Setting blocks
A Socratic walk-through of setting blocks — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why is a heavy pane of glass set on two small blocks rather than supported evenly all along its frame?
A glazier lowers a heavy pane into its frame and rests the whole thing on two small plastic pads a few centimetres long. The rest of the sill, a metre of perfectly good metal, carries nothing.
Every instinct says that is backwards. Spread a load and you reduce the stress; concentrate it and you invite a break. So why does the trade insist on two points, and treat "bedded solid along the sill" as a defect rather than extra care?
Reasoning it through
REASONING #Test the instinct first. Spreading a load reduces stress if the load actually spreads. For that you need either a surface flat enough to touch everywhere, or a material that yields locally until it does. Which of those does a window offer?
Not flatness. An extruded or rolled sill has a manufacturing tolerance, it deflects under the pane's own weight, and it sits in a building that settles and moves. Its bearing surface is flat to a fraction of a millimetre at best, less once sealant, swarf or a bead of paint is on it.
Not yielding, either — and this decides the question. Glass is elastic right up to fracture and then stops; it has no plastic range in which an overloaded contact patch flows and hands its load to a neighbour. Nor does it break where it is squeezed. It breaks in tension, from a flaw, and its worst flaws are the microscopic chips left along the cut edge — so the weakest part of the pane is exactly the part the sill asks to carry everything.
Put those together and "bedded solid" is not the option you imagined. A rigid pane on a not-quite-flat sill lands on whatever high spots exist: unknown in number, position and area, glass hard against metal. That is not distributed support but an arbitrary set of point loads chosen by the tolerances of the day, applied to the flawed edge, and it can be neither specified nor inspected because nobody knows where the contacts are.
Two blocks are not a concession, then. They make the load path known: chosen position, chosen area, chosen material — a resilient polymer of specified hardness, stiff enough not to squeeze out under the weight, soft enough to conform over the edge's irregularities and keep glass off metal.
Where should they go? Here the trade rule is worth checking rather than trusting. Treat the pane as a uniformly loaded beam of length L on two supports set a distance a in from each end. The overhanging ends give a hogging moment at each support of wa squared over 2; between the supports the sagging moment is w(L - 2a) squared over 8, less that same hogging term. The largest moment is smallest when the two are equal, which gives a squared = (L - 2a) squared over 8, so a(2 + 2 root 2) = L, and a = 0.207L. The trade says quarter points: 0.25L. The rule of thumb sits a couple of per cent of the pane's length from the calculated optimum — close enough that the arithmetic vindicates the habit rather than correcting it, which is the usual outcome when you check a rule that has survived a century of glazing.
Two more jobs fall to the same block, and they are not structural. A glazing rebate gets wet — driven rain, condensation running down the inner face — and drains through weep holes at the bottom; a pane sitting flat on the sill dams that channel and stands its own edge in water. For an insulating unit that is a durability question with money attached, since water attacks the edge seal that keeps the cavity sealed. The blocks also preserve clearance, so that when frame and glass move differently with temperature the metal never arrives against the glass.
Is any of this inertia? Some of the detail is contractual rather than physical: manufacturers specify block material, size and position as a warranty condition, and a glazier follows it on a small light where the loading would forgive almost anything. That is a liability rule riding on a mechanical one.
The analogy
THE ANALOGY #Think of setting a slab of stone down on a gravel drive and calling it evenly supported. Every stone in the yard is "supporting" it, in the sense of being underneath; in fact three or four high pebbles carry the lot, and which ones is a matter of chance. Two flat timbers laid down first support less of the slab and protect it far better.
Gravel is visibly irregular, whereas a machined sill looks flat — the high spots are thousandths of a millimetre, and the reason that is still enough is that glass, unlike almost anything else on a building site, cannot yield locally to share what it is given.
Clarifying the model
THE MODEL #The model is not "concentrate the load". It is that support you cannot specify is worse than support you can, and that with a material which neither yields nor tolerates edge tension, the achievable choice is between two known contacts and an unknown number of accidental ones.
Two refinements connect this to its neighbours. A flaw in glass raises local stress enormously — the ordinary story of stress concentration — but here the flaws cannot be removed, since cutting the pane to size makes them. The design move is not to eliminate flaws but to keep the load off them. And toughening changes the tolerance without changing the logic: a toughened pane carries a compressed skin that suppresses edge flaws, yet cannot be cut afterwards, so its edge condition is fixed at the factory and a bad bearing is permanent.
The test that decides it is where breakages start. If uncontrolled edge contact is the mechanism, failures in hard-bedded glazing should originate at the bottom edge, at a contact point, should be commoner in heavy or long panes, and should depend on what the pane rests on rather than on the quality of the glass. Find breakages originating in the middle of the pane, or hard-bedded and block-set panes failing at the same rate in the same weather, and this account is wrong — the cause would be wind or thermal stress, and the blocks would be doing nothing but drainage.
A picture of it
THE PICTURE #How to readThe five boxes are the jobs the bottom of a glazing rebate has to do, not stages in a process — read each as a condition to be met. The two elements below are the rival ways of meeting them. Follow the arrows out of each and count: the block pair reaches all five, the continuous bed only the first, and the four arrows it does not have are the answer to the question. Note that only one of the five is about strength; the rest are about knowing where the load is, and about water.
What became clearer
WHAT CLEARED #Support that cannot be specified is not support. A pane bedded along its whole sill really rests on a random handful of high spots, bearing on its most flawed surface, with nothing soft between and its drainage dammed. Two blocks give up the imaginary benefit of even bearing and buy a load path someone chose: known area, known position, a resilient interlayer, a clear route for water. And the quarter-point rule the trade teaches sits within a whisker of the position that minimises the pane's own bending — a rule of thumb that survived because it was very nearly the calculation.
Where to go next
ONWARD #- Why toughened glass cannot be cut after treatment, and what that forces on everyone handling it.
- How the same "make the contact known" logic governs bearings under beams and machine feet.
Key terms
TERMS #| Term | What it means |
|---|---|
| Setting block | a resilient pad carrying a pane's weight at a chosen position on the sill. |
| Glazing rebate | the channel holding the pane's edge, drained through weep holes. |
| Edge flaw | a microscopic chip left by cutting, from which a brittle fracture starts under tension. |
Every term the collection defines is gathered in the glossary.