends on which way the tolerances happen to fall: sometimes it fits, sometimes it does not, and the note says which failure you would get.
The clearance minimum and maximum are the two numbers to carry into a discussion. A design whose minimum clearance is 0.095 has room to spare. The same design with a minimum clearance of 0.002 is technically inside the target and will produce arguments on the floor.
Worked example
The sample has three segments: 10.000 ±0.002, 20.000 ±0.003 and 0.375 ±0.001, and a target of 30.475. The nominals sum to 30.375, and the tolerances sum to ±0.006, so the assembled length runs from 30.369 to 30.381. Against a target of 30.475 the clearance runs from 0.094 to 0.106, and the verdict is clearance: it always fits. Change the target to 30.000 and the same parts become interference, because the shortest possible stack is already longer than the target.
Worst case, not statistical
The root-sum-square method gives a smaller, more realistic band, and it is the right tool for setting tolerances during design. It is the wrong tool for answering whether a specific assembly can fail, because it assumes the errors are independent and normally distributed, and a stack of three parts from one machine on one afternoon violates both assumptions. Worst-case addition needs no assumptions and never understates the range.
Direction matters too. Segments that run in opposite directions subtract rather than add, so a chain with a negative-going segment has to be entered with that segment negated. The tool adds what it is given.
What it will not do
It does not handle geometric tolerance, angular contributions, or the effect of a datum shift, and it does not know which segments subtract. Those require the drawing. What it does is remove the arithmetic from a question that is asked constantly and answered by hand with a calculator and a certain amount of hope.
Questions
Why worst case rather than root sum square?
Because the question this page answers is whether an assembly can fail, and worst-case addition is the only method that never understates the range. Root sum square is smaller and more realistic over a production run, but it relies on the errors being independent, which is not true for parts made together on the same machine.
How do I enter a segment that subtracts from the chain?
Enter it with a negative value. The checker sums what it is given, so a segment written as -0.375 in the length column reduces the total. It does not try to infer direction from the drawing, because the drawing is the only place that information exists.
What if I leave the target empty?
The page still sums the chain and reports the worst-case high and low limits, just without a clearance verdict. That is useful on its own for checking what a stack of dimensions actually adds up to before deciding what the target should be.
Does the target have its own tolerance?
Yes, and it has two fields so an asymmetric target can be stated. A target of 30.475 with +0.000/-0.020 is common on a slot, and dropping that asymmetry would overstate the clearance on one side.
Can a row carry its own unit?
Yes, the same as the other checkers on this site. Values with a unit attached override the dropdown, and the sum is computed in a single base unit so a chain that mixes inches and millimeters still adds up correctly.