Font Forensics

No matter where or how you write, a bit of you stays behind.

The phone suddenly began beeping with SMSes. Your account has been debited by ten thousand rupees, the first one said. It was followed by another. And another. And another. Finally, there came a different message from the bank: Your credit-card has been blocked due to undue amount of transactions.

The credit-card number, it seemed, had been leaked, and was being used by some online agent to buy ₹50,000 worth of ads on Facebook.

In India, as in most countries round the world, a credit-card number is not enough to make a payment. You also need a PIN, or Personal Identification Number — a secret four-digit code that you punch in to complete your transaction. Online transactions also need a “one-time password” that’s sent to your mobile number to prove it’s really you.

The credit-card PIN is common in most places, but there is one country in particular where it is rarely found. And that country is the USA.

In the past few years, credit-cards have been switching to the new “chip and PIN” technology. Each card has a special electronic chip that’s hard to copy, and needs a PIN to complete the transaction.

In the US, the situation is different. Most cards don’t have PINs, because Americans don’t want “yet another password” to remember. Besides, they have a more convenient alternative: chip and signature.

Long before the advent of modern banking, people were already using signatures to prove their identity. Emperors and noblemen may have special seals to stamp on their documents, but ordinary people simply wrote their names.

This might seem a bit counter-intuitive. If everybody knows your name, surely anybody can copy it and pretend to be you? But no. A signature isn’t about what’s written; it’s about the way it’s written. Everyone has their own style, and, if you tried copying someone else’s, your uncertainty would show in the pressure of your pen and the thickness of your strokes.

Forging a signature is not easy — in fact, it’s easier these days to forge someone’s email.

The difficulty of forging a signature is a special case of a more general rule: it’s hard to forge any handwriting of another person. And, everyone’s handwriting is unique.

How do you write? Do you press your pen down hard on the paper, or let it glide lightly past? How long are the tails of your ‘a’s, the curls of your ‘y’s, and the round loop on top of your ‘o’? Do you cross your ‘t’s to the left, or to the right, or right in the middle, or not at all?

These are only some of the more obvious ways your handwriting can vary: there are so many details to differ in, it’s practically impossible for all of them to line up in two different people.

Handwriting is not just unique; it’s also hard to disguise. Even if you try to change your handwriting, some of your true style shows through. To get an idea of just how deep it runs, researchers looked at people whose hands had been paralysed. They had learnt to write with their toes or mouths — and, with practice, their ‘new’ writing style ended up the same as the original.

Not surprisingly, handwriting analysis has become a useful part of the police toolkit.

On 4 July 1956, Mrs. Beatrice Weinberger, who had left young Peter outside for a few minutes, was shocked to find the baby gone and a green-inked ransom note in his place.

A worried Weinberger called the police, though the kidnapper had told her not to, and left a dummy parcel of shredded newspaper, although the kidnapper had asked for $2000. The kidnapper never showed up, although he said he would, and, finally, the FBI was called in.

Without delay, a manhunt was launched — not across the streets of New York, but through two million papers, records, and hand-filled forms in government bureaus. The handwriting experts had studied the ransom-note, and were trying to find a match.

Finally, the culprit was identified: a certain Angelo La Marca, whose style, especially a distinctive stroke over the ‘y’, matched that of the kidnappers. When Angelo was shown the evidence, he broke down and confessed. The baby had been left in a “safe place” near the park, he said, but when the police rushed to the spot, all they found was Peter’s lifeless body. Angelo was tried for murder and sentenced to death.

Not all cases are quite so dramatic: handwriting analysis is usually just one piece of evidence, to prove that someone wrote something, or that someone else didn’t.

But the threat of identification meant that kidnappers were probably happy when a new technology came along: one that, at least in the beginning, let them write truly anonymous ransom notes.

The trial for Angelo La Marca was short and swift, but most court cases tend to drag on — and someone needs to be there, taking notes all the time.

That’s why court reporter James O. Clephane was very enthusiastic when he heard of a device that could help him write notes faster, as well as make multiple copies. Inventor Christopher Sholes wanted him to try out his latest models of typewriter, and give him feedback.

And try it he did. Clephane tried those typewriters as they had never been tried before; he tested them so hard that they broke as fast as new ones could be assembled for testing. Clephane’s feedback was ruthless, and he always listed many faults within the latest design.

At one point, so the story goes, Sholes was fed up with Clephane and wanted to stop sending him pieces for review. But his business associate Densmore replied that “This candid fault-finding is just what we need. We had better have it now than after manufacturing”. Let us fix every lever and rod, he said, and then “depend upon Clephane for all the praise we deserve”.

Sholes’ typewriter was by no means the first one invented, but it was one of the first to be popular and mass-producible. Of course, with the rise of typewriters came the rise of typewriter forgery. And then, the rise of typewriter forensics.

