Glazing Bars Explained: Georgian to Edwardian Patterns

A glazing bar is the one part of a sash window doing two jobs at once, holding glass and carrying the look of the whole elevation, and it has the least timber in it of anything in the window. On a fine late-Georgian bar the solid neck between the two glazing rebates is about the thickness of a pencil. That is why bars snap, why they rot from the middle outwards under sound paint, and why matching one properly costs more than anybody expects.

In Scotland they are called astragals; in an English joiner's shop an astragal is a half-round moulding and a bar is just a bar. Either way they are the most informative component on a window, because both the width and the profile moved with the decade.

The neck, and why everything follows from it

Cut a bar in half and it is a single piece of timber, usually 20 to 22mm from the outer face to the inner face on a domestic sash. The outer face is rebated on both sides to take the glass and the putty. The inner face carries a moulding stuck along each edge, running the full length. What is left in the middle, between the two rebates, is the neck, and on good Georgian work it can be as little as 6 to 8mm.

Six millimetres explains most of what happens to glazing bars afterwards. It is why one snaps when a ladder leans against it, why they let go at the crossings when a painted-shut sash is forced, and why a bar with any decay in it has no useful strength left at all.

It also settles the timber specification. The stock has to be straight-grained and quarter-sawn, with no knot anywhere along its length: a knot in a 7mm neck is a fracture that has not happened yet, and quarter-sawn material moves least across a section this small, so the mouldings stay tight where the bars meet. Ask about that before anyone starts machining, or it gets decided by whatever is in the rack.

Where bars actually fail: the crossings

Bars carry very little load. What they do is hold four small pieces of glass and their putty, and the joint where two bars cross decides whether the window lasts a century or twenty years.

On good work they are halved together where they meet, a housing cut in each so the two interlock, and the mouldings are then scribed so each one runs through the crossing without a break. Done properly you find the joint with a fingernail rather than with your eye. On cheap work the mouldings are mitred instead, which is quicker and needs far less skill.

A mitre in a moulding this small will open. The timber moves with the seasons, the joint is end grain against end grain, and once there is a hairline gap the paint bridges it for a winter and then splits. Water then sits in the middle of the sash, a long way from anywhere it can drain, behind a paint film that still looks sound from the room.

Three checks, from inside, with the sash closed:

• Press a thumbnail into each crossing. Sound timber gives nothing. A crossing that feels corky has been wet for years.
• Sight along the bar at a shallow angle and look for a dark hairline running diagonally out of the crossing. That is an opened mitre, not a crack in the paint.
• Check the putty on the four panes around any crossing you are suspicious of. Putty nearly always goes first, and the bar goes because the putty went.

Widths and mouldings, decade by decade

Bar widths shrank steadily as glass got better and cheaper. These are the widths commonly found in London, measured across the face on the inside.

Treat that as a guide and not a rule. Bars were run with whatever planes the joiner owned, and suburban work ran a generation behind the fashionable squares. But if a house dated 1875 has 25mm ovolo bars in a twelve-pane sash, the window is either much older than the house or much newer, and the rest of the evidence will say which. Knowing what a genuine Georgian sash is supposed to look like is half of that judgement.

The moulding worked on the inner face carries the period more precisely than the width does, and it is the part a stock joinery profile gets wrong.

Ovolo. A quarter-round with a small fillet each side. Early, heavy, the default on late-seventeenth and early-eighteenth-century work. From inside it reads as a fat, blunt bar.

Lamb's tongue. A cyma reversa with a flat quirk, so the moulding rises and then rolls back on itself. The characteristic mid-to-late Georgian profile, and the one most people picture when they picture a Georgian window. It throws a far finer shadow than an ovolo of the same width.

Ogee. A double curve, common right through the nineteenth century. Easier to run than a lamb's tongue and slightly less crisp in the shadow it casts.

Gothic and margin bars. Curved and pointed patterns in the upper sash, used from the 1790s in Regency work and revived in Victorian and Edwardian villas, usually with narrow margin lights down the sides. The Edwardian arrangement gives itself away: small panes above, one clear sheet below, sometimes bordered with coloured or textured glass. It is one of the quickest ways to separate Edwardian windows from Victorian ones.

None of these are interchangeable, and the reason is light rather than pedantry. A bar catches light along its whole length, so the wrong moulding reads as a different colour from its neighbours even under identical paint, and an elevation repeats that difference thirty or forty times. Take the profile off a sound bar with a gauge, a card pared to fit, or a pencil rubbing. Any of the three beats a name: two joinery works can both call their moulding a lamb's tongue and neither will have run it the same way.

Why so many London sashes have none

A great many houses built with divided sashes now have plain single panes, and it is usually not vandalism by a previous owner. Sheet glass became cheap in the middle of the nineteenth century, large panes became what people wanted, and sashes were reglazed or renewed accordingly across whole streets. All that glass then hung on one meeting rail instead of being spread through a lattice of bars, and the joinery had to answer for it, which is what the horns on a Victorian sash are doing there.

If you want the bars back, start with evidence rather than a pattern that looks about right. The rear elevation is usually the least altered part of the house, and top-floor windows were often left alone because they were small and not worth the money. Look along the terrace for the house nobody modernised, and ask the borough archive about street photographs. A conservation officer will want something of that kind, and so will any joiner who would rather not guess.

