Read this before you copy any number out of this article.
Every dimension, U-value, bar count and authority reference below is either quoted from a publicly published source and labelled as such, or is our own arithmetic with the assumptions printed next to it. None of it is a product claim. Do not paste any of it into a drawing, a specification or an order without checking it against your own opening, your own panel sizes and your own authority having jurisdiction.
One more thing up front, because it decides how you should read everything after it: we build sliding windows and doors. Our swing door, folding door and entry door category pages are empty at the time of writing. We cannot sell you a French pair, and this article will say so again before the end. Everything here about hinged French doors comes from published public sources and from arithmetic we show you, not from our own catalogue.
The phrase "French patio sliding door for luxury homes" carries two adjectives, and both of them describe a photograph. Nobody has ever written a measurable definition of luxury: brochures spell it with material names instead, a mahogany front entry door, a slab of honed stone, a lever in unlacquered brass. "French" is the same kind of word. It is shorthand for a picture: two tall leaves, glass broken up into small panes, standing open onto a terrace.
That matters to anyone trying to order one, because nearly every feature that made a French door French belongs to the hinge. Once the leaves travel in a track, those features are not compromised, they are gone. One of them survives, and it survives precisely because it never did any structural work: the grid.
So the whole buying question collapses into a much smaller and far more answerable one. Which grid, built how, on which panel sizes, with which consequences, requesting which documents? That is a question that can be written into a purchase order. "French" cannot. This article exists to convert the first into the second, and it is written by a factory whose answer to "can you build me a French door" is no.
Take the picture apart and lay the parts next to a slider. Most of them do not make the trip.
| What makes it French | In the hinged pair | In the sliding version | Survives? |
|---|---|---|---|
| Nothing permanent in the middle of the opening | Both leaves swing clear; the centre of the reveal is empty air | The interlock is there whether the door is open or shut | No |
| Both halves operable | Both leaves move independently | Typically one panel stays put; a multi-track unit can stage more | Partly |
| The whole width opens | Yes, from jamb to jamb | The panels have to be somewhere when open | No |
| A leaf far taller than it is wide | Roughly two and a half to one, from published standard sizes | Set by panel width, which is the thing the client came for | No, and the arithmetic follows |
| Glass divided into small panes | Originally structural muntins holding small sheets | Entirely available, on any panel size, built three different ways | Yes |
| A reveal that stays clear | The leaves take a quarter circle out of the room or the terrace | Nothing projects into either space at any point in the travel | Sliding wins this one |
The last row is the reason this product exists at all. On a terrace where furniture, planters, a grill and foot traffic all live inside the swept area, a hinged leaf is a permanent argument with the floor plan. Buyers who want the French picture are usually buying it after losing that argument, and they should say so out loud, because it changes which compromise they are willing to accept next.
Everything else in the picture is gone. What remains carrying the entire cost of the word is a pattern of bars, and from here to the end of this article, that is what "French" means.
Start with published sizes, because the shape is measurable. Exterior French pairs are commonly stocked at sixty and seventy-two inches wide, with ninety-six inch pairs offered for wider patios, and individual leaves at thirty, thirty-two or thirty-six inches. Standard height is eighty inches, with eighty-four and ninety-six inch leaves offered for taller ceilings. One retailer describes the seventy-two by eighty unit as the most commonly stocked exterior French size, which means the default leaf is thirty-six by eighty: exactly the rectangle a catalogue would print as an entry door 36x80.
