How to Choose Book Security Tags for Your Library Collection
Jan 22, 2026
Leave a message

Book security tags fall into three families: EM strips, which only trigger an exit gate; HF RFID labels at 13.56 MHz, which identify the item and drive self-service; and UHF labels at 860–960 MHz, which read a shelf section at a time. Collection size matters less than workflow. If you need item-level identification at a desk or kiosk, HF is the default answer.
What follows is the selection detail: a comparison of the three families, a matrix that maps them to workflow drivers, the label-format rules that fail most often, the ISO 28560 clauses that keep you out of vendor lock-in, and a read-rate acceptance test to run before the full order.

What Each Tag Family Can and Cannot Do
The three families are not competing versions of the same product. An EM strip is a magnetic marker - it answers one question at the exit, and it answers it without knowing which item it belongs to. RFID book tags answer a different question: which item is this, and is it currently authorised to leave. Choosing between them is not a technology upgrade decision, it is a decision about how many questions your workflow needs answered.
| Attribute | EM strip | HF RFID label | UHF RFID label |
|---|---|---|---|
| Operating band | Low-frequency magnetic field | 13.56 MHz | 860–960 MHz (region-dependent) |
| Air interface standard | De facto industry practice, no ISO air interface | ISO/IEC 15693 / ISO/IEC 18000-3 Mode 1 | ISO/IEC 18000-63 (EPC Gen2v2) |
| Library data model | Not applicable | ISO 28560-2 or ISO 28560-3 | ISO/TS 28560-4 |
| Item-level identification | No | Yes | Yes |
| Typical working distance | Gate aisle width only | Typical range 10–30 cm at a desk or kiosk pad | Typical range 1–3 m handheld in dense stacks; far less than free-air datasheet figures |
| Security mechanism | Strip physically deactivated / reactivated | Security bit written on the tag at checkout | Security bit or ILS lookup at the gate |
| Where it fails | Alarm fires but staff cannot tell which item | Too slow for aisle-by-aisle stocktaking | Cross-reads items the patron never presented |
One consequence is worth stating plainly, because it decides migration budgets: HF and UHF are not interoperable. A collection already converted to 13.56 MHz cannot be moved to UHF by swapping readers. The Danish standards body's RFID in libraries Q&A is explicit that the two are different technologies and do not offer a cost-effective migration path for a library that has already implemented one of them.
Selection Matrix: Match the Tag to the Workflow, Not to the Collection Size
Volume counts are a poor primary criterion. A 30,000-volume school library running six periods of class-set circulation a day has a harder workflow problem than a 200,000-volume reference collection that circulates almost nothing. Start from what the staff actually do repeatedly.
| Primary driver | Indicated choice | Condition that overturns it |
|---|---|---|
| Exit theft only; barcodes stay; no self-service planned | EM strips + gates, keep existing barcodes | A self-service or automated-returns project is already funded for the next budget cycle |
| Self-checkout, automated returns or sorting | HF, encoded to ISO 28560 | Your existing installed kiosks are UHF - then follow the installed base |
| Shelf inventory and misplaced-item search dominate staff time | UHF handheld, with UHF item tags | Reading rooms are crowded and shelving is dense, where cross-reads become the larger cost |
| Both circulation and frequent full-collection audits | HF for circulation, UHF handheld as a second layer | Budget cannot absorb two encoding regimes and two sets of spares |
| Special collections, archives, fragile paper | Non-invasive security; adhesive review before any label | Never overturned by cost - preservation policy governs |
| Retail bookstore loss prevention | EAS labels sized for softcover spines | Inventory accuracy, not shrink, is the actual pain point |
Five Questions to Answer Before Requesting Quotes
1. What is already installed? Frequency follows the installed gates, kiosks and sorters unless you are replacing all of them in the same project.
2. Which ILS, and over which protocol? SIP2, NCIP or a vendor API - confirm before hardware, not after.
3. How many distinct binding types are in the collection? Count them. Each one is a separate read-rate risk.
4. Who encodes - factory or staff? Factory pre-encoding removes a station from the conversion line but requires a clean item-identifier export.
