Security glass is rated in blows, not minutes

Every physical security control does one of three things: it deters, it delays, or it detects. Glass is a delay device. It is not going to stop anyone determined, and the industry does not pretend otherwise. A technical paper presented at Glass Performance Days puts the objective plainly: while it may not be possible to completely prevent intrusion, the objective should be to provide sufficient delay to be able to detect and respond to the intruder.

If delay is the product, then time is the unit. That is what makes the rating landscape so strange. The codes that appear most often in quotes and brochures, on both sides of the Atlantic, are not denominated in time.

What the common European rating actually counts

EN 356 is the European standard for security glazing against manual attack, and it splits into two halves that test two completely different things.

The lower half, classes P1A through P5A, is a ball drop. A 4.11 kilogram steel sphere, 100 millimetres across, is dropped onto the pane in a triangular pattern from increasing heights. A specification published by Arts Council England for museums and heritage buildings gives the endpoints: for P1A, three balls from 1.5 metres; for P5A, three sets of three balls, nine in total, from 9 metres. Its table lists the impact energy per stroke rising from 62 joules at P1A to 370 joules at P4A and P5A, with suggested glass thicknesses from 6.8 to 10.3 millimetres.

The upper half, P6B through P8B, switches weapons. The pane is struck repeatedly by a mechanical hammer and then an axe, with the force controlled throughout, and the rating is the number of blows the glass survives. The Arts Council table gives minimum blow counts of 31 for P6B, 51 for P7B and 71 for P8B. The Glass Performance Days paper describes the same procedure from the laboratory side, a minimum of twelve hammer strikes to break the glass in a rotated square pattern, then the axe head working the same square until it cuts through, with the total tallied: 30 to 50 strikes earns P6B, 51 to 70 earns P7B, and more than 70 earns P8B.

A trade data sheet for the standard fills in the mechanics, and the detail is worth having because it shows how tightly controlled the "attack" is. The hammer head is 40 millimetres square and weighs 2 kilograms, the axe head also weighs 2 kilograms, each blow is delivered with exactly the same energy, and the axe is sharpened every ten blows. The target is a 400 millimetre square aperture: the rating is how many identical blows it takes to cut that hole.

This is a good comparative test and a strange model of a burglary. Nobody sharpens their axe every ten swings, and nobody stops when the hole is 399 millimetres.

Notice what is missing from both halves. Nobody is holding a stopwatch, and nobody chose the tool.

The sentence that should be on every quote

The Arts Council specification, having laid out both tables, states the limitation directly:

BS EN 356 does not classify glass according to its resistance to different levels of attack, different tools or time available to an intruder.

That is a public body telling its own institutions that the most widely quoted security glazing rating does not answer the three questions a security plan is built on. It is not a criticism of the standard. EN 356 does what it says, which is to compare glazing products under a repeatable mechanical test. The failure is in how the result gets used downstream, where a P-number arrives in a proposal carrying an implied promise about attackers and minutes that the test never made.

The American pair has the same shape

In the United States the two names you are most likely to meet are UL 972 and ASTM F1233, and they sit at very different points on the same scale.

UL 972, burglary resisting glazing, is the ball-drop family member. The Glass Performance Days paper describes it as subjecting the sample to multiple impacts of an 82 millimetre steel ball weighing 2.26 kilograms, across four categories that differ by drop height and test temperature: multiple impact, outdoor use, indoor use, and high energy. Pass means no ball penetrates. The threat it models is a smash and grab, someone breaking glass fast to take what is behind it, which is a real threat and a well-designed test for it. It is not a test of a person with a pry bar and a quiet hour.

ASTM F1233 is the other end. The paper describes it as a comprehensive test method with 41 classification levels, using blunt and sharp tool impacts, thermal stress and chemical deterioration. And unlike the ball drops, it is not a machine test:

Unlike the previous test methods discussed, this standard does not use a machine for the testing. Instead, it employs six muscular males weighing between 82 and 113 kg (180 – 250 lb.) to enthusiastically assault the glass.

The failure criteria are the interesting part, because there are two and they answer different questions. The first is contraband passage: an opening large enough to pass a 3 millimetre diameter shape is a failure. The second is body passage: an opening large enough to pass a rectangular object of 20 by 20 by 13 centimetres, given as 8 by 8 by 5 inches, is a failure. One of those tells you whether someone can reach through and work a latch. The other tells you whether someone can get in.

