LEGO’s letter to parents, and how not to tell a fake when you don’t see one

According to the website of the Independent newspaper, LEGO UK has verified the 1970s ‘letter to parents’ that was widely tweeted last weekend and almost as widely dismissed as fake. Business as usual in the Twittersphere — but there are some lessons here about dating type.

‘The urge to create is equally strong in all children. Boys and girls.’ It’s a sentiment from the 1970s that’s never been more relevant. Or was it?

Those of us who produce or handle documents for a living will often glance at an example and have an immediate opinion on whether it’s real or fake. That first instinct is worth holding on to, because it comes from the brain’s evolved ability to reach a quick conclusion from a whole bunch of subtle clues before your conscious awareness catches up. It’s OK to be inside the nearest cave getting your breath back when you start asking yourself what kind of snake.

But sometimes you will flinch at shadows. Why did this document strike us as wrong when it wasn’t?

First, because the type is badly set in exactly the way early consumer DTP apps, and word processor apps to this day (notably Microsoft Word), set type badly — at least without the intervention of skilled users. I started typesetting on an Atari ST, the poor man’s Mac, in 1987. The first desktop publishing program for that platform was newly released, running under Digital Research’s GEM operating system. It came with a version of Times New Roman, and almost nothing else. Me and badly set Times have history.

In the LEGO document, the kerning of the headline is lumpy and the word spacing excessive. The ‘T’ seems out of alignment with the left margin, even after allowing for a lack of optical adjustment. The paragraph indent on the body text has been applied from the start, contrary to modern British typesetting practice; the first line should be full-out. The leading (vertical space between lines of text) is not quite enough for comfort, more appropriate to a dense newspaper column than this short blurb.

There’s also an error in the copy: ‘dolls houses’ needs an apostrophe. Either before or after the last letter of ‘dolls’ would be fine, depending on whether you think you mean a house for a doll or a house for dolls. But it definitely needs to be possessive.

It wasn’t just that the type looked careless. It was that it stank of the careless use of tools that shouldn’t have been available to its creators.

In 1974 (the date, if you squint hard, at the foot of the LEGO text), one did not simply walk into typesetting. There were no personal computers with software that let you enter text on screen, style it using proportionally spaced fonts and then print the result. Setting type was a clunky specialist job requiring dedicated equipment costing hundreds of thousands of pounds.

This was the age of hot metal, and larger presses were using Linotype machines or similar. Hell’s Digiset – the first system featuring digitised fonts and a CRT screen – had been around for just a few years (and wouldn’t last many more). The relatively affordable Compugraphic machines had yet to appear. Letterpress was still commonly used for small jobs.

The Apple Mac was a decade away.

Typesetting a booklet such as this was not part of its designer’s skill set, but a separate task entrusted to a trained craftsperson. Copy would be checked by an editor before being supplied marked up with notes on typeface, point size and layout. Few design studios had their own typesetting machines or operators (though the giant LEGO might be an exception); the task would be contracted out. Galleys would be returned with corrections.

No matter how lazy the designer or how tolerant the client, it was impossible merely to press Command+N, type some crap in the first style that came to hand, whack up the headline a few sizes and hit Print.

Yet the LEGO blurb looked as if someone had done exactly that, and using the first style that would come to hand with bog-standard desktop publishing or word processing software — once it was invented.

Note here that Aegir (like me, an experienced designer for whom digital typesetting is part of the job) describes the document as a ‘letter’. Looking at the first tweet in my timeline that linked the blurry photo, on my iPhone, I failed to register the fact that I wasn’t seeing an A4 sheet, but a square page with a spine at the right held by a staple.

This, when appreciated, made it more likely that the document was real. Knocking up a fake branded letter in Word to make some sort of point about sexism and children’s play over the past 40 years is one thing; cropping and binding it for realism would be taking it quite a few steps further.

But did the reprographics feel quite right?

I agreed with Aegir that the rub-off looked like laser (although others thought the text looked like inkjet). Laser printers melt toner, a fine plastic dust, onto the surface of the paper; when the paper is creased, it cracks off like oil paint, leaving a white gap. Ink, whether applied by an inkjet print head or by a web offset press, soaks into the paper; creasing the printed area won’t immediately produce a gap in the print.

From the photo, it was impossible to come to clear judgements on such fine details. Which was rather convenient, wasn’t it?

Talking of that logo, everyone knows companies change their branding over time — sometimes dramatically, other times just a tweak. So it ought to be possible to date the exact variant of the LEGO logo on the document. I turned to the relevant entry in Brickipedia, an open repository of LEGO information.