5. What happens in year eight? If a different vendor cannot read and rewrite your tags, the answer is a full re-tagging project.
Before sizing the kiosk fleet, it helps to walk through how a self-service borrowing transaction actually executes end to end - the tag only carries one step of it.
Label Format, Placement and Substrate
This is where deployments go wrong quietly. A gate is commissioned once and then works. A label format decision is repeated across every item in the collection, and a bad one surfaces months later as an intermittent read failure that nobody can reproduce at the desk.
Size Selection by Item Type
Two square-ish formats dominate the library market for paper items, roughly 50×50 mm and roughly 50×80 mm, with dedicated circular formats for discs. A larger antenna generally buys read margin at the cost of fitting fewer item types - which is exactly the trade-off a mixed collection cannot make globally.

| Item type | Indicated format | Failure mode if you get it wrong |
|---|---|---|
| Standard hardcover | ~50×50 mm, inside back cover | Placed over the hinge, the antenna flexes and fractures after repeated opening |
| Trade paperback | ~50×50 mm, or narrower where the cover curls | Label edge lifts, then catches on shelving and peels |
| Thin journal or A5 magazine | Narrow format; verify it fits before ordering the batch | Label overhangs the trim line and is torn off in handling |
| Dense shelving, marginal read range | Larger ~50×80 mm antenna | Undersized antenna reads on the bench and fails between packed spines |
| CD / DVD / Blu-ray | Dedicated circular disc format, balanced design | Off-balance label causes vibration and playback errors |
| Foil-stamped or metallised cover | Test individually; may require repositioning to an interior page | Metallic layer detunes the antenna and the item reads intermittently |
Antenna geometry, chip choice and converting format all change the read result, so confirm the available custom HF and UHF library label formats before you fix a single size in the tender document. Discs, plastic cases and laminated covers do not behave like paper, which is where special-substrate tags for discs and non-paper media belong in the specification rather than as an afterthought.
Placement Rules, and Where Security Tags in Books Are Normally Hidden
EM strips are traditionally inserted deep into the gutter between pages or along the inner spine, precisely because they are meant to be hard to find and hard to remove. RFID labels take the opposite approach: they sit on a flat interior surface, usually the inside back cover, because the antenna needs to stay planar and unstressed.
Three placement rules carry most of the risk. Keep the label clear of the hinge, so that opening the book does not flex the antenna. Keep it clear of metallic or foil layers, which detune it. And keep the placement consistent - if staff apply labels at varying positions, a marginal reader configuration will produce a failure rate that looks random and is impossible to diagnose later.
Adhesive, and When Acid-Free Is Not Optional
Standard acrylic adhesive is adequate for general circulating stock. For special collections it is not a cost question but a policy question: an acid-free RFID label for rare books exists because conventional adhesives can yellow and embrittle paper over decades, and because removal may damage the substrate.
Two practical rules for preservation-sensitive material. First, run an accelerated ageing check on low-value items with comparable paper and coating, and inspect for edge lift, discolouration and removal damage before committing the collection. Second, if the item cannot tolerate any irreversible adhesive, do not solve it with a different label - solve it with a non-invasive security approach and accept that the item will not be part of the RFID circulation workflow.
Encoding: ISO 28560 and the Tender Clauses That Prevent Lock-In
The most expensive error in a library RFID project is rarely a weak tag. It is a tag encoded in a proprietary layout, discovered three or four years later when a different kiosk vendor cannot read the existing collection and the only options are re-tagging or buying from the incumbent forever.
