Glass Bottle Colour Consistency: Shade Acceptance

What Has to Be Defined Before a Shade Can Be Judged

Most colour arguments between a buyer and a factory are not arguments about colour. They are arguments about missing definitions. Two people can look at the same bottle, disagree honestly, and both be right, because they are comparing it against different things under different conditions. Until five things are defined, no judgement about shade is repeatable and no tolerance can be enforced.

The first definition is the reference object. Something has to be nominated as the thing the shade is supposed to match, and that something has to be a physical object rather than a description. A colour name, a Pantone-style reference, a photograph on a screen or a memory of the previous delivery will not do: screens shift, prints shift, and memory does not survive a year. The reference is a signed or otherwise identified physical sample with a date and an identity attached to it. The second definition is the route, meaning whether the colour sits inside the glass, on the glass, or is produced by a decoration that is not really about colour at all. The route determines which variables can drift and therefore which acceptance rule makes sense. The third is the viewing condition: the light source, its intensity, the background, the viewing angle and the distance. Glass is a specular material, and the same bottle changes shade depending on whether it is backlit, side-lit, on a white surface or on a dark one.

The fourth definition is the decision rule – what constitutes a pass and what constitutes a fail – expressed as an allowed difference rather than as an instruction to match. The fifth is the sample size and the sampling rule: how many bottles are examined from a lot, and how the lot is judged from what is found. Where the parties want a formal structure for that last element, the sampling schemes described in ISO 2859-1 and the associated AQL idea are the usual public reference points, and the inspection level and the acceptance number have to be agreed in writing rather than assumed. None of this requires a laboratory. It requires that the five definitions exist before the first delivery is judged rather than being invented during the first dispute.

Where Batch-to-Batch Colour Variation Actually Comes From

Understanding the sources of variation is what makes a tolerance possible, because each source can be attacked differently. A buyer who knows that a drift is coming from the melt has a different conversation from one who suspects the coating line.

For colour that sits in the glass body, the dominant source is the melt itself. Glass colour of this kind is produced by the composition of the batch and the redox conditions in the furnace, and furnace conditions move gradually with campaign age, raw material lots and the balance of cullet that is returned to the melt. That means a colour that was matched at the start of a campaign can drift slowly across months, in a direction that is usually consistent rather than random. A second source is the forming stage, because wall thickness and the distribution of glass in the body change how much coloured material the eye has to look through. A thin wall reads lighter than a thick one in the same colour, and a change in weight or in the mould’s condition enough to alter wall thickness will therefore read as a colour change even when the colour itself has not moved. A third source is annealing, which affects the internal stress state and can affect apparent clarity rather than hue. Finally, raw material lots can shift the shade slightly, particularly where the colour depends on iron or another minor component whose level varies between suppliers.

For colour that sits on the glass as an exterior coating, the dominant source is the coating process rather than the glass. Film thickness is the variable that matters most, and it is affected by spray parameters, the position of the part in the booth, transfer efficiency and the viscosity of the material. A film that is slightly thicker reads deeper, a film that is slightly thinner reads lighter, and a gradient across the surface can read as a colour difference between two bottles that were coated in the same run. The glass underneath still contributes, because a coloured film over a clear bottle and the same film over an amber bottle are two different products. Curing or baking conditions then affect the final appearance, and small shifts in oven temperature or time can change the shade of some systems. A further source is the surface preparation, since contamination on the glass can cause local differences in adhesion and in the way the film lays down.

For colour produced by decoration rather than by colouring the pack – organic or ceramic inks, transfers, or a printed layer – the sources are different again: ink batches, film or screen wear, curing, and registration. These processes are usually judged against their own appearance specification rather than against the glass shade, and they are mentioned here only so that a dispute is not misattributed. An apparent colour difference caused by decoration will not be fixed by adjusting the melt.

How a Shade Is Judged: Reference Object, Light Source, Viewing Angle

A judgement about colour is only repeatable if the conditions are fixed. Four conditions do most of the work, and all four are cheap to standardise.

The reference object should be physically present at the point of judgement, not described remotely. Where the reference is a sealed sample held by the factory and a duplicate held by the buyer, both should be identified in the same way so that a photograph of one can be matched to the other without ambiguity. The reference should be handled rarely, stored away from light and heat where its own material can age, and replaced on a defined cycle rather than when it is finally noticed to be wrong.