The logo on the left, Brickipedia informs us, dates from 1973; the one on the right from 1998. Still on the iPhone screen, I reckoned the variant in the photo, with clearly defined glyphs and a strong yellow stroke, looked more like the 1990s cut, still in use today — another indication that the document was more recent than it purported to be.

Here I was just flat wrong, as I would realise when I got back to my Mac. As @sirtwist pointed out:

The unfilled counter of the ‘O’ (along with the similar treatment of the gap between ‘L’ and ‘E’) is the killer. This is definitely the 1973 cut. I’d set too much store by the weight of the yellow outline. It is heavier in the photographed document than in Brickipedia’s image. But this was the 1970s. Logo masters existed as ink drawings on desk-sized sheets of film, not Adobe Illustrator files. Reproduction was optical, mechanical, fallible.

And, just as metal fonts were once cut differently for larger and smaller point sizes, logos would be adjusted for scale. This inch-square copy would naturally get a thicker stroke, for clarity. Not so careless.

By now, tweets questioning the provenance of ‘LEGO’s letter to parents’ were proliferating. Even without close analysis, it just seemed too good to be true.

Then the plot thickened.

A different version of what appeared to be the same document was also extant. Written in German, it was similarly typeset, and the text, although not a straight translation, covered the same points. A better photo was available, clearly showing that it was the back page of a saddle-stitched booklet.

Note the show-through: the headline on the previous page can be made out running backwards across the top (‘Am Tag, als…’). This looks right for the kind of thin paper the brochure appears to be printed on, judging from the creasing around the staple. It all fitted.

Then again:

This kind of difference in artefacting can often be useful in identifying composite images. But it’s rarely cut-and-dried, and in this case I was sceptical about George’s sharp-eyed point, for two reasons.

First, if the foreground and background type didn’t match, what did that imply about the way the image had been constructed? The show-through looked right: big type at the top left of the page (on the right here); smaller type at the bottom right; a dark shape, evidently an image of something, obscuring other parts; uneven light and texture across it all… Putting this together convincingly would have taken a lot of work. Why bother?

If the show-through was genuine, on the other hand, the basis for the fake must be a real photo of the back of a German booklet. So what was originally printed on its outside back cover? There was no sign of anything having been removed, and given the mottled texture of the paper and the show-through covering almost the entire area, to erase anything invisibly would, again, have required a good deal of skilled work.

High-volume leaflets are rarely designed with blank pages. Nobody wants to pay hundreds of thousands of times to convey nothing.

Second, it’s a fallacy that artefacting should be evenly distributed. What George broadly referred to as ‘pixellation’ — the making visible of the colour blocks making up an image — is recognisable here as the kind of compression artefacting produced by JPEG algorithms.

Most photographic images (as opposed to solid-colour line work) on the web are JPEGs. Their file size is dramatically reduced by throwing away information — particularly information conveying colour, rather than lightness — where the human eye is unlikely to notice the difference.

It’s a clever software trick, and mostly invisible. But one of its tell-tale signs is speckling around high-contrast edges. That’s what we’re seeing here on the black lettering. And I’ve just answered the question of why the same effect might be absent from the lettering showing through faintly from the other side of the paper: lower contrast.

This is worth a brief digression into Photoshop. If you’re viewing this on mobile, I can only apologise for Medium’s lack of zoom support. On desktop, embiggen your window now.

Above, I’ve imported a crop of the German booklet photo. The artefacting around the lettering is reasonably clear. To make it clearer, I add a Levels Adjustment Layer to expand the range of tones in the light areas:

As proposed, much more speckling is evident around the dark foreground lettering than the faint reversed type on the back of the page. Having established this, I’ll hide that Adjustment Layer for now and reduce the opacity of the photo so I can work over it.

I’ve added a Type Layer and entered the same words. I could use any shapes filled with black to test the pixel theory, but reproducing the type will be an interesting and perhaps informative exercise in itself. As expected, Times New Roman Bold is a good fit. A bit of manual kerning aligns the glyphs pretty much exactly with the original, seen in grey behind.

After adding a second Type Layer matching the visible part of the text from the reverse of the page, flipping it horizontally, adjusting the kerning and reducing the opacity, I have a pretty decent facsimile of the original crop:

I’m not finished, but I almost have my answer. Already, it’s obvious that the pixellation of the type, at the same low resolution as the original image, is far more noticeable in the high-contrast foreground text.

Don’t be led astray here: I’m not demonstrating that this effect is seen in type that I’ve created on a computer, and therefore the type in the image must have been created on a computer. On the contrary: I’m finding that pixel artefacts in any high-contrast area will be more visible than in any low-contrast area, in any image, of any provenance. So the effect seen in the photo is not a telling anomaly.

