The Bureau of Engraving and Printing builds three levels of security feature into a note: covert features, features for banknote equipment manufacturers, and public features. Feel the paper, tilt the note, hold it to the light: that campaign is the third level. The note acceptor on the self-checkout was built by a company that gets the second.
Start with the word the instructions keep using. Three of the features on the current $100 begin with tilt. A validator cannot tilt anything. It grips the note between rollers and pulls it in a straight line past fixed sensors.
The whole public check fits under four headings in the Currency Education Program’s guide, and the summary line is one sentence long: feel the paper, tilt the note, and check with light and with magnification. Three of the four are optical and one is tactile. Every one assumes a reader who can hold the note and change their mind about the angle.
“Tilt the note back and forth while focusing on the blue ribbon. You will see the bells change to 100s as they move.” The ribbon is woven into the paper, not printed on it.
“Tilt the note to see the color-shifting bell in the copper inkwell change from copper to green, an effect which makes the bell seem to appear and disappear within the inkwell.”
The numeral 100 in the lower right corner of the front “shift from copper to green.” Present on the $10 and above.
“Hold the note to light and look for a faint image of Benjamin Franklin in the blank space to the right of the portrait. The image is visible from both sides of the note.”
Embedded vertically, “imprinted with the letters USA and the numeral 100 in an alternating pattern.” It “glows pink when illuminated by ultraviolet light.”
Run a finger down Franklin's shoulder. “It should feel rough to the touch, a result of the enhanced intaglio printing process used to create the image.”
Quotations above come from the $100 note feature sheet and the Dollars in Detail guide. The sheet adds microprinting and the red and blue fibres worked into the paper. The BEP puts the paper itself at 75 percent cotton and 25 percent linen.
The security thread is embedded in a different position for each denomination and glows a different color when held to ultraviolet (UV) light. That is a published feature written in a language a machine already speaks: position and fluorescence, both of which a fixed sensor can read at speed. The $1 and the $2 have no thread.
The published feature sheet for the $1 runs to raised printing, the paper and its red and blue fibres, two seals, the portrait and the serial numbers. No security thread. No watermark. No colour-shifting ink. No microprinting. And it is not going to change, for a reason that has nothing to do with counterfeiting technique.
“Because the $1 note is infrequently counterfeited, the government has no plans to redesign this note. In addition, there is a recurring provision in the annual Financial Services and General Government Appropriations Act that prohibits the redesign of the $1 note.”
The BEP names this level in the same sentence as the other two and then stops. What it does publish is the route by which manufacturers get it. Approved firms receive non-monetised sample notes through the Test Deck Program, and the BEP runs a device lab where a manufacturer’s own equipment is fed BEP samples and, sometimes, counterfeits. Who qualifies is drawn broadly: companies building equipment for “point of sale, counting and sorting, transit systems, vending, gaming, and note validating and dispensing in commercial environments.”
What that equipment does is documented, just not by the issuer. It sits in the patent files of the companies that build it. Read a few and the same short list of physical measurements keeps coming back, and no two designs are quite alike. The ECB defines a machine type as one that can be told apart from another by its specific detector systems, software and other components. Treat everything below as one firm’s engineering rather than a standard.
Magnetism is nowhere on the public list. It could not be. There is nothing a person can do with it, and in the design below it is the first question the machine asks, from a head sitting under the paper path before the note ever reaches the optics.
“Currency printers typically attempt to deter counterfeiting by giving the currency predefined magnetic signatures. The magnetic signature can be realized by using ink or dyes which have magnetic properties, for example, the ink can contain magnetized particles which produce a magnetic flux.”
That patent’s design does not measure the field the note gives off. It sets up a reference field and watches the note disturb it, which keeps working on currency “in which the magnetic field produced by the ink has been weakened.” The sensing head it specifies is a magnetic erase head of the type found in audio cassette tape recorders.
A validator’s optical head carries its own lamps, because it cannot rely on the room.
“The optical sensor arrangement may include visible, infrared and ultraviolet light sources, and one or more image sensor to receive and record reflected, transmitted and/or fluoresced light from the banknote under investigation.”
The preferred embodiment takes three exposures in sequence: white LED, then the infrared LEDs, then the UV LED. Infrared is where the interesting work is. The patent exists because some current currency banknotes contain inks that exhibit absorption spectra which vary over the infrared sector of the electromagnetic spectrum, and one infrared lamp is not enough to separate them. It prints the same note twice, once in visible light and once in infrared.
The obvious use of a UV lamp is to make the thread glow. The older use is the opposite: to catch paper that glows when it should not.
“Counterfeit bank notes are usually printed on customary paper containing optical brighteners. In contrast, the paper of authentic bank notes is generally free from optical brighteners. Detection of optical brighteners in a bank note can thus serve as a sign of the presence of a forgery.”
Ordinary paper is brightened so that it looks whiter. Banknote paper generally is not. On a $5 or above, the UV picture is a dark field with one bright line down it, and both halves of that are worth reading.
Counting is a physics problem. The standard way to tell one note from two stuck together is to shine light through and see how much comes out the far side, and it fails in a way anyone who has handled worn cash will recognise.
“For example, an especially dirty bill may pass relatively little light through, thus appearing as two bills to the light sensor. Conversely, doubled bills that are especially clean or worn may let more light through than an average bill, giving the sensor the impression that a single bill is being transported.”