Not all typewriters are unique, but some are.

Put simply, typewriters work like this: you press a button on the keyboard, activating a lever connected to a metal stamp block of the corresponding letter. This stamp block, or “typeface element”, bangs down onto an inked ribbon, which leaves its impression on the paper.

Now, some typewriters have typeface elements permanently attached. You can’t change them around. That’s great news for detectives, because you can look out for any chips or scratches or worn-our letters that serve as tell-tale marks for that particular machine. You can’t tell who typed the document, but you can at least tell whose typewriter was used.

Other typewriters have changeable typefaces, which makes them harder to identify. There are other methods, though, including looking at the inked ribbon to see impressions of what’s been typed before, or — though slightly dubious — analysing how hard letters were pressed to see if it was typed by a particular person or someone with different typing habits.

But the world has moved on; gone are the days when typewriters were commonplace. Documents are now written and forged as computer printouts, and even digital files, with their author metadata, don’t tell you much. All you can conclude is that a certain person writes all sorts of documents in all sorts of places, and goes by the name of “Admin”.

So it might come as a surprise that, when Prime Minister Nawaz Sharif was convicted of corruption, one of the key witnesses was Microsoft Word.

Digital documents are a bit like typewriters, using sequences of bytes instead of ink from ribbons. But they have many letters that typewriters don’t: the Unicode standard defines letters for 146 different scripts, not to mention extras like punctuation and emoji.

And then, there are the zero-width characters.

Zero-width characters are letters that can’t be seen on the screen, like “zero width non joiner”, which are there for technical reasons. But they can also be used to catch whistleblowers.

Some years ago, Tom Ross was part of a videogame tournament team, which had a private message-board for discussions and appointments. Suddenly, the announcements started appearing in other places on the web, posted to mock the team or, worse still, reveal their secret tactics. The message-board website was pretty secure, so the messages were probably being leaked by one of the logged-in users.

It was an inside job.

So, Tom made a script to modify the messages each user saw. Hidden in the text was a unique set of zero-width characters, which the website changed depending on which user was loading it. When the messages were copy-pasted to other sites, the hidden zero-width characters went along too — allowing the culprit to be identified.

This is a neat trick, but journalists and whistleblowers beware: zero-width characters can be used on you as well.

The document that convicted the Prime Minister didn’t have zero-width characters. It had something much more elementary: a font.

You might not realise it, but fonts never stay the same. They keep getting minor updates over time— the way two letters join, a tweak in the curve of an ‘a’, or something even more subtle.

Type designer Thomas Phinney didn’t plan to be in the detective business. But as part of Adobe’s type design team, he was called in by a lawyer to check the validity of a will. Since then, he often spends his time poring over sheets of paper, armed with a magnifying-glass and a pair of calipers, painstakingly analysing to find out which version of what font was used where.

Phinney gives his cases Sherlock Holmes-esque titles: The Dastardly Divorce, The Respected Rabbi, The Presidential Plot. Each turned out with a different solution: use of a “typewriter” font instead of a real typewriter, or a document printed with an inkjet printer that wasn’t available at that time.

In the case of Prime Minister Nawaz Sharif, the accusation was of collecting money illegally, and then sending it overseas to avoid paying tax. Tax-evasion isn’t always illegal — super-rich people hire “wealth managers” to use the optimal loopholes for them — but in this case, it was.

During the trial, the Prime Minister’s daughter, Maryam, produced a document. It was an ordinary one — written using Calibri, the default font of Microsoft word — and the writing inside proved them innocent. Or so it seemed.

The Calibri font was released in November 2006. Maryam’s document was dated February.

Instead of collecting and hiding money, some people just steal it when they need it. That’s what happened in the case at the beginning of this article, where ₹50,000 was spent off a credit-card to buy ads on Facebook.

When my friend— to whom the card belonged — phoned the bank, the receptionist seemed to have heard it all before. “We’ll give you a temporary refund”, she said: Facebook would get a 90-day grace period to “prove the transaction was genuine”, failing which the refund would be made permanent.

Some time later, the Reserve Bank of India blocked all international credit-card payments, unless the users specifically asked for the feature. Apparently, it’s not an uncommon problem.

It seems that, if you enter your credit-card number online, some websites allow you to complete the payment without PINs, even if a PIN is mandatory for transactions at home. All it takes is a leak of your credit-card number from one website, and there it goes.

The banks seem to be ready for it, though. Their algorithms monitor every transaction, ready to step in if something seems amiss. They’re adapting to a world without signatures, or typewriters, or passwords.

A world where you can authorise a payment by writing nothing at all.

Have something to say? At Snipette, we encourage questions, comments, corrections and clarifications — even if they are something that can be easily Googled! You can also sign up for our weekly email updates.

Want to know more? Sources and references for this article can be found here.