Divide height by width and the form stops being a feeling. Our own arithmetic, on those published figures converted at twenty-five point four millimetres to the inch:
| Element | Width | Height | Height divided by width | Where the number comes from |
|---|---|---|---|---|
| French leaf, thirty inch | 762 mm | 2032 mm | 2.67 | Published French door size guides |
| French leaf, thirty-six inch | 914 mm | 2032 mm | 2.22 | Published French door size guides |
| Whole French pair, seventy-two inch | 1829 mm | 2032 mm | 1.11 | Published French door size guides |
| Whole French pair, ninety-six inch | 2438 mm | 2032 mm | 0.83 | Published French door size guides |
| Published maximum single sash, one of our series | 2500 mm | 3000 mm | 1.20 | Our catalogue, published against an assumed glazing build-up |
| Published maximum single sash, another of our series | 2800 mm | 2600 mm | 0.93 | Our catalogue, published against an assumed glazing build-up |
Read the last two rows against the first two. A French leaf is between 2.2 and 2.7 times taller than it is wide. The largest single panels we publish sit at 1.20 and 0.93, which makes them roughly 1.9 and 2.4 times less slender than the leaf that gave the form its name. One of those panels is wider than it is tall.
Nothing here is a complaint about anybody's product. It is the observation that the two words in the title are pulling at different things: French wants slender leaves, luxury in a patio wall wants wide ones, and the bigger client is the one writing the cheque. Proportion loses first, quietly, before anybody has drawn a bar.
Assume the default reference leaf from published sources: thirty-six by eighty inches, carrying a grid of three columns and five rows, which is fifteen lites. Each lite comes out at about 305 by 406 millimetres, a little more than 1.3 times taller than wide.
Now put that grid onto one of our largest published panels, 2500 by 3000 millimetres, dividing the panel evenly and ignoring bar thickness for the sake of the arithmetic. You can hold one thing constant, and whichever one you pick, the other one moves:
| Choice | Columns by rows | Lites per panel | Each lite, mm | Lite height divided by width | What it costs you |
|---|---|---|---|---|---|
| The reference French leaf | 3 by 5 | 15 | 305 by 406 | 1.33 | Nothing. This is the shape the word describes. |
| Hold the lite proportion | 6 by 5 | 30 | 417 by 600 | 1.44 | Twice the lites, and every consequence of having more of them |
| Hold the lite count | 3 by 5 | 15 | 833 by 600 | 0.72 | Every lite turns landscape, one to zero point seven two against the reference |
There is no third option. Keep the panes French-shaped and you buy about twice as many of them. Keep the count and every pane becomes wider than it is tall, which is the shape the divided lite was invented to avoid in the first place. Either way, the grid you end up with on a luxury panel is not the grid on the photograph, and both versions then have to be paid for.
Counting is free and it is the most useful thing you can do before you choose a pattern. Again on our own arithmetic, using the same 2500 by 3000 millimetre panel, whose Perimeter is eleven metres: an interior vertical bar runs the full panel height of three metres, and an interior horizontal bar runs the full width of two and a half metres.
| Pattern on one panel | Vertical bar length | Horizontal bar length | Total added bar | Multiple of the panel Perimeter | Rectangles to wipe per face | Separate sealed units if built as true divided lite |
|---|---|---|---|---|---|---|
| None | 0 m | 0 m | 0 m | 1.00 | 1 | 1 |
| 3 columns by 5 rows | 6.0 m | 10.0 m | 16.0 m | 1.45 | 15 | 15 |
| 6 columns by 5 rows | 15.0 m | 10.0 m | 25.0 m | 2.27 | 30 | 30 |
The modest three-by-five pattern adds more linear metres of bar than the entire Perimeter of the pane it decorates. The six-by-five version adds more than double. Those metres are the subject of section six.
Trade sources describe three mainstream constructions, plus a fourth that is usually offered as an accessory. They look different from across the garden, they clean differently, they are treated differently by a heritage review board, and they behave very differently when something goes wrong.
| Construction | Where the bars live | Sealed units per panel | Cleaning | If a unit fogs | The catch |
|---|---|---|---|---|---|
| Grilles between the glass | Inside the cavity of one unit, nothing on either face | 1 | Easiest, both faces are flat glass | One unit, and the pattern has to be reordered exactly | No shadow line. Up close it reads as the single sheet it is. |
| Simulated divided lite with a spacer bar | Bars bonded to both faces, with a bar aligned inside the cavity | 1 | Each rectangle wiped individually, both sides | One unit; external bars can usually be re-supplied | Without the internal bar you can look through at an angle and see straight into the gap |
| Simulated divided lite without a spacer bar | Bars bonded to the faces only, nothing between the panes | 1 | Same as above | Same as above | The illusion breaks at an angle, which is exactly where a seated guest looks from |
| True divided lite | Each pane glazed separately between real muntin bars | One per lite, so fifteen or thirty per panel | Every pane cleaned separately | One small pane at a time, if the construction allows it | Fifteen or thirty times the edge seals, which is why trade sources call it the hardest construction to make efficient |
| Removable interior grille | Clipped or magnetically fixed to the room-side of the sash | 1 | Lifts off | Unit replaced as a plain pane | From outside it sits behind glass and reads much like a between-the-glass grille |
Now the part that ruins every quote comparison. A published trade comparison says it plainly: manufacturers do not name these consistently. One company calls the version with an internal bar "simulated divided lite with spacer bar"; another calls an equivalent construction "full divided light". The word "divided" appears in all three constructions and distinguishes none of them.
So ask two questions in writing, and accept only in writing: is the external bar permanent, and is there a bar aligned between the panes? Those two answers tell you how the thing cleans, how it looks from an angle, and what has to be reordered when it fails. A pattern name tells you none of that.
The whole-window figure that most authorities care about is built from four inputs. Published guidance gives the formula as Uw = (Ag x Ug + Af x Uf + Lg x Psi-g) / (Ag + Af), where Ag is the glass area, Af the frame area, Ug the centre-pane value, Uf the frame value, Lg the length of the glass edge and Psi-g the linear thermal transmittance of that edge.
Look at those four inputs and then look at a grille sealed inside the cavity. The glass area has not changed. The frame area has not changed. The Perimeter has not changed. A between-the-glass grid is arithmetically invisible to that equation, which is the polite way of saying that a decorative element carrying sixteen metres of bar on a single panel can be absent from the number on the datasheet.
A national research council publication described the consequence more than twenty years ago, and nobody has improved on the wording since. Muntin bars inserted between the panes of an insulating glass unit come with reduced R-value and reduced interior glass surface temperature, and therefore reduced condensation resistance, and all of it because the bars cause thermal bridging. The same publication adds a second effect that almost nobody expects from decoration: under ultraviolet radiation the bars can contribute to fogging, because volatile materials evaporate and then condense on the glass.
That last clause deserves a pause, because it names something the industry has its own provision for. Internal deposits that are a material problem rather than a seal problem are distinguished by their own test method in parts of the ISO 20492 family; we have written about that distinction elsewhere. Here the narrower point is that a national research body associates a decorative grid with two outcomes most buyers would not predict: a colder interior glass surface, and deposits forming on a pane you cannot touch.
Some regulators have begun closing that hole by ordering the components into the calculation. A UK specification blog reporting a trade federation seminar set out a single window carried through a series of configurations: 1.27 W/m2K as a standard window, 1.27 with a dummy sash, 1.32 with dummy transoms, 1.45 with PVC Georgian bars and 1.60 with aluminium Georgian bars. The accompanying remark is the one worth keeping: the window has not changed, the calculation has. Take that as direction rather than digits. We do not know whether those particular bars were inside the unit or applied to it, and a value that depends on one panel size will not survive being quoted at another.
Ask for the unit nobody asks for. The glass-edge figure has one, watts per metre kelvin, and published guidance puts a typical band around zero point zero four to zero point zero eight depending on the spacer, with warm-edge spacers at the lower end. Ask for the same number for the grid, not only for the spacer, and ask which method produced it. We are deliberately not printing one here for anybody's grid, including our own.
An insulated glass unit is not a rigid box. Its panes bow in and out as atmospheric pressure and temperature change, so the cavity between them widens and narrows continuously. That is why a between-the-glass grille has to be made thinner than the nominal gap, and why an insulating glass manufacturer published a note explaining the consequence in detail.
