Pick the closure by what has to sit under it. A two-piece push-turn cap is the normal route when a dropper, roller or pump is fitted; a one-piece push-turn cap suits a plain lined pack on a standard threaded finish; squeeze-turn caps with side tabs need lugs on the bottle that must be designed into the mould from the start. Whichever you choose, the claim is only proven by a test on the finished, filled pack with a child panel and a senior-adult panel, for example under 16 CFR 1700.20 in the United States or EN ISO 8317 in the European Union.
Child resistance belongs to the filled pack
A cap is not child resistant because of how it is built or what a catalogue calls it. The property exists only once the assembled pack, holding the actual product and closed at the torque the capping line really applies, has been put through the panel study the relevant protocol describes and has come out inside its limits. Most errors in this category start from treating the cap as the thing being certified.
Three points follow for a buyer.
- A component report is not enough. A bare closure can pass on a bench and the finished unit can still fail. Gaskets compress differently once the bottle is filled and sealed, a heavier bottle gives a child a better grip, and a shrink band or tamper ring alters what the fingers touch. A report with no filled bottle in it shows that the mould works and nothing more.
- Two panels, one report. The child panel checks that the pack resists opening. The senior-adult panel checks that an ordinary adult can open it, unprompted, in reasonable time. Failing either one makes the pack unacceptable: the first because it is unsafe, the second because the intended user cannot get in.
- Repeated use is part of the test. Many protocols have the pack opened and reclosed a set number of times before and during the study. That exposes a closure that works only when new, such as a push-turn shell whose lugs have worn smooth after a dozen cycles. Closing torque, lug depth and shell material therefore belong in the specification, not in the assembler's discretion.
Terms to use and terms to avoid
"Child resistant" is the tested claim. "Child proof" is everyday speech with no technical definition, and it should stay off purchase orders. When people search for child resistant glass bottles they nearly always mean a glass container supplied with, or prepared for, a safety closure; the glass does nothing on its own. "CR glass bottles wholesale" is the same combination bought in volume. Child resistant dropper bottles are the narrow case in which the dispensing part and the safety closure are one assembly, and that is where most structural trouble turns up.
Push-turn, squeeze-turn and the less common alternatives
Almost every child-resistant screw closure in the pharmaceutical, nutraceutical and essential-oil trade is one of two types. Side by side they look alike. In the hand and on a capping line they have little in common.
Push-turn
The user presses the cap down along the axis of the bottle and turns it while it is held down, so the release movement is vertical. In the one-piece lined form, pressing compresses a spring element in the moulded body and frees the inner thread or lug path. In the two-piece form, an outer shell turns idly until it is pushed onto an inner core; that core can carry a pipette collar, a dropper insert or a restricted orifice. Because the shell rotates without dragging what is underneath, push-turn is the usual choice whenever a dispensing insert lives below the cap.
Push-turn depends on axial room and on the spring recovering. The neck has to offer enough thread engagement to hold the cap at its seated position plus additional travel for the push, and the inner element must spring back every time. On the line, seating depth has to be controlled: capped short, the closure sits above its release point; capped deep, the spring or liner is damaged. A specification that gives a torque value but no seating depth covers only half of the assembly.
Squeeze-turn
Here the user pinches two opposite walls, or two marked points, until internal tabs clear a pair of lugs, then turns. The release is sideways. These caps are generally a single moulding, simpler and shorter than push-turn, and they are common on detergents, household chemicals and some cosmetics that are poured or squeezed out.
The design rests on wall flexibility and lug geometry. The shell must flex far enough to release and recover far enough to lock again, which is a function of the polymer, the wall thickness and the temperature around it. A shell that feels right at 22 degrees Celsius turns stiff in a cold warehouse and soft, and easier to defeat, in a hot one. The same flexing distorts anything mounted inside, so squeeze-turn pairs badly with droppers and pumps.
Tool and alignment closures
A smaller group gets its resistance from a tool or a sequence: a cap released with a coin, or a mark on the cap that has to line up with a mark on the bottle before it will turn. They are legitimate and appear in some regulated categories, but they nearly always sacrifice day-to-day ease for people with arthritic hands. Where the market permits them and the contents are truly hazardous they can be correct. For a wellness product in a health-food shop they seldom are.
Closure systems compared
The glass often stays constant across a range while the closure changes from one SKU to the next, so the comparison below is arranged by closure system. The protocols named are the ones commonly cited; treat them as points to raise with the plant and with your regulatory advisor, not as evidence that a given pack complies.
| System | How it opens | What the neck finish must provide | Dropper, roller, pump or insert | What the test should cover | Typical complaints |
|---|---|---|---|---|---|
| One-piece push-turn with flat liner | Press down on the bottle axis, keep pressing, turn | Standard threaded finish, wide flat sealing band, thread height that lets the cap travel beyond its release point; no locking feature on the bottle | Cannot hold a dropper or pump as one unit. A loose dropper or orifice reducer may sit in the neck but the cap does not retain it, so it falls out or goes missing | Finished pack, child panel and separate senior-adult panel, under the destination market's protocol such as 16 CFR 1700.20 or EN ISO 8317 | Slow seepage where the liner is unevenly compressed; a loose feel after a few cycles once the liner has taken a set; no release when capped too deep |
| Two-piece push-turn with inner core | Push the outer shell onto the core, then turn the shell | Threaded finish sized to the core, not the shell; axial clearance above the thread for shell travel; sealing band as above | Takes a dropper collar, roller fitment, pump or sprayer on the core. The only system here in which an integrated dropper is routine | Finished pack with both panels; the report should name the dispensing element fitted, since another insert changes what the panel handles | Shell spins without releasing because the engagement has worn or a tight carton has shortened the push; dropper pulls from its collar and leaks into the cap; parts separate in packing |
| Squeeze-turn, side tabs over bottle lugs | Pinch two opposite points until tabs clear lugs, then turn | Thread plus bottle-side lugs at a defined angular position, toleranced for height and position; lugs are cut in the mould and cannot be added afterwards | Poor match for a dropper or pump: the pinch deforms whatever is inside, more so on a wide cap than a narrow one | Finished pack with both panels; room temperature recorded, because shell stiffness and therefore the result vary with it | Cracking at the hinge point after repeated squeezing; hard to open cold, easy to defeat warm; lug position drifting within tolerance so some packs open on a light squeeze and others not at all |
| Squeeze-turn with integral pump or fine-mist sprayer | Pinch to release, turn, then work the pump underneath | Finish and shoulder must clear the pump body and dip tube; the shoulder is often widened so the pump skirt does not stand proud of the glass | Works by design, but only as a matched assembly from one supplier. A shell from one source on a pump from another is a regular cause of leaks and failed panels | Assembled, filled pack with both panels, pump actuated as a consumer would use it | Product drying in the pump well so the shell sticks; users pressing the actuator instead of the shell and assuming the pack is broken; removal torque rising in storage as residue glues the parts |
| Push-turn, metal shell over moulded core | Press the metal shell down and turn | Standard threaded finish with a generous sealing band; deeper push travel than plastic because metal deforms less | Not normally paired with a dropper. Chosen for a metal look on a regulated pack, usually with a plain liner and a separate, non-captive insert | Finished pack with both panels, in the finish and colour to be sold, as coatings change finger friction | Dents in transit that stop the release; coating wear at the grip points; condensation in a cold chain making the surface slippery |
Fitting a dropper, roller or pump under a CR cap
It can be done, and whether it works is a structural question before it is a regulatory one. This is also the requirement that most often sends a bottle back for redesign.
Start with height. A safety cap stands taller than a plain cap because it needs release travel, but the extra interior space is not automatically useful. A dropper collar has to grip inside the neck. A pump needs a flat land on the shoulder and a straight run for its dip tube. Put a tall cap on a short neck and the pump has no support, so the assembly wobbles.