The light source is the single most common source of honest disagreement. Daylight, cool white fluorescent, warm interior lighting and a phone camera flash all render a coloured glass surface differently, and a factory inspection area lit for general work is not the same viewing condition as a retail shelf or a photographic studio. The practical answer is to agree a defined source for acceptance – a daylight-equivalent lamp with a stated correlated colour temperature is a common choice – and, separately, to check the pack under the light in which it will actually be sold. If the product is sold under strong warm light, a drift that is invisible under a neutral lamp may be obvious to the consumer, and that is worth knowing before the batch is made rather than after.

The viewing angle and background matter as much as the light. Glass is transparent and reflective at once, so the same bottle read against a dark surface and against a bright one shows two different shades, and a bottle held at an angle to the light shows more of its surface reflection and less of its transmitted colour. The agreed condition should state the background, whether the assessment is done in transmission or in reflection, and roughly where the viewer stands. A simple and repeatable convention is usually enough, provided both sides use the same one.

The observer and the sample complete the set. Colour judgement by eye is a human measurement, and individuals differ in their sensitivity, particularly in the blue region; the same person should ideally make the comparison, or at least a small number of trained people should have agreed beforehand what they call a match. The number of bottles examined also changes the answer, because a lot that contains a small proportion of off-shade bottles will look perfect if two are examined and imperfect if fifty are. That is the point at which a defined sampling plan turns a matter of opinion into a decision with a known risk on both sides.

How to Express a Tolerance Instead of Saying “Match the Sample”

The instruction to match the sample is the most expensive sentence in a colour specification, because it makes every delivery a matter of opinion. A tolerance does not have to be complicated to be better. It can be expressed in one of two ways, and many successful programmes use both.

The first way is visual grading against the reference. The reference defines the target, and a small number of permitted steps away from it are defined in words: a match, an acceptable but noticeably lighter or darker step, and anything beyond that is a rejection. The advantage is that this approach maps directly to what the consumer sees and requires no instrument. The disadvantage is that it depends on the judgement being made under the agreed conditions by people who share the same interpretation, which is why the wording and the viewing conditions have to be written down and the retained samples have to be shared. Where the product is decorative and the shade is not carrying functional meaning, visual grading against a well-kept reference is usually the right level of control.

The second way is instrumental measurement expressed on a colour difference scale. CIELAB is the usual colour space for this purpose, and the concept of a colour difference – the distance between two measured colours on that scale, usually written as a difference value – allows a tolerance to be stated and a result to be checked numerically. Most of the recognised ASTM methods for colour measurement describe how a sample is prepared, measured and reported rather than what tolerance to apply, so the method name belongs in the specification for consistency while the tolerance figure belongs to the buyer. Instrumental measurement has two real limits worth knowing: it needs a defined measurement geometry that suits a curved, translucent, reflective object, and it cannot fully represent the visual impression of a whole bottle on a shelf. Its real value is as a referee – a way to show objectively whether a shift has occurred and in which direction, which is exactly what is needed when two parties are looking at the same bottle and seeing different things.

Whichever route is chosen, the tolerance should be written as an agreed allowance rather than as a target with no margin. A target of zero difference is not achievable in glass production of any kind, and a specification that demands it will simply be enforced by argument. The practical structure is a target shade plus an allowed deviation, with a defined sampling plan for how many bottles are examined and how the lot is judged. It is also worth writing down what happens at the boundary – whether a lot that sits at the edge of the allowance is accepted, rejected, or accepted with a note – because boundary lots are where most disputes originate. Suggested tolerances are indicative only; what governs is the sealed sample together with the viewing and judgement conditions that both parties have signed.

Colour in the Glass versus Colour on the Glass: Why the Route Changes the Acceptance Rule

The same tolerance wording cannot be copied from one colour route to another, because what is being measured is different in each case.

Where the colour is in the glass body, the shade is a property of the material throughout its thickness and it is read in transmission. That gives the assessment a strong dependence on wall thickness and on how the bottle is lit, and it means that a shade difference can appear between two bottles that were formed from the same melt but with different wall distribution. It also means that the reference sample ages slowly and predictably, and that the main drift to manage is the furnace campaign rather than a surface process. Acceptance is usually visual against the reference under an agreed transmission condition, with instrumental measurement used when a disputed batch needs a referee.