There is a level above that, which the paper calls enhanced forced entry, where the sample is first weakened with ballistic impacts, and the bullets are allowed through. The stated intent is to mimic an active shooter shooting the glass to soften it before attacking it with tools. And above that sits multiple forced entry assault, combining ballistic resistance with no penetration allowed and then forced entry, the category used for embassies, where the US Department of State's own SD-STD-01.01 applies alongside ASTM F1233.

The standard that does count minutes

There is a rating system built the other way round, and almost nobody outside the insurance and institutional world quotes it. The Loss Prevention Certification Board's LPS 1270 is described by the Arts Council as an alternative in which the test is carried out by a person able to vary the choice of tools and method of attack, which it says is generally considered to better represent an actual attack than the robotic method used in BS EN 356.

It has eight security ratings, SR1 to SR8, and each one is defined by two things a homeowner would actually recognise: which tools the attacker gets, and how long they have. The tool categories run from A, an opportunist using bodily force and minimal tools including knives, levers and glass cutters, through B with claw hammers and hand drills, C with axes, cold chisels, crowbars and cordless drills, D with disc grinders, fire axes, hooligan bars and sledgehammers, and on to D+ through G with circular saws, reciprocating saws, oxyacetylene cutting kits and chainsaws. The maximum working time attached to each rating in the Arts Council table climbs from 1 minute at SR1 to 20 minutes at SR8, inside maximum test durations of 10 minutes and 60 minutes respectively.

And a single product does not get one number. It gets three, one for each of three outcomes the standard defines separately: local penetration, described as making a hole big enough to pass a wire or screwdriver to operate a panic bar or lever handle; hand hole access, big enough to reach the lock or an item of value on the protected side; and complete access, big enough to pass a 400 by 225 millimetre test block, replicating the size a person may be able to get through.

Three numbers, because the glass will fail those three ways at three different times. That is the level of resolution a real security design needs, and it is precisely the resolution that gets compressed away when a spec says "security glazing, P5A."

Two more standards worth recognising on a quote

EN 1627 rates the whole doorset or window rather than the pane, in classes RC1 to RC6, against manual burglary attempts that assume the intruder uses stealth. Its test duration counts contact time, rest time, observation time and tool change time, which is a more honest model of a break-in than a continuous assault. It also contains a detail I like enormously, because it shows a test written by people who wanted the truth rather than a pass:

The test attackers are given access to door plans, allowing them to identify potential vulnerabilities and giving them an advantage over an actual attack situation.

The testers are deliberately given better information than a burglar would have. The rating is therefore conservative by construction. That is what a well-designed security test looks like.

PAS 24 is the opposite kind of document and it is worth knowing why. The Arts Council describes it as a minimum standard that does not have grades or performance scores, assessing windows and doorsets against an opportunistic burglar using tools that are easy to conceal, on a pass or fail basis. A PAS 24 window is not a bad window. It is a window that cleared a floor. If a proposal offers it as the security tier, the correct question is what is above it.

The rating belongs to the assembly, not the pane

Here is the failure mode I would bet money on in a residential installation, and both of my sources flag it independently.

The Glass Performance Days paper attaches a caveat to its own results that generalises to every number in this report:

The testing was conducted on glass only samples whereas certification may require the full frame and glazing assembly to be tested.

The Arts Council makes the same point from the installation end, and gets specific. Its framing guidance calls for an adequate depth of rebate or edge cover for the glazing, typically 25 to 30 millimetres; bonding the glazing into the frame with a suitable compatible neutral curing silicone; securely fastening the frame into the wall or the display; and framing the glass on all sides. Its blast section is blunter still, noting that blast-resistant glass is designed to work together with specially designed frames and that the frame must be approved for use with it for the glass to be effective.

None of that is exotic. It is the ordinary observation that a pane which survives 71 axe blows is irrelevant if the sash it sits in can be levered out of the opening in twelve seconds. A rating tests a system. An installer who reads it as a property of the glass has quietly transferred the certificate to a component that was never tested alone.

Retrofit film, and the one line that disqualifies most of it

For an existing house, particularly a listed or architecturally significant one, replacing the glazing may not be an option, and security film is the standard answer. The Arts Council endorses it for exactly that case, and then adds two qualifications that between them decide whether the money did anything.

The first:

Note: film which provides UV filters only adds no security properties to glass.

A great deal of film sold into homes is solar or UV film. It is doing a real job, and the job is not security. The word "film" on an invoice is not a security claim.

The second is the installation requirement, and it is the same assembly point in miniature:

Window film should be applied in combination with an edge retention system which anchors the film to the surrounding frame.