Next, work out which part holds the seal. On an ordinary dropper bottle the cap clamps the collar to the neck. On a two-piece safety cap the rotating piece is not the one gripping the collar, and a collar held only by friction against a turning shell loosens within a few dozen openings. The dependable layouts are a core keyed to the collar, or a collar that snaps positively into the neck, so the seal stays still while the shell turns.

- Rollers are short and sit low, so they are easier to house than droppers. Watch the ball diameter against the orifice, and make sure the neck gives the fitment a vertical stop. Without one, a capping head that seats to a fixed depth will crush it.
- Pumps and sprayers are the hardest. Each needs a crimp or screw engagement, a sealing land and a dip tube cut to the bottle height, and a separately bought safety shell adds an interface nobody is responsible for. If the two parts must come from different sources, test the finished assembly as one unit instead of relying on separate part tests.
- Tamper evidence, whether a shrink band, drop ring or tear strip, adds thickness exactly where fingers reach for the release. A band laid over the squeeze points turns a working closure into one nobody can open, and the first person to find out is usually a consumer.
Neck finish dimensions to settle before the mould is cut
Four groups of measurements decide whether a system will work, and all of them are cut in steel. Once the mould exists they are, in practice, fixed.
- The finish itself. This is the moulded mouth: thread or lug layout, outer diameter over the threads, height from sealing surface to thread root, and the flat ring on top where the gasket lands. Two bottles sharing a nominal diameter may differ in pitch, thread starts and sealing-band width, so a cap matched to one may not engage the other. Finish codes, industry or proprietary, are abbreviations. The drawing is what counts.
- Thread form and number of starts. Single-start threads take more turns to seat and tolerate more angular error. Multi-start threads seat in part of a turn, which suits fast lines, but offer fewer lock positions and can make the locked state feel vague. A push-turn cap needs a thread shallow enough for the inner element to pass it; a squeeze-turn cap needs its lugs where the hand naturally grips.
- Sealing band. The top of the neck must be flat and wide enough for the gasket or moulded sealing cone. Too narrow, and the gasket is partly unsupported and creeps under pressure. Chipped, and no closure will seal it, which is why incoming glass inspection matters as much as capping torque.
- Locking features. Some systems need a lug below the thread, a shoulder step or a vertical rib on the bottle. A glass mould can carry these, but they have to be drawn together with the finish. Leaving them until later is the most frequent reason a preferred closure will not fit a bottle that has already been designed.
Two further dimensions sit outside the mould. Headspace between the fill level and the sealing surface changes how the pack behaves under pressure. Total closure height above the shoulder decides whether the finished pack still fits the carton, shelf tray and shipping case the brand has already drawn.
Tolerances, gauging and sampling
A nominal diameter with no tolerance leaves room for the tooling to drift across the permitted range, and the result is a cap that fits one delivery and leaks on the next. When we check a drawing for a CR project, these are the dimensions we expect to see with explicit bands:
- thread outer diameter, and thread pitch or lead
- finish height
- flatness and width of the sealing band
- on lug systems, lug height and the angular position of every lug
- the locking feature dimensions
The last item is where safety is won or lost. A tab slightly too deep refuses to release; one slightly too shallow gives way under light pressure.

Glass is normally checked with go and no-go gauges, a plug gauge in the bore, and a caliper or optical comparator on the thread form. During a run, control comes from functional checks at set intervals, not a full dimensional study of each pallet. For CR packs the checks that matter are application torque, removal torque while locked, removal torque after the release action, and a leak test at the pressure the filled product will actually experience.
Write the sampling plan into the purchase order. The common basis is an acceptance quality limit scheme at an agreed inspection level for cosmetic and dimensional attributes, tightened for critical ones such as the sealing surface and locking geometry. Naming the standard and the limit per attribute at enquiry means the supplier inspects against the same document. Leaving it out means arguing about "normal industry practice" in the middle of a dispute.
Test protocols and what the report must contain
Rules vary by market and by how the product is classified, and nothing here is legal advice. What a buyer can do is understand the form the obligation takes and judge whether a supplier's answer matches it.
In the United States, the Poison Prevention Packaging Act and its implementing rules cover certain product classes. The protocol in 16 CFR 1700.20 describes how the panel study runs, how many children take part, how long they have, and what share may open the pack before it fails; a senior-friendly panel is required alongside. In the European Union the usual reference for reclosable packs is EN ISO 8317, which likewise sets a child test and a separate adult test. Other markets run their own schemes, and some impose no mandate on a particular product class.
In practice this means four things.
- The duty generally sits with the pack and with whoever places the product on the market. That party needs a report referring to the finished pack with the real contents.
- Bottle, closure, insert and gasket should each be identified by drawing or part number, so that a later swap is recognised as a change needing a retest and not waved through as equivalent.
- Change control is part of compliance. A new liner compound, gasket thickness, dropper collar or bottle supplier makes it a different pack. Buyers with a compliance file keep a short list of such triggers and require notice before any is acted on.
- Classification, not glass, determines whether child resistance is compulsory. One botanical extract may fall into different classes in different countries. That is for the brand's regulatory advisor to settle; an honest supplier offers a set of options and asks for the destination and classification to be confirmed.
Torque, release force and ease of opening
Resistance and usability pull against each other, and design can only manage the conflict. Anything that makes a pack harder for a child makes it harder for an adult with arthritis too, which is the reason the senior-friendly panel exists.
The quantities to measure are application torque, removal torque locked, removal torque after release, and the force and travel of the release movement. Application torque is set on the line and governs both the seal and what the consumer feels. Set low, the pack weeps in transit. Set high, an adult with small hands cannot open a closure whose geometry is perfectly sound.
Before redesigning anything, look at two things. One is grip: ribs, a scalloped skirt or a textured band let the user deliver the same force with less effort and change how a panel experiences the pack. The other is what the contents do in the thread. A buyer cannot control it but should ask, because a product leaving dry residue there behaves very differently after a month in a warm warehouse.
There is a commercial side as well. A compliant pack that people find difficult gets forced open with a tool, sent back or reviewed poorly. Between two compliant systems, the one with a clearer grip and a more positive lock generally does better on the shelf.
Failures that can be designed out
Complaints about CR packs repeat themselves, so most can be prevented at the specification stage.
- Seepage at the sealing band. The cause is nearly always the gasket, the band width or the capping torque, not the thread. Thin liquids and alcohol-based formulas escape through paths water would not, and a part-empty pack breathes as temperature changes. Check with a cycle test lasting weeks, not a bench test lasting minutes.
- Residue lock. Sticky or resinous contents dry in the thread and the cap well, so removal torque climbs in storage and the release stiffens. Choose a cap whose well drains back into the bottle, or one with little exposed thread volume.
- No release after shipping. A pack that works on the bench will not budge after a container voyage. The shell has usually been deformed in a stack-loaded carton, or the carton lid has pressed on the cap and shortened its travel. A heavier carton or a lower fill height cures more of these cases than altering the closure.
- Colour shifts between batches. They show more on a cap than on glass, since a cap is a small, saturated patch of colour. This is process control: agree a reference sample and a stated method of tolerance instead of a visual promise.
- Parts missing or separated. Shell and core of a two-piece closure can come apart in transit or in semi-automatic capping. Closures supplied pre-assembled and bagged remove the problem.
- Late revalidation. The costliest failure is finding, shortly before launch, that the pack on the line is not the pack that was tested because someone substituted a lower-grade dropper collar. A brief change-control list shared with every supplier in the chain prevents it.
Retention samples and batch records
A test report shows how the pack performed on the test day. It says nothing about the shipment now at sea. Retention samples bridge that gap, and they cost very little next to a recall or a retailer chargeback.