Where the colour is a coating or sprayed film on the exterior, the shade is a surface property read largely in reflection. Thickness, coverage and uniformity become the controlling variables, and a batch can be off-shade in a way that is not only about the colour material but about how evenly it was applied. Acceptance then needs to consider not only the average shade but also consistency across the surface of a single bottle, because a gradient or a light patch is a defect in its own right, not merely a colour difference. Exterior coatings raise their own questions about adhesion, abrasion and solvent exposure, and where the film is intended to remain on a food contact surface the food contact position needs separate treatment. The two colour routes also interact: a coating over a coloured body multiplies the effect of variation in both, and a buyer who specifies both should expect the tolerance to be harder to hold than either alone.

Where the colour comes from decoration or printing, the reference is an appearance specification rather than a glass shade, and the controlling variables are the ink, the tool and the curing conditions. These processes are judged on their own terms, and a colour dispute in a decorated area should be routed to the decoration supplier and the decoration specification rather than to the glass plant. Sorting out which route is responsible is the first practical step in any dispute and is treated again below.

Reference Samples and Retained Samples: Who Keeps What, How Many, How Long, How Updated

Reference management is the unglamorous part of colour control and the part that decides most disputes. Two distinct kinds of sample should exist, and confusing them is a common mistake.

The reference sample is the governing object: the thing that defines the target. It should be a sealed or otherwise protected set from an approved production run, identified by a reference number, a date and the batch it came from. One set stays with the factory and one with the buyer, and the two are recorded as being the same thing. The reference should be handled as little as possible, kept out of direct light and away from heat, and stored so that it cannot be confused with ordinary stock. Where the product is sold against a retail listing, it is worth confirming that the nominated reference is the pack that was actually presented and approved, because a reference that predates a listing review can lead to a batch being rejected against a standard that no longer applies.

The working sample is a duplicate that gets used: handled, photographed, shown to the filling line, and compared against incoming deliveries. It will age and dirty, which is exactly why it must not be the governing object. Keeping one working sample and one sealed reference per colour separates the sample that gets used from the sample that decides.

How many to keep is a judgement about risk rather than a fixed rule. A reasonable structure is one sealed reference set per colour per site, one working sample per site, and a retained sample from each approved production run for the length of time the product is expected to be sold. That last set is what allows a later complaint to be investigated, because it ties a physical bottle to a production identity. Where several sites or several contract fillers are involved, each location that judges colour needs its own working sample, and the governing reference should be duplicated and its identity recorded, since a single reference sitting in one country cannot settle a dispute in another.

Updating the reference is the step most often neglected. A reference should be replaced when it has aged or been damaged, when the specification legitimately changes, when the product is reformulated or relaunched, or when the agreed shade itself is re-tuned – and each replacement should be recorded with a date, a reason and the identity of the new reference. Replacement should never be informal, because an unrecorded replacement means that from that moment the specification has silently changed. Where a colour has drifted slowly over a long production campaign and the buyer has accepted the drift at each delivery, the honest thing is to acknowledge that the effective reference has moved and to reissue a reference that reflects the accepted current shade, rather than to keep pointing at an original sample that can no longer be met.

Filling Line and Shelf Conditions That Make the Same Shade Look Different

Two bottles can be identical in colour and still fail in practice, because the conditions around them differ. Three situations are worth planning for.

The first is the shelf set. Retailers and distributors frequently display several units of the same product together, and sometimes units from different deliveries. Individually acceptable bottles can read as a mismatched group when placed side by side in a row under bright light, and the mismatch is more visible in a clear or pale product than in a dark one. Where this risk matters, the practical measures are to plan deliveries so that a shelf is not filled from two batches at once, to keep the older stock forward in the rotation so that a batch is consumed before the next one is opened, and to agree a somewhat tighter allowance for products that are routinely displayed in sets. This is a stocking discipline as much as a colour specification, and it is worth writing into the receiving instructions rather than leaving to the warehouse.

The second is the filling line. Colour itself does not change on the line, but everything around it does. