Stack ten chairs from a good order and the column stands plumb, each seat sitting the same distance above the one below it. Stack ten from a bad order and by the seventh chair the whole thing leans, the gaps run uneven, and somebody has to hold the top while somebody else wheels the dolly. Same drawing, same nominal dimensions, completely different behavior in a corridor.
The explanation is not workmanship in the way most buyers assume. Venues that stop having this problem tend to have moved to commercial banquet chairs from a manufacturer who controls the stacking geometry as a tolerance question rather than a styling one. What separates the two orders is an accumulation effect that civil and mechanical engineers know well and furniture buyers almost never hear named.
Every Dimension Has a Range, Not a Number
A drawing that says a leg is 17 inches does not mean every leg will measure 17.000 inches. It means the leg should fall inside an allowed band around 17, and the width of that band is the tolerance.
General tolerance standards exist precisely so this does not have to be written next to every dimension. ISO 2768 lets a manufacturer put one callout in the title block that sets the default permitted variation for every feature without its own tolerance, across four classes running from fine to very coarse. The American equivalent for geometric characteristics, ASME Y14.5, covers the same territory for form and position. Reading how dimensioning and tolerancing is formally specified makes clear how much a single title-block note is quietly deciding.
Choose a coarse class and you save money at the machine. You also widen every band on the drawing at once.
Stacking Is Where Small Variations Stop Being Small
Here is the part that surprises people. A variation that is invisible on one chair is not invisible on ten, because the chairs sit on each other.
If each chair’s nesting surface can sit a sixteenth of an inch high or low within tolerance, a single chair looks perfect. Ten chairs in a column can accumulate more than half an inch of drift, and because the deviations are not all in the same direction, the column can also twist. That accumulation is what engineers call tolerance stacking, and it’s the reason a set can pass a per-unit inspection and still behave badly as a stack. The formal treatment of engineering tolerance covers exactly this: the interesting number is rarely the individual part, it’s the assembly.
A stack is an assembly. Nobody puts it on a drawing, which is how it ends up untested.
Where the Drift Concentrates on a Banquet Chair
Four features do most of the damage, and they’re worth naming because they’re the ones to ask about.
The nesting shoulder, the small ledge or bend where the next chair’s frame comes to rest, sets the pitch of the whole column. The rear leg splay decides whether chairs sit concentric or gradually walk sideways. Weld fixture repeatability determines whether chair number 400 has the same geometry as chair number 4. Frame straightness after powder coat matters because heat cycles move steel, and a frame measured before the oven is not the frame you receive.
Any one of those held loosely produces a stack that leans. All four held loosely produces a stack that has to be built in fives.
Why Two Orders of the Same Model Behave Differently
This is the question that brings people to the problem, usually a year after the first order.
Chairs bought in one production run share a fixture setup, so their deviations tend to cluster in the same direction and the stack stays tidy. Chairs bought two years later, from a different run or a different plant, cluster differently. Mix the two and the accumulated drift roughly doubles, because you’re now stacking parts from two distributions instead of one. The chairs are all in tolerance. The stack is still wrong.
Manufacturers who take seating seriously test to the contract seating standards published by BIFMA, which measure strength, stability and durability. Those protocols tell you a chair will survive. They do not tell you a hundred of them will stack straight, so that question has to be asked separately.
What to Ask Before a Bulk Order
Five questions get at the geometry rather than the finish:
- What tolerance class governs the frame, and is it stated on the drawing?
- Is the nesting shoulder dimensioned and inspected, or is it inherited from the bend?
- Are frames checked for straightness after powder coat, not before?
- Will a repeat order in two years come off the same fixtures?
- Can you supply a photograph of ten units stacked, from the actual production lot?
That last one costs a supplier five minutes and tells you more than any spec sheet.
The Storage Consequence Nobody Budgets
A stack that leans gets shortened for safety, and that decision is made on the floor rather than in a meeting.
Crews stop at six instead of ten because the eleventh is unstable. A room’s worth of chairs then needs half again as many stacks, and the storage closet that was sized correctly is suddenly too small. Nobody logs this as a furniture defect. It shows up as a space complaint, or as a slow reset, or as chairs stored in a corridor where the fire marshal eventually finds them.
The cost lands somewhere else than the cause, which is why it survives for years.
Precision Is Cheaper Than It Sounds
Tightening a tolerance class does raise unit cost, and past a point the return falls away sharply. The useful move is narrower than that: hold the tolerance tightly on the two or three features that govern nesting, and let the rest run to a general class.
That’s ordinary engineering practice and it’s how a manufacturer who understands the problem already works. It’s also why two chairs that look identical in a catalog, weigh nearly the same and carry the same load rating can differ by a factor of two in how well they stack. The difference isn’t visible in a photograph, isn’t in the price, and doesn’t appear until four hundred of them are in a building.
The Column Tells You What the Drawing Didn’t
There’s a simple field test that requires no instruments. Stack ten, stand back twenty feet, and sight down the column. If the seats step evenly and the front edges line up, the geometry is controlled. If the gaps open and close as your eye travels up, you’re looking at accumulated deviation, and it will get worse as the set ages and gets mixed with replacements.
Furniture rarely gets discussed in these terms, which is a shame, because a banquet chair is a welded steel assembly produced in the thousands and it obeys the same rules as any other one. Buyers who ask about tolerance get better stacks, shorter resets and smaller storage rooms. Buyers who ask about finish get a chair that photographs well and leans at the seventh row.