Normal practice is one sealed sample per production batch, stored in the conditions the product meets in trade, for the product's shelf life plus a margin. It must be a complete filled pack. Seepage, torque decay, gasket set and residue in the cap well do not show in an empty bottle.
The batch record should let you rebuild the pack years afterwards. Record:
- the glass batch, and the mould cavity where the supplier tracks it
- the closure lot
- the gasket or liner lot and compound
- application and removal torque measured during the run
- the capping head setting
- date and line
A record listing only the glass is useless when the complaint concerns the cap.
Include the outer carton in the sample whenever the carton forms part of the claim. That happens more than brands assume: an internal tab, a blister or a locked flap may itself be the child-resistant feature, with an ordinary bottle inside. The tested object is then carton and bottle together, and a retained bottle alone proves nothing.
What to send, and where related questions are answered
We do not publish quantities, unit costs, tooling charges or timings for CR packs, because none can be stated honestly until the specification is defined. They are quoted per project and depend on the mould, the closure system, decoration, quantity and freight terms. To get from enquiry to a workable specification, send:
- the market where the product will be sold
- the contents, with approximate strength, pH or acidity
- the fill volume in millilitres
- whether a dropper, roller, pump or plain liner is intended
- whether the pack must be tamper evident
With those details we can name a closure system, the finish it requires and the test route that applies. A photograph of a competitor's pack is usable but slower, since finish and insert geometry then have to be inferred.
Some neighbouring questions are handled elsewhere. How a pipette collar physically fits a neck, and how many drops an orifice delivers, are covered in our dropper bottle engineering guide, which uses the same drawing convention for finishes. For a botanical extract bound for a market with a packaging mandate, the tincture bottle buying page works through contents, gasket and destination together. Oil-based products under the same closures, including the matter of aroma families, are on the essential oil bottle page. If all you need is a stock amber or flint bottle with standard caps and no safety closure, go to the cap and finish list for Boston round bottles with caps.
Frequently asked questions
Does my product have to be in child-resistant packaging?
The destination market and the product's classification there decide it; the bottle and cap do not. Some markets mandate CR packaging for defined classes and publish a protocol, some rely on general product-safety duties, and some have no mandate for a given class. State both market and intended classification when you enquire and have the requirement confirmed for that pairing.
Which is better, push-turn or squeeze-turn?
Neither, in general terms. Push-turn fits packs with a dropper, pump or insert under the cap, since its axial release leaves the insert undistorted. Squeeze-turn is simpler and shorter and suits poured or squeezed products, but it goes stiff in the cold. Let the dispensing method and the climate of sale decide.
Can a dropper be combined with a child-resistant closure?
Yes, normally as a two-piece push-turn assembly whose inner core carries the collar while the outer shell rotates. Order it as one matched set including the dropper, make sure the collar is not held by friction against the shell, and have the filled pack tested as a whole.
Does the bottle mould need special features?
That depends on the system. A one-piece push-turn cap generally runs on a standard threaded finish with a wide, flat sealing band. Squeeze-turn caps with side tabs require lugs on the bottle at a set angular position, cut into the mould, so the decision has to be made together with the finish.
What test report should a supplier show?
One covering the finished, filled pack with a child panel and a separate senior-adult panel under the destination market's protocol. It should identify bottle, closure, insert and gasket. A closure-only report does not support the sale.
How long should retention samples be kept?
For the product's shelf life plus a margin, one sealed filled pack per batch, under trade conditions, with the carton where it carries the CR feature. Keep the closure lot, gasket compound, torque readings and capping head settings with it; those are what you will need when a complaint surfaces two years later.
What does "child proof glass bottles" mean on a purchase order?
Technically, nothing. Use "child resistant", defined as passing the applicable finished-pack protocol, and name the market. Buyers who accept catalogue wording tend to learn the difference when a retailer or customs broker asks for the report.