Jumbos and MD-80s, measured from the factory drawings
When we gave aircraft their real shape, every model came out of a family template. A twin-engine widebody looked like other twin-engine widebodies, scaled to its published span, length and tail height. For most types that works. For the rare ones people actually stop to look at, it shows. A 747 drawn from a generic four-engine template has its engines in the wrong places and a wing that sweeps too little. An MD-80 from the generic rear-engine template has a stubby tail cone.
Aircraft makers publish airport-planning documents for every type they build, so airports can plan gates and taxiways. Each one opens with a three-view drawing with dimensions printed on it. We went through those drawings for the rarest types in the sky and wrote the measurements into the table.
What changed. These types now use their own proportions instead of the family's: 747-100, -200, -300, -400 and -8; 777 Freighter; MD-11; MD-82, MD-83, MD-88 and MD-90; 737-500, 737-600, 737 MAX 7 and MAX 10. That covers fuselage width and depth, where the engines hang, wing sweep and dihedral, how the wing tapers, how high the tail sweeps up, and the fin's size and lean. Where a drawing prints a number we use it as printed. The rest is measured off the drawing at a known scale.
The 747-400 wingspan. The two public databases we use disagreed about the 747-400. One gave 59.65 m, which is the span of the older 747s without winglets. The other gave the right figure. Boeing's own drawing settles it at 64.44 m, so the model now has its full span and its winglets. When a value comes from the manufacturer's drawing, it overrides both databases. The 737-600 span is corrected the same way.
New types in the sky. The 747-100, the MD-90 and the 737 MAX 10 had no shape before, because neither database had a layout page for them. They now borrow the layout from a close sibling and get their dimensions from the drawing. Before this, the MD-90 was drawn as a plain narrowbody with engines under the wings.
What we didn't do. The drawings themselves stay with their publishers. We store numbers only, each with the document and revision it came from. Airbus types came later, see the ninth update. A couple of simplifications remain. Our winglets are a bit longer than the real ones, and the MD-80 tail cone ends at the stabilizer tips.
Update, the same evening: the common ones too. We went on to the types you see most often: the 737-700, -800 and -900 with their blended winglets, the 737 MAX 8 and MAX 9, and the 777-300ER. The 737NG sheets show a wing with a little less sweep than the MAX (27 degrees at the leading edge against 28), a taller fin with a shallower lean, and a fuselage that sits about 25 cm lower on its gear. The wing and tail read the same on all three NG sheets to within 10 cm, so the only thing that changes between a -700 and a -900 is where the fuselage plugs move the wing. The MAX 8 and 9 use the MAX 7's wing and tail, placed along the longer fuselage. The 777-300ER uses the 777-200LR's wing and tail, moved forward by the 5.3 m plug ahead of the wing. To check the method, we ran the same script on two sheets we had already measured by hand. It landed within 13 cm on the MAX 10 and within 30 cm on the 777-200LR.
Second update: the widebodies and the 757. Next came the 787-8, -9 and -10, the 777-200 and -300, the 767-200, -300 and -400ER, and the 757-200 and -300. Two of them got corrected sizes. One of the two databases listed the 777-300 at the length of the 777-200, almost ten metres short; Boeing's sheet prints 73.86 m. For the 757 and the 767-300, the databases give the span with the winglets many airlines retrofitted. Boeing's sheets show the wing as delivered, without them, and that is what the models now have. A 757 with winglets will look slightly short-winged until we add a variant for it.
We also corrected our own earlier numbers. The 777-200LR and -300ER fin was placed about 1.7 m too far aft, and their dihedral was a little too steep. The 777-200 sheet and the 777-200LR sheet now read the same with the method we use everywhere, and both models follow them.
One thing that cost us time: on some sheets the side view and the top view aren't drawn at exactly the same scale. On the 777-300 sheet the side view is about 1% larger, and on the 787-9 sheet about 1% smaller. Measuring the fin's position with the top view's scale put it 0.5 to 0.6 m off. We now check the scale of each view separately and measure the fin and the tail cone in metres from the end of the aircraft. Done that way, every variant of a family gives the same answer within about 20 cm.
Third update: the Embraer jets. Embraer publishes airport-planning manuals too, so the E190, the E175 and the newer E190-E2 and E195-E2 now use their own proportions. The E175 comes in two versions in the sky: the original wing with small winglets, 26.00 m across, and the later "enhanced wingtip" at 28.65 m. One of the databases had the long-wing E175 at 31 m, which is the span of the E175-E2, a different aircraft. The models now follow Embraer's drawings for both. The E2 jets have raked wingtips instead of winglets, and that is how they're drawn.
These drawings had a trap of their own. On the E190 and E175 sheets, the page is stretched: one metre along the fuselage takes more paper than one metre across it. Using a single scale made the E190 fuselage 2.7 m wide instead of 3.0, so we measured every view along both axes from its own printed dimension lines. The E170, the original E195 and the ERJ-145 family are still on the template, because we couldn't find a public manual for them.
Fourth update: the ATR turboprops. The ATR 42 and ATR 72 are the most common turboprops over Europe, and they now have their own shape too: a high wing that sits on the roof of the cabin, a T-tail, and propellers 8.10 m apart. ATR doesn't publish an airport-planning manual openly, so we used the fact sheets on its own website. They print the span, length, height, wing area and propeller spacing, and the views are drawn to scale, which was enough to measure the rest. The top view on those sheets is rotated by 45 degrees, so we turned it back before measuring. One detail fooled us for a minute: the black stripe along the front of the wing looked like a shadow, but it's the rubber de-icing boot, so the wing starts where the black starts.
This also fixed a real mistake. One of the two databases gave the ATR 72-500 the wingspan of the smaller ATR 42, 24.57 m instead of 27.05. Until now the model followed it, so every 72-500 in the sky flew with wings two and a half metres too short. ATR's sheet settles it. If ATR's full airport-planning document turns up, we'll check these numbers against it.
Fifth update: the CRJ family. The Bombardier CRJ200, CRJ700 and CRJ900 now use the drawings from Bombardier's airport-planning manuals, which are still on its customer site. All three keep the rear engines and T-tail, but the wings differ: the CRJ200 is swept 27.5 degrees, the CRJ700 and CRJ900 share a newer wing at 30 degrees, and the later CRJ900s carry larger winglets tilted 34 degrees outward. That shared wing was a useful check: we measured it on two separate manuals, and the root chord, sweep and fin came out within 2% of each other.
These pages needed more care than any so far. Each view is drawn at its own scale, and the little scale bar printed on the CRJ200 sheet matches none of them. On the CRJ900 sheet the printed heights don't fit the drawn outline at all, so we took the stance from the nose clearance that both the CRJ700 and CRJ900 manuals print, 1.73 m. The CRJ200's span is now 21.23 m with its winglets; one database had 20.94, which leaves them off.
Sixth update: the Dash 8. The De Havilland Canada Dash 8 Series 100, 200, 300 and 400 now use the drawings from their airport-planning manuals, which sit on the same Bombardier customer site as the CRJ ones. They keep the high wing, the T-tail and the long engine nacelles, and the Q400 now looks like the stretched airliner it is: 32.83 m long on a 28.42 m wing, with its propellers 8.80 m apart against 7.87 on the smaller ones.
The Series 100 manual is a 1993 scan on graph paper, so the 100 and 200 share the cleaner Series 200 sheet; they are the same airframe. Two of the sheets needed straightening first. The Q400 side view is drawn tilted nose-down by 1.35 degrees and its top view is turned by about 32 degrees, so we rotated both back before measuring. The Series 300 top view doesn't agree with itself along its length, and its tailplane is drawn 9% wider than the dimension printed next to it, so the 300 takes its fin from the Series 200 sheet, moved back by the 3.45 m stretch. All three manuals print a wing dihedral of 2.5 degrees, and that is the number the models use.
