Start where somebody actually has to decide. Not the paperwork. The arrival.
A bill of exchange comes into a London counting house. Translation, if you need one: a written order to pay a stated sum, a wire transfer with a short-term loan bolted onto it, and the counting house is the back office where merchants kept the books and handed over real money. The bill came from Limerick. Or Newcastle, or Hamburg. Nobody standing in that room has ever seen the hand that signed it.
They get one decision. Money against paper, or no.
And everything upstream of that moment is useless to them. The partnership deeds, the sealed power of attorney locked in a drawer three hundred miles away, the private letter of instruction that says don't accept anything above five hundred pounds. All of that exists somewhere. It isn't in the room. What's in the room is one sheet and one signature, and whatever that signature can't carry by itself does not get carried.
Sometimes it carried two extra letters.
Per procuration, through the agency of another, abbreviated by a working hand to "per pro." or "p.p." It made one claim, and a narrow one at that: this is not the principal's own hand, and this hand signs under authority granted by somebody else. Nothing in that impugns the paper. What it told you was that a boundary on this authority existed somewhere, and that locating it had just become your job.
Nobody sat down and designed that. It accreted, the way every convention worth having accretes, and the courts showed up afterward to explain what it had already been doing for years. In Attwood v. Munnings (1827), London bankers took a £1,560 bill accepted in a merchant's name "by procuration" while the merchant was off abroad. The authority he'd actually left behind didn't stretch that far. The bankers ate it, and Holroyd J. dropped the loss exactly where the mark had already put it:
"gave due notice to the plaintiffs, and they were bound to ascertain, before they took the bill," whether the acceptance matched the authority actually granted.
Before they took the bill. The whole mechanism lives in those five words.
I've spent a career trying to get useful signal into the place where the decision actually happens, which is why the economics here still delight me. Emitting the mark cost nothing: a few extra strokes inside a signature block somebody was writing anyway. Ignoring it cost £1,560, because it took a judgment you could make on appearances and converted it into a judgment that required asking a question.
It specified almost nothing, which is the other half of why it held up. The hand on the pen might belong to a bank's manager, or a merchant's brother minding the shop while he travelled, or an attorney acting under a sealed instrument. The mark didn't sort them. It said the hand wasn't the principal's, and that was the entire payload. Somebody holding that sheet up to a bad window, running it against a ledger, learned nothing about the identity behind the pen and quite a lot about the class of question he now owed himself.
Traders read the field closely enough that grades developed. In Ulster Bank v. Synnott (1871) the court laid out a working trade distinction: "p.," "pro," and "for" suggested general agency, while "per pro." and "p.p." pointed at authority under a formal procuration or power of attorney. Nobody develops grades for a field nobody parses.
There was a field order, too. A bill litigated in Alexander v. Mackenzie (1848) carried the indorsement "Per proc. Newcastle-upon-Tyne Joint-Stock Banking Company, H. Bleckley, Manager." Warning first. Principal second. The individual hand and his office last.
It was never a fence, and pretending otherwise misreads it. When a "per proc" signature on an 1857 charter-party produced Smith v. M'Guire, the principal argued his agent had blown straight through private instructions, and he lost anyway, because the agent genuinely ran the business. So the signal enforced nothing. It routed. Ask, and whatever you turn up governs. Which likely shaped how a careful reader handled unfamiliar paper a good deal more than it shaped any particular outcome.
Maybe that's how it survived long enough to get written down. When Parliament codified, section 25 of the Bills of Exchange Act 1882 said a signature by procuration "operates as notice that the agent has but a limited authority to sign." The Act's draftsman described his method as reproducing existing law "whether it seemed good, bad, or indifferent in its effects." He found a habit already bearing load and declined to move it.
How often anyone actually went and checked, we mostly can't know. One thing about it isn't in doubt, though. The warning travelled on the instrument. Not in the archive.