Film that is not anchored to the frame turns broken glass into one sheet of broken glass, which then leaves the opening as a single piece. Anchored film keeps the fragments in the frame. The difference between those two outcomes is an attachment system, and it is the line item most likely to be value-engineered out of a retrofit quote.

Why this beat cares

I help design the AI security systems for a veteran-owned (SDVOSB) home-security company run by fellow veterans. I do not own that company and earn nothing from this link. Full policy here.

My first career was counter-IED and electronic warfare, and blast glazing is the corner of this subject I have the most direct feel for. In that world nobody asks whether a window is strong. They ask how far the fragments travel into the room, which is exactly how the blast standards score it, on hazard to occupants rather than on whether the glass held. The same discipline applies one threat level down. The question is never "is this security glass." It is: against which tools, for how long, and until who arrives.

That last clause is the one that makes the glass rating useful or useless. Delay is only worth what the response is worth. I have written before about whether anyone actually comes when your alarm trips, and about why buying more gear does not integrate into protection. A pane rated to hold a determined attacker for five minutes is an excellent purchase if detection is instant and a response arrives in four. It is decoration if nobody is watching the sensor, and it is over-specified if the door beside it opens to a pry bar in forty seconds. Glazing is a delay budget, and a delay budget only balances against a detection time and a response time.

What I could not confirm

I did not open any of the standards themselves. EN 356, EN 1627, UL 972, ASTM F1233, LPS 1270 and PAS 24 are all sold rather than published. Every number in this report is taken from two documents that describe them, and I have said which one each figure came from in the source notes. Where the two overlap they agree, which is reassuring but is not the same as reading the source text. Treat every figure here as an orientation to what the standard measures, not as a specification to build against.

One of my two sources has a commercial interest. The Glass Performance Days paper was written by an author at Kuraray, which manufactures the laminate interlayers the paper evaluates, and its conclusion is that thicker interlayers deliver higher security levels. I have used it only for descriptions of test procedures, which are checkable against the other source where they overlap, and not for any product or material recommendation. Its own test data is explicitly presented as a guide and not as certified security products, and I have not reproduced any of it.

My sources disagree on one figure in a way worth stating. The Glass Performance Days paper gives the EN 356 ball as weighing 411 kilograms, alongside a parenthetical of 9 pounds. Those cannot both be right, 9 pounds being about 4.1 kilograms, and the third source I opened gives 4.11 kilograms for the 100 millimetre ball. I have used 4.11 and I am treating the other as a typographical error rather than a disagreement. That same paper's figure captions also refer to UL 975, and to UL 972 ballistic specifications, where UL 972 and UL 752 appear to be meant. None of this affects its procedural descriptions, but it is the reason I have not leaned on its figures.

The blow counts for the EN 356 B classes are reported slightly differently across sources, 30 to 50 strikes for P6B in one and a minimum of 31 in the others. Those are consistent readings of the same band rather than a conflict, and I have given both. The suggested glass thicknesses do genuinely differ: the Arts Council table gives 8.8 mm at P2A and 9.1 mm at P3A where the data sheet gives 8.1 mm and 8.5 mm. The endpoints they share, 6.8 mm at P1A and 10.3 mm at P5A, agree exactly. These are suggested constructions rather than requirements of the standard, which is presumably why they drift, and it is why I have quoted only the endpoints and would not specify a thickness from any of them.

The Arts Council document is a UK museum security specification, reviewed in May 2024, and its guidance is framed for collections and heritage buildings under a government indemnity scheme, not for private homes. I am using it for what the standards measure, which does not change with the building. Its own recommendations about which class to specify are aimed at insured objects and should not be read as a residential recommendation.

I did not test, evaluate or inspect any glazing product, film, frame or installation, and none is recommended here. I am not a glazing engineer and this report is not a substitute for a security consultant or a structural specialist assessing a specific opening. My published research is in microwave spectroscopy and has nothing to do with glass.

The signal

Security glazing is sold as a grade and bought as a feeling of safety, and the grade usually describes a machine dropping a ball. That is a legitimate comparative test and a poor description of a burglary. The standards that describe a burglary, the ones with tool categories and a clock and separate numbers for reaching through versus climbing through, exist, and are the ones you have to ask for by name.

So when a quote says security glass, the questions are: which standard and which class, tested as glass alone or as the full frame and glazing assembly, and what is the rebate depth and fixing detail of the frame it is going into. If it is film, ask whether it is security film or solar film, and whether an edge retention system is in the price. And then ask the only question the glass cannot answer for itself, which is how many minutes of delay you actually need, given how long it takes someone to notice and how long it takes someone to arrive.