To confirm, I tweak the tone of both type layers (using the extra Levels Adjustment Layers listed in the Layers panel above) to roughly match the original, comparing the colour values via the Info panel, and also use the Eyedropper tool to sample the paper colour — well, one middling colour from the many that it contains — and fill the Background layer with it. Then I save the image as a JPEG, selecting a medium to low quality setting.

Opening this JPEG file and, for convenience, pasting it into the same Photoshop document, I get this:

Look closely, and you should recognise those sparkles around the characters. Applying the same Levels Adjustment Layer as I used on the original photo reveals this:

It’s not a precise match, but clearly the same kind of pattern as before. Around the high-contrast lettering, distinct pixel noise. At the top of the image, the lower-contrast type shows different and subtler blocky artefacts.

The German LEGO photo isn’t a composite. That’s just how JPEGs are.

Let’s wander back a moment to where we laid Times New Roman over the original German headline type. Did you notice that the ‘f’ didn’t match? In Times New Roman, its ascender continues straight up, curling around at the very top to form the ball, which sits entirely to the right of the vertical stroke, the curve of its left edge barely overhanging the bar below.

In the original ‘f’ (on the left, below), the ascender recoils to make way for a ball that sits over the bar, reducing the glyph’s overall width.

There’s also something slightly different going on with the ‘k’. And although it can be made to align, the original font is a little heavier and more compressed. These variations aren’t surprising: Times New Roman was first released in 1931, and numerous cuts have been sold since then. Adobe and Monotype each have their own official versions, and there are many clones, partly because typefaces are poorly protected by copyright in the US.

But that ‘f’ is downright odd. For some reason it made me think of Berthold, and sure enough there it is, in Arbiter, Mikaway, Simone, Tyfa… It’s by no means unique to that foundry, of course. The recoiling vertical is seen in a number of italics, and cutouts making way for balls are commonplace in super-heavy weights; but in a basic bold like this it’s unusual.

In cuts of Times, still more unusual: the only other example I can find anywhere is in Monotype’s Small Text variant, which I bet nobody’s touched since the heyday of newspaper classifieds.

What does this prove? Well, whoever typeset this was using a weird cut of TNR that doesn’t seem to match those currently available digitally. That could support the document being real, or it could just mean whoever faked it was using a free knockoff font from some dodgy website. We’re no further forward, but if you’re still reading this post I’m going to assume you’re as fond of little rabbit holes like this as I am, so there’s no need to apologise for wasting that last 40 seconds of your time.

While we’re talking typefaces, let’s take a closer look at the body text of the German blurb. It looks quite different from the English version. (The quality of the photo happens to be better, but let’s try to ignore that.) The spacing is similar: normal to slightly loose tracking (horizontal), normal to slightly wide word spacing, slightly tight leading (vertical). Again, it hasn’t been hand-kerned — look at the ‘Ta’ in ‘Tatendrang’ (a lovely word that translates as ‘zest for action’).

But it looks nicer, doesn’t it? That’s mainly because it’s not set in ugly Times New Roman. What exactly it is set in, I’m not quite sure: it’s another transitional serif, but feels closer to its late Renaissance roots. Note the tapering ‘J’, descending below the baseline. It might remind Adobe users of Robert Slimbach’s Minion, which wasn’t released until 1990 — but don’t get excited, it’s not a match. If I’m missing an obvious identification, someone comment and tell me so I can kick myself.

Anyway, the two documents, though similarly laid out at a glance, are differently typeset. And remember this is at a time (if we believe the date) when a ‘font’ is either an actual box of metal type, or a set of matrices for hot metal, or a newfangled Digiset template — all eye-wateringly expensive. This isn’t just two designers on their respective Macs picking different fonts from the list and thinking they look near enough the same.

Both LEGO documents say they were ‘Printed in Germany by Mühlmeister & Johler, Hamburg’. Founded in 1876, this firm no longer appears to be trading, but was a big deal in its time. You might guess that if LEGO was printing everything for the European market in Germany, that was because its print design function was headquartered there, and one team would be producing localised versions of each document for the various territories where LEGO products were sold.

So why would it typeset two versions of the same document using different fonts?

And why, in 1974, would a document be labelled ‘Printed in Germany’, when there was no such country? It had been divided since 1945 into West Germany (the Federal Republic of Germany), where LEGO no doubt operated, and East Germany (the German Democratic Republic).

At this point my gut feeling, based purely on the typography, was that the English document was fake and the German document was real. The former, though it looked more convincing as a physical object when I had chance to look at the photo again on a desktop screen, was just too much like a lazy DTP effort. The latter was a plausible product of pre-digital commercial typesetting.

Was one, then, a later copy of the other? This seemed conceivable. The English version couldn’t be a later localisation by LEGO of the German, since it was dated ‘1974’; as blurry as the picture was, there was no mistaking that for any other number. And if we were right about the laser printing, it was likely to be a one-off mockup. High-volume laser printers are used for short-run digital printing these days, up to a few thousand copies, but not for jobs like a leaflet boxed with a mass-market toy.

Someone had to have found the German document and thought it was too good not to share, but realised it wouldn’t go viral in a language unknown to 94% of the internet. So they’d recreated it in English, massaging the text to make its point about gender stereotyping more clearly and prune the marketing-y repeated references to LEGO. It made a degree of sense. Fake letters go viral all the time.

But the footer was the same on both documents, so it had to be explained. I had a feeling that ‘Printed in Germany’ wasn’t wrong, and had been the standard form even during the country’s partition. Since I’d studied German as a child and had been to the country a few times, the feeling might have some basis in fact, even if I couldn’t recall where.

Finding a reference to confirm this proved tricky: I’m mercifully unfamiliar with the databases of historic European trade regulations that might have provided a definitive guide, and the web is not jam-packed with fully annotated examples of post-war German printed ephemera.

Still, I eventually got lucky. At, Masanori Yokono (no relation — what did you think?) maintains a catalogue of Beatles record releases. Inside the picture sleeve of 1967’s ‘Magical Mystery Tour’ Double EP, we find the legend:

Germany. No ‘West’ or ‘Federal Republic’. It’s just one example, but a prominent one, and at the height of the Cold War. So ‘Printed in Germany’, in 1974, wasn’t an anachronism after all. There was no reason to believe the German document was wrong.

As it turned out, nor was the English one. The Indy, without any forensic investigation at all, eventually established this by the frankly cheaty method of phoning LEGO and asking.

There were several good arguments for questioning the photo of the ‘LEGO letter to parents’. They just all happened to be mistaken.

The text itself seemed a bit off. I had LEGO in the 1970s, but don’t remember seeing this or anything like it. As marketing blurb, it now seems oddly defensive and curiously drab, but laboured long-form copy was an accepted trope of the period; people had more time to kill back then:

People also had time to set type properly, as this illustrates. (Check out that hand-kerned head, set TNT — ‘tight not touching’ — in the characteristic advertising style.) But LEGO’s composers, or Mühlmeister & Johler’s as the case may be, evidently had an off day. Maybe the machine with the nice type was being serviced. Or the UK arm commissioned differently from the Germans. Someone should track down the original employees and find out. Myself, I think I may have spent slightly too long on this already.

Even with hindsight, I think it would be impossible for anyone who’s paid attention to type for the last few decades to look at the LEGO letter and not hear alarm bells. But asking ourselves why mediocre mechanical typesetting in 1974 should look mysteriously similar to mediocre digital typesetting in 1994 is getting the question backwards.

Desktop publishing software was designed to produce something resembling contemporary bog-standard typography. That was its model. We may now associate the ubiquity of Times New Roman with products like Microsoft Word, but it dates back much further. It was because it was already in widespread commercial use that it was included as a ‘professional’ feature of the new systems. Now you could set type that looked just like all the other type that had been set.

I wonder if the German origins of both documents might also explain the two other problems with the English type: the first-line indent and the missing apostrophe. If the German-language copy came first, and was then swiftly localised for the English market, the formatting might have been left as-is, ignoring the niceties of typography and grammar.

‘Put the right material in their hands’ sounds a little stilted. It’s a direct translation from the German.

Whoever wrote the English copy, anyway, that person may have had no involvement after submitting their typewritten text. I’ve seen otherwise competent English-language documents with German punctuation („like so“) and vice versa, a comparable error to a missed apostrophe or non-standard indenting. Such production errors will happen.

The absence of sourcing beyond a single low-resolution photo and unevidenced description meant fakery couldn’t be ruled out, but didn’t prove it. When reasonable doubts were raised about the reprographics — the cracked fold, the below-offset-quality text — they couldn’t easily be dismissed, but nor were they substantiated.

A white line can appear through ink-based print if a crease is worked to the point where the paper fibres fray. And the thick inks used in commercial printing can give a shiny finish similar to laser toner, depending on the substrate. The kind of cheap semi-gloss paper often used for in-box leaflets could exacerbate both of these effects.

As for the crispness of the type, or lack thereof: judging from a low-resolution image is pushing it, but I’d guess this job was run on a gravure press, not offset litho. This process, also used for catalogues, newspaper colour supplements and weekly magazines, gives a dirtier line quality.

Prop-makers will tell you these high-print-run publications are among the hardest to mimic convincingly, because the methods used to produce them are prohibitively expensive for one-offs. Ironically, if someone did want to fake their own ‘LEGO letter’, it would be no small feat to get it right.