The chain runs like this. The grille is sized thinner than the gap so it cannot break the glass. When the cavity expands as the panes bow, the grille is no longer a snug fit. Movement of the glass, whether from opening and closing the panel, wind load, or air vibration from traffic, then lets the grille touch the glass and produce a tapping noise. Plastic or felt pads at the joints muffle it but cannot eliminate it. And the sentence that should stop any designer specifying a large format panel: deformations are greater in cases of larger sheets compared to smaller panes. The same source is careful to say this cannot be attributed to a fault in the glass or the window.
Read that next to the panel sizes in section two. A between-the-glass grille is at its most troublesome precisely where a luxury project puts it: on the widest panels the system will allow. This is not a defect report and not a reason to avoid the construction. It is a reason to decide the pattern and the panel size together, in writing, before either is cut, and a reason to ask how the grille is retained at the edges rather than accepting "it is sealed inside".
Open both leaves of a French pair and the middle of the opening is air. In the sliding version there is always a vertical member there: the interlock, the meeting stile where the moving panel closes against its neighbour. It is the one line in the elevation that no specification can ｄｅｌｅｔｅ, and every photograph of the room has it in it.
A divided-lite pattern is the cheapest available way to make that unavoidable line look deliberate. Turn one vertical line into six and nobody studies any of them. There is nothing dishonest about this; it is composition, and architects have always used it.
But composition has arithmetic. Divide the closed elevation into an even number of equal columns and there is a bar exactly at the centre line of the panel pair, so the interlock hides inside it. Divide into an odd number and the centre falls in the middle of a pane, and the one line you cannot remove sits in the middle of an otherwise clean rectangle:
| Columns across the closed elevation | Is there a bar at the centre | What the eye sees |
|---|---|---|
| 4 | Yes | The interlock reads as one member of the pattern |
| 6 | Yes | Same, with a finer grid |
| 3 | No | One line across the middle of the central pane |
| 5 | No | Same, and harder to forgive on a wide opening |
There is a second condition, and it is the one that usually breaks. This only works if the two panels that meet are the same width, because otherwise the interlock is not at the centre line at all. Widening one panel to make a picture-window module moves it off centre, and no amount of grid arithmetic recovers it. "All panels equal" against "one wide view panel" is therefore a decision that has to be made before the pattern, not after it.
One honesty note before you read our own numbers for this. We publish sash width and frame width for our sliding series; one series, for example, lists a seventy-five millimetre sash and a thirty-six millimetre frame. Those are profile dimensions, not the head-on sightline you see when you sit down in the room, and we are not going to imply a twenty-millimetre shape by mixing them up. If the visual width of the meeting stile decides the project, ask for it measured the way you intend to look at it, and ask for it from whoever is making the panels.
There is a second thing the word luxury reliably produces on a terrace, and it is not decorative at all: the demand that the interior floor simply continue to the paving. Our own catalogue shows why that request is not free. One of our series publishes an eight millimetre low track and a forty-three millimetre high track. Another offers a forty-five millimetre high-ｄｒｏｐ bottom track with an optional eighty-five millimetre water barrier. Those numbers exist because of gravity, not because of style.
Published detailing guidance puts the trade-off crisply. A raised threshold sheds water by its own geometry; a flush threshold has no geometry to help and has to rely instead on external fall, a continuous membrane tied into drainage, and where the terrace is not sheltered a channel drain in front of the door. One installer's summary is worth more than most technical papers: stepped thresholds insulate well, flush thresholds largely do not, because a flush threshold has almost nothing for a seal to press against and ends up depending on brush seals. That same source adds the sequencing rule that governs the whole detail: install the doorset first and build the floor levels up to meet it, because the drainage mouths have to stay visible and clear. A system supplier's technical note names the failure that follows when the order is reversed, listing sealant, mortar, insulation and terrace finishes set too high as the things that block drainage openings.
Luxury asks for the floor to carry on. Water asks for somewhere to go. Both are answered before the slab is poured rather than at the moment the door is chosen, and the interface itself is not ours. We supply the track, the drainage route through the profile and the drawings. The floor build-up, the fall, the channel drain, the membrane and how they meet belong to the building and to whoever signs that part of it.
This is the table we would rather you read than any other. A hole in a published table is not an admission that something is weak; it usually means nobody has put a number on the thing you are about to compare.
| Item | What we publish | Why there is a hole | What to do |
|---|---|---|---|
| Thermal values on window series | System-level figures for named window series | A system value belongs to a configuration, not to your opening, and none of them include a grid | Ask for the value at your panel size and glazing, and ask for it with the grid |
| Thermal values on sliding series | Not published | A moving panel has to keep clearance to move, so its sealing philosophy differs from a compression casement | Request a calculation at your size. Unpublished is uncomparable, not worse. |
| Air, water and wind ratings on sliding series | Not published in our catalogue | They are not part of our published product data | Ask for the report and check what configuration the test panel actually was |
| Roller and carriage load figures | Per carriage figures, including a large multi-wheel and a smaller driving capacity on different series | They are quoted against an assumed panel build-up, and the two figures answer different questions | Never sort them into one column. Ask for both, per sash, at your glass. |
| Maximum single sash sizes | Published per series | They move when the glass does, and that relationship is not printed | ｄｅｃｌａｒｅ your glass build-up first, then ask for the size |
| Salt spray hours and composite shear figures | Published for one specific series | One measures coating and substrate corrosion, the other the bond between insulation strip and aluminium; neither is a water, air or wind rating, and neither transfers to another series | Use them for what they measure, and ask for the ratings separately |
| Hardware cycle counts | Bench figures for named hardware | A bench count is not a service life, and handle cycles are not the same test as cycling a loaded panel | Treat them as a hardware comparison between components, nothing more |
| Meeting stile sightline | Sash width and frame width per series | Profile dimensions and head-on appearance are not the same measurement | Ask for the visual width from a stated viewing position |
| Track heights | Published per series, including low and high options | Which one you need depends on your terrace, not on the door | Decide it with whoever is designing the external drainage |
| U-value including a grid, and a linear value for the grid | Nothing | We do not hold a published figure for any product | Request it. If nobody can produce one, the value you were given came off a plain pane. |
| Line | A good answer looks like | If you leave it blank |
|---|---|---|
| Grid construction | Between the glass, applied both faces with or without an aligned inner bar, true divided lite, or removable | You will be quoted against whichever construction is cheapest for the shop |
| Pattern, expressed numerically | Columns by rows per panel, plus a drawing | The word "colonial" is a family, not a specification |
| Whether a bar lands on the interlock | Yes or no, with the reason stated from the column count | The one vertical line you cannot ｄｅｌｅｔｅ ends up mid-pane |
| Panel sizes in millimetres, per panel | Width and height for every panel, plus which ones move | Every downstream number gets calculated on a reference size |
| Glass build-up | Ply thickness, cavity, coating, fill, and resulting weight per square metre | Maximum panel sizes and roller figures lose their basis |
| Maximum single sash revalidated | The size confirmed again against your declared glass | You inherit a catalogue maximum that belongs to a lighter pane |
| U-value with the grid, at your size | One number, one size, one standard, grid stated as included | You are comparing a plain-pane number against a compliance target |
| Linear transmittance for the grid | A value, a unit, and the method that produced it | The longest thermal bridge in the assembly remains unpriced |
| Track height and external drainage interface | Low or high named, plus who owns the channel and the membrane | The flush floor arrives and the drainage has nowhere to go |
| Panel widths, equal or unequal | A module table, or a stated single wide view panel | Symmetry is decided by accident, after it is too late to change |
| Screen type and where it lives | Mesh specification and which side of the panel it sits on | A second frame line appears in the elevation nobody has drawn |
| Replacement path | A drawn pattern, bar profile, whether contoured, colour code and batch | One fogged pane becomes a whole elevation with two different grids |
| Ask this | A real answer contains | This sounds plausible and tells you nothing |
|---|---|---|
| How is the grid built? | One of the named constructions, and where the bars actually are | It has a colonial grid |
| Is the external bar permanent? | Yes or no, plus the fixing method | It looks authentic |
| Is something aligned between the panes? | Yes or no | It has a spacer. That word also means the edge spacer every unit has. |
| At which size was that U-value calculated? | A named size and a named method | It is below one point four |
| Does that number include the grid? | Yes, and how it was treated | The glass is high performance |
| How is the grille retained? | A description of the fixing and the clearance to the panes | It is sealed inside |
| What happens if one unit fogs? | What is replaced, by whom, on what lead time | Do not worry, it is covered |
| What is the linear value for the grid? | A number, a unit and a method | There are no thermal bridges in aluminium of this quality |
| Are all the panels the same width? | A module table | It opens fully |
| Which side do the panels park, and what do they park against? | A side, a plan drawing, and what occupies that wall | It slides beautifully |
Not in the way the dictionary means it, and any supplier who says otherwise is selling you the photograph. A French door is a pair of leaves hinged at the jambs that meet in the middle with nothing permanent between them. Put those leaves in a track and four of the five defining features are gone. What we sell is a sliding door carrying a divided-lite pattern, and that is a perfectly good product to specify as long as it is specified by its own name.
Structurally yes, physically no. It adds linear metres of conductive material inside your insulating cavity, sometimes more than the Perimeter of the pane. A national research council associate it with reduced thermal resistance and a colder interior glass surface, and therefore reduced resistance to condensation. It is decoration with a measurable bill.
The honest answer is that it depends on which of three failures you would rather live with. Bars bonded to the faces give the best shadow line and let you ｒｅｐｌａｃｅ bars without replacing glass, at the cost of cleaning each rectangle. Bars sealed between the panes are the easiest to clean and the hardest to repair, and are at their most restless on large panels. True divided lite is the most honest and multiplies the number of sealed edges by the number of panes. Ask how the grille is retained before choosing, and ask what happens when the unit fogs.
No. We build sliding windows and doors, and our swing door, folding door and entry door category pages are empty at the time of writing. The thing the photograph remembers is a pair of wooden swing door leaves, and we would rather tell you that in the first paragraph than quote you for something we do not make. If what you actually need is the French elevation over a wide terrace opening with no swing arc, say that instead, and then specify it by grid construction, panel modules and threshold detail.
Everything about numbers with units does. The published sizes we quoted for French doors are largely North American retail conventions, and the U-value discussions above are largely drawn from United Kingdom regulatory material. Both travel badly. Carry the definition with the number, and ask your own authority which of these things it regulates at all.
Before the shop drawings are frozen, definitely. The pattern interacts with panel widths, with whether a bar lands on the interlock, with replacement codes and with whether the thermal number you have been given still describes the unit once the grid is added. Deciding it late does not only cost money; in some cases it changes which panel sizes are still available to you.
Projects of this kind tend to be glazed from floor to ceiling and seen at dusk from across a garden. If you want to see how the panels, the sightline matching and the finish actually come out on built work, our completed villa glazing packages in the Gulf and Australia are the most honest reference we can offer, because they show what happens to these decisions after installation rather than before it.
Do not take anything from this page into a drawing or an order unopened.
Threshold behaviour, drainage detail, thermal methodology and the date any of it takes effect are decided by your authority having jurisdiction. The regulatory material quoted here was published as United Kingdom guidance in one case and North American retail convention in another, and neither travels to your project without checking.
And to repeat the sentence we would most like you to remember: we are a sliding door and window factory. We do not make hinged French doors, we do not hold anyone's local product approval, and we cannot take responsibility for what the building does around the track.