The fix in that patent is to read the note in reflection first, decide how grubby it is, and adjust what the transmitted reading is expected to be. The machine has to model the note’s life before it can count it.
Size is on the sensor list. A note acceptor senses reflectivity and transmission patterns, size of the note and the magnetic characteristics of the inserted note, and the ECB refers plainly to sensors checking the size of the euro banknotes. How much that is worth depends entirely on the currency.
Every U.S. denomination has been the same rectangle since the 1928 series, when the large-size notes were reduced to their present size, 6.14 by 2.61 inches. Euro notes are graduated by value instead. The ECB’s technical specifications run from 120 by 62 mm for the €5 up to 153 by 82 mm for the €200. Measure a euro note and you have narrowed it to one denomination. Measure a dollar note and you have learned that it is a dollar note.
The clearest published account of what a note-handling machine has to decide is a piece of European law. Decision ECB/2010/14 governs any machine a bank uses to put euro notes back into circulation, and its Annex IIa sorts everything a customer feeds into a deposit machine into five categories. The wording below is the Decision’s own.
| Cat. | Meaning | Properties | Treatment |
|---|---|---|---|
| 1 | Objects not recognised as euro banknotes | “non-euro banknotes”, “euro banknote-like objects”, “wrong image or format”, “large folded corner(s) or missing part(s)”, or a “feeding or transportation error of the machine” | “Return by the machine to the customer” |
| 2 | Suspect counterfeit euro banknotes | “Image and format recognised, but one or more authentication feature checked by the machine not detected or clearly out of tolerance” | Withdrawn, handed to the national authorities. “Do not credit to the account holder” |
| 3 | Euro banknotes that are not clearly authenticated | “Image and format recognised, but not all authentication features checked by the machine are recognised because of quality and/or tolerance deviations. In most cases unfit euro banknotes” | Withdrawn and sent for authentication. “May be credited to the account holder” |
| 4a | Genuine and fit | “All authenticity and fitness checks carried out by the machine giving positive results” | “Can be used for recirculation” |
| 4b | Genuine and unfit | “All authenticity checks carried out by the machine giving positive results. At least one fitness criterion checked giving a negative result” | Returned to the national central bank. Still credited |
Category 2 is the machine calling a note suspect. Category 3 is the machine declining to call it anything: the image and format came back right, but not every feature it checked came back cleanly. The Decision then says what that usually means in practice, and it is not fraud.
“Image and format recognised, but not all authentication features checked by the machine are recognised because of quality and/or tolerance deviations. In most cases unfit euro banknotes.”
The note is withdrawn from circulation and handed to the competent national authorities for authentication. The customer, meanwhile, may still be credited for it.
A machine type only reaches the ECB’s published list once a national central bank has run it against a standardised deck. The deck is not only genuine notes and seized forgeries.
“genuine euro banknotes, representative counterfeits found in circulation and ‘additional documents’, reproductions created by the Eurosystem that imitate certain properties of euro banknotes”
The central bank manufactures objects that are deliberately part-right, then checks whether the machine is fooled by them. The deck is refreshed approximately every two months to ensure that tests are carried out with counterfeits recently found in circulation.
A cash recycler asks a real note a second question before it hands it to the next customer: is this thing still good enough to be checked again? Annex IIIa of the same Decision lists twelve defects that make a euro note unfit, and it is a much more physical vocabulary than anything on the public side.
Graffiti is the third item on the machine list, and the entry under it is one sentence: at present there is no mandatory requirement to detect graffiti. Turn to Annex IIIb, the list for a trained human doing the same job by hand, and graffiti carries no such exemption. Write on a note and the teller has to pull it. The machine may hand it straight back out.
A person knows a dead note the moment they touch it. A machine has no sense of touch, so the Decision routes around the problem: as limpness normally correlates with soiling, limp euro banknotes are generally also detected via soil sensors. And soil itself is not dirt to the machine. It is a number: “Soil increases the optical density of euro banknotes.”
Annex IIIa is a legal text, not an engineering one, yet it keeps naming hardware. De-inked notes “might be detected by image detectors or UV detectors.” Folds can be caught “by sensors checking the size of the euro banknotes” and “by thickness sensors.” Soil has its own sensor. Five kinds of detector, named almost in passing in the Official Journal of the European Union.
The same second question runs at the top of the chain. The Federal Reserve’s own account of the cash cycle: each Reserve Bank has high-speed machines that process the banknotes to look for rips, soiling, and general wear and tear. Notes that fail are shredded.
An Advanced Counterfeit Deterrence Steering Committee recommendation from 2013 set the order, and the BEP still publishes it: the $10 first at approximately 2026, then the $50 (2028), $20 (2030), $5 (2032–2035), and $100 (2034–2038), pending developments in counterfeiting threats or technology. Each of those is a date by which acceptors already in the field have to recognise a note they have never seen. That is what the sample notes are for.
The $1 is not on the list. Congress renews a rider each year that forbids redesigning it, so the note with the shortest published feature list of the seven is also the one that stays exactly as it is.
The BEP’s own definition of who is eligible:
“companies that produce any type of equipment that handles banknotes (cash handling equipment) for commercial purposes, which includes but is not limited to, point of sale, counting and sorting, transit systems, vending, gaming, and note validating and dispensing in commercial environments”
A transit ticket machine and a slot machine sit in the same programme as a bank’s cash recycler.
Several of these documents carry thresholds, tolerances and calibration values. Those are not reproduced here.