Then understand that putting bars back means new sashes rather than a modification to the ones you have. A plate-glass sash has no housings in its rails and stiles for bars to sit in, and its members were sized for a different job. Cutting fresh mortices into hundred-year-old rails is not a repair anybody should be offering you. New sashes worked to the recovered pattern and hung in the original box is the honest version, and a fair amount of the sash window work we take on across London turns out to be exactly that.

The weights have to be reset either way, because a bar-divided sash weighs a different amount from the single-pane one coming out. Anyone who fits new sashes without opening the pockets is handing you a window that creeps.

True bars, dual bars and the grid stuck on the glass

If you are being quoted for new sashes or a glazing upgrade, this is where the money and the appearance both sit.

The reflection is the giveaway with anything other than a true divided bar. On a genuinely divided sash each pane reflects on its own, so the sky breaks at every bar and the break moves as you walk past. Behind a grid stuck to one large unit, the reflection runs unbroken across the whole sash and the bars appear to float on the surface of it. Conservation officers read this from the pavement, and once you have seen it you cannot stop seeing it.

The constraint behind that table is depth. A standard sealed unit runs 24 to 28mm overall, which no period sash rebate was ever cut to take. Slim units at roughly 11 to 14mm can be made to work in a full-size pane, but a 19mm bar has to carry two unit edges, two beds of compound and two rebates in a neck that was never drawn to lose any timber at all. That is most of the argument in our guide to what slim double glazing will and will not fit.

Which panes need safety glass, and which are let off

Two things decide this, and the age of the window is neither of them.

The first is where the glass sits. Approved Document K treats glazing as being in a critical location if it is wholly or partly within 800mm of the floor, or within 300mm of the edge of a door and up to 1500mm above the floor. In a London house that catches basement and lower-ground sashes, anything glazed down to a low cill, and the sidelights beside a garden door. Glass in a critical location is normally expected to reach Class 3 when tested to BS EN 12600, which in practice means toughened or laminated, and Class 2 where the pane is more than 900mm wide or high.

The second is the size of the individual pane, and this is where divided sashes are let off. Ordinary annealed glass is accepted in a small pane, meaning an isolated pane or one of a number held in glazing bars or in traditional leaded or copper lights, whose smaller dimension is no greater than 250mm and whose area is no greater than 0.5m², both measured between the glazing beads or similar fixings. That annealed glass has to be at least 6mm thick, or 4mm in leaded and copper lights where fire resistance is not a factor. A Georgian six-over-six with panes around 250 by 350mm sits well inside those limits, which is why a low divided sash does not have to be filled with toughened glass to be lawful.

Repair is a separate question again. Putting a new pane into the sash you already have is not building work, so nothing here is notifiable, and the boundary between repairing and replacing a window is worth reading before somebody tells you a certificate is needed.

Two practical points if you do end up buying safety glass for a divided sash. Each pane has to carry a permanent mark identifying the maker, the product standard and the BS EN 12600 class, and that mark has to remain visible once the glass is in, which is discreet on a full-size pane and conspicuous on one 250mm across. Agree where it goes before the order is placed. And toughened glass cannot be cut after it is made, so every pane has to be ordered to a finished size taken off the sash itself. On a hand-made window no two openings are quite the same, and a millimetre of optimism buys a new pane rather than a trim.

Replacing a broken bar: the sequence on the bench

Once the sash is off the cords and on the bench:

1. Hack out the putty either side of the broken bar and take the adjoining panes out. Keep the original glass and label it. Old cylinder and crown glass cannot be bought back, and it is a large part of what the window is worth.
2. Cut the damaged bar out at the crossings, leaving the halving housings in the sound bars intact wherever you can.
3. Take the profile off a sound bar in the same sash. A card template or a lead gauge pressed over the moulding is enough.
4. Machine or hand-run the new bar in matching timber, watching the grain direction. Quarter-sawn stock moves least across a 7mm neck.
5. Halve the new bar into the existing housings, scribe the mouldings at the crossings so each runs through, then glue and pin.
6. Prime everything, rebates included, before glazing. Bare timber draws the oil straight out of the putty fillet and the putty is dead within a couple of years. This is the commonest reason re-puttying fails early.
7. Bed the glass on a thin back-putty, sprig it, and run the external fillet to the same line as the neighbouring panes.
8. Paint the fillet and take the paint 1 to 2mm onto the glass, sealing the joint between putty and pane.

Steps three to five carry the cost. The machining is quick. Getting the moulding to sit down and run through at the crossings is not, and it is bench work rather than site work. Repairs of this kind are ordinary sash window restoration rather than anything exotic, and almost always cheaper than the replacement quote sitting next to them.

Taking the template yourself

You do not need to know whether your bars are lamb's tongue or ogee, and be wary of anyone who asks you to settle it over the phone. Three things let a joiner price the work properly, and you can gather all of them in ten minutes: a pencil rubbing or a card template taken off a sound bar in the same window, that bar's width across the face measured with a rule rather than a tape, and two photographs of the sash, one square on and one from about forty-five degrees in good daylight. The angled shot shows the shadow the moulding throws, and it usually shows whether the bars are real.

If a bar has snapped rather than rotted, keep the pieces. A broken bar with its moulding intact is the best template there is, and it saves the one part of the job that costs real money.