Seventh update: the 717 and the 727. Boeing still publishes the airport-planning documents for both, the 727 one as a 1978 scan. They have something the other jets on this page don't: a T-tail that sticks out past the end of the fuselage. The 717 is 37.80 m long to the tips of its tailplane, but its fuselage ends at 34.34; on the 727 it's 46.68 against 41.50. Until now our models ran the fuselage all the way to the overall length, so both had a tail cone several metres too long. The model now ends the fuselage where the drawing does and lets the tail overhang it.
The 727 now has its own fin, swept back 56 degrees above the centre engine intake, a 35-degree wing, and a span of 32.92 m, the wing as delivered. One of the databases lists 33.31, which is the span with the winglets some operators added later. The 1978 sheet is printed on graph paper, so we straightened the scan by a quarter of a degree and filtered the grid out before measuring.
Eighth update: the older jets and the winglets. The 727-100, two DC-9s (the short -15 and the -32), the MD-87 and the DC-10 now have their own proportions. The 727-100 is new to the table: one of the databases listed it with two engines, but every 727 has three, and the model now draws all three.
Most of the older Boeing and Douglas sheets are scans on graph paper. On the 727-100 and DC-9 pages the outline breaks up once the grid is filtered out, so we couldn't trace them the way we traced the rest. These aircraft share their wing and tail with a sister we had already measured from a clean sheet, though, and the drawings print the distances that differ: where the nose gear sits and how far it is from the main gear. The 727-100 is the 727-200 with about 3 metres less in front of the wing and 3 behind it, and the DC-9-32 is close to the 717. We moved the wing and the fin by those printed distances and wrote down that these four are worked out that way, not traced. The DC-10 sheet was clean enough to measure directly, and its wing lines up with the MD-11 drawing.
We also fixed two things that every model shared. Winglets were drawn at a fixed share of the span, so on some jets they were twice the real size. The split winglets of the 737 MAX had no lower half. Both now follow the drawings. The centre engine on the MD-11 and DC-10 sits higher on the fin, where it really is.
Ninth update: Airbus. Airbus publishes its airport-planning documents only to a browser, so for a while we thought we'd have to wait for someone to download them by hand. Then we found copies. Archived copies of the official "Aircraft Characteristics" sit on archive.org, saved by perma.cc with the original airbus.com address and the date of the copy, and SKYbrary keeps the older ones on its bookshelf. Twenty-one Airbus types now use their own proportions: the A319, A320 and A321 with their neo versions, the four A330s, all four A340s, the A350-900 and -1000, the A380, and the three older widebodies, the A300B4, the A300-600 and the A310.
The newer Airbus sheets are drawn as filled silhouettes with highlights rather than outlines, so we taught the measuring script to read a solid shape. The A300-600 and A310 sheets are 1994 scans with no text, and both are slightly skewed. On the A310 the whole page is turned by about half a degree, so we rotated it back. On the A300-600 only the vertical lines lean, which is a scanner shear, so we straightened the columns instead. On the older sheets the flap fairings and slat lines confuse an automatic fit of the wing's trailing edge, and there we picked it by hand and wrote down which line we used.
One dispute is closed. The two databases gave the A330-300 heights almost three metres apart, 13.93 and 16.85 m, and Airbus's drawing settles it at 16.94 m. For the height we use Airbus's clearance table at maximum taxi weight, which is the tallest the aircraft stands; the outline on the page is drawn about half a metre lower.
Two things are left as they are drawn. The A320 family is shown with sharklets, though older aircraft carry small wingtip fences instead, and the A300-600 and A310 sheets show plain wingtips, though many of them fly with fences. The callsign data doesn't tell us which one is overhead. The A220 is the one Airbus type still on the template, because we haven't found a copy of its document anywhere outside Airbus's own site.
If a type you spot looks off, tell us in Discord. We'll check it against the drawing.