ISO 28560 is a family, not a single document, and the part numbers matter in the tender wording:
- Part 1 - the set of data elements, independent of encoding
- Part 2 - fixed-length encoding on 13.56 MHz tags
- Part 3 - ISO/IEC 15962-based encoding on 13.56 MHz tags, allowing national or consortium profiles
- Part 4 - encoding for UHF tags with partitioned memory
Parts 2 and 3 are both HF and are not interchangeable with each other; a system built to one does not automatically read the other. The published summary of ISO 28560 sets out the differences, including the fact that Part 3 does not specify a locking strategy. For historical context on why library standardisation converged on HF rather than EPC Gen2, the ALA Library Technology Reports chapter on library RFID systems is the reference most integrators cite.
| Tender requirement | Exact question to the supplier | Acceptable answer |
|---|---|---|
| Data model | Which part of ISO 28560 is the encoding written to? | A named part (2, 3 or 4) plus any national profile, in writing |
| Air interface | Which ISO/IEC air interface and mode does the chip implement? | ISO/IEC 15693 and/or ISO/IEC 18000-3 Mode 1 for HF; ISO/IEC 18000-63 for UHF |
| Locking policy | Which memory blocks are locked at the factory, and by whose decision? | A documented block map; nothing locked without the library's written instruction |
| Third-party read/write | Can a different vendor's reader read and rewrite the required fields? | Yes, demonstrated on sample tags with hardware you nominate |
| Chip continuity | What is the chip model, memory size and stated data retention? | A named chip model with a published datasheet, not "compatible equivalent" |
| Exit clause | What does it cost to re-encode the collection to another compliant profile? | A quoted figure now, not a conversation in year five |
Acceptance Testing Before the Full Order
Suppliers quote read rates as a marketing figure. Turn it into a contractual one by sizing the sample properly. The useful rule here is the "rule of three": if a test of n items produces zero failures, the 95% upper confidence bound on the true failure rate is approximately 3/n.
| Sample size with zero failures | Failure rate you can claim (95% upper bound) | What that means on 100,000 items |
|---|---|---|
| 100 | ≈ 3% | Up to 3,000 problem items - not a usable guarantee |
| 300 | ≈ 1% | Up to 1,000 - the practical minimum for a pilot |
| 1,000 | ≈ 0.3% | Up to 300 - appropriate for a large multi-branch conversion |
Stratify the sample rather than drawing it at random: the pilot must contain every binding and cover type in proportion, because the failures cluster in the awkward categories, not in the hardcovers. Then test each item against every device that will touch it in production.
| Test | Method | Pass criterion to write into the contract |
|---|---|---|
| Single-item read | Each stratified sample on the kiosk pad at normal patron speed | Zero failures across the full sample |
| Stack read | Stacks of the maximum number of items the kiosk advertises | All items resolved on first presentation, repeated ten times |
| Gate detection | Un-checked-out items carried through at varied orientation, height and walking pace | Alarm on every pass; document the orientations tested |
| In-situ shelf read | Your actual shelving, fully loaded - never a sample board | A stated read rate per shelf metre that you accept as the operating baseline |
| Cross-read boundary | Items held at 0.5 m and 1 m from the kiosk while another item is presented | No unintended reads at the distances patrons actually stand |
| Interoperability | Read and rewrite with a second vendor's reader | Full read and successful rewrite of required fields |
| Adhesive ageing | Sample labels on representative stock, inspected after several weeks | No edge lift, no discolouration, no substrate damage |
Cost Model: What to Count Besides the Label
Comparing unit prices across three quotes is the least informative thing you can do with them, because the label is typically a minority of first-year project cost. Build the model from your own pilot numbers instead of a supplier's throughput claim:
First-year cost = (N × unit price × [1 + rework buffer]) + (N ÷ measured items-per-hour × loaded labour rate) + gates + kiosks and staff stations + ILS integration + training + project supervision
Annual hours returned = (transactions/yr × self-service share × seconds saved per transaction ÷ 3,600) + (audits/yr × hours saved per audit)
Three variables decide the answer, and only one of them is quoted to you. Measured items-per-hour must come from your own pilot on mixed stock, not from a desk test on new hardcovers - publicly reported library conversions span an enormous range, from tens of thousands of volumes over a year to hundreds of thousands in a matter of weeks, which tells you the number is a property of your collection and staffing, not of the tag. Rework buffer covers labels that fail verification or are applied to the wrong item; carry an explicit percentage rather than discovering it mid-project. Self-service share drives the entire benefit side, and it depends on kiosk placement and patron communication far more than on the label.
Encoding at the factory removes one full station from the conversion line, which changes the items-per-hour term directly - worth asking what pre-encoded label production and custom printing requires in export format and lead time before you plan the labour schedule.
If aisle inventory rather than circulation is the driver, specify the reader first and let the tag follow: compare UHF reader configurations for shelf-level scanning against your shelf depth and aisle width, because a tag chosen without reference to the reader is a guess.
Libraries that later extend tagging to equipment loans, study rooms and furniture often find it useful to see where item-level tagging sits in a wider asset strategy before fixing the encoding scheme for books alone.
Privacy and Data Minimisation
Store an item identifier and an operational status. Nothing else. Titles, patron records and borrowing history belong in the library management system, where access is controlled and auditable - not on a tag that can be interrogated by any compatible reader within range, including one outside the building.
A defensible specification answers four questions in writing: which data elements are written to the tag, which memory blocks are locked and when, who is permitted to read and rewrite them, and how a tag is retired when the item is withdrawn from the collection. Regional data-protection law may impose additional requirements, so route the specification past whoever owns privacy compliance before it goes into the tender.
FAQ: Book Security Tags
Q: Are RFID book tags the same as anti-theft security tags?
A: No. An EM strip is a security device only - it triggers a gate but cannot say which item it belongs to. An RFID label carries an item identifier and a security bit, so it performs both roles: identification for checkout, returns and inventory, plus exit detection. Many collections run both, using EM where identification is not wanted.
Q: Where are security tags hidden in books?
A: EM strips are usually inserted deep into the gutter between pages or along the inner spine, where they are difficult to see and difficult to remove. RFID labels are normally applied flat to the inside back cover, clear of the hinge and of any metallic cover layer, because the antenna must stay planar to read reliably.
Q: Should a library use HF or UHF RFID tags?
A: Use HF at 13.56 MHz for circulation, self-service and gates, where a short controlled field is an advantage rather than a limitation. Use UHF where aisle-by-aisle inventory dominates staff time. The two are not interoperable, so if equipment is already installed, follow the installed frequency instead of the better datasheet.
Q: What size RFID tag do library books need?
A: Roughly 50×50 mm covers most hardcovers and trade paperbacks; a larger format near 50×80 mm buys read margin in dense shelving at the cost of not fitting thin items. Discs need a dedicated circular format. Count the distinct item types in the collection first - a single size rarely serves a mixed collection.
Q: Can security tags in books damage rare or old material?
A: The chip and antenna do not, but the adhesive can. Conventional adhesives may yellow and embrittle paper over decades and can damage the substrate on removal. For special collections, use an acid-free construction or a non-invasive approach, and run an ageing check on comparable low-value stock before committing the collection.
Q: How many samples should be tested before ordering the full quantity?
A: At least 300 stratified items, tested with zero failures, before a conversion of any size. Under the rule of three, 300 clean reads support a claimed failure rate of roughly 1% at 95% confidence; 100 items support only about 3%, which is too loose to be a contractual guarantee. Large multi-branch projects should test 1,000.
Next Steps
For a library or archive project at the specification stage, we can supply:
Samples across formats, so you can run the stratified read-rate test on your own stock and shelving
Technical documentation: chip model, memory map, antenna drawing, adhesive type and air-interface standards for the tender file
Written confirmation of the ISO 28560 part your encoding is written to, and which memory blocks are locked
Quotation by format and volume, so the label line in your cost model uses real numbers

Email: ruby@synteksmart.com
Tel: +86-755-84876052
Enquiries: Send a project enquiry
Syntek Smart Technology Co., Ltd
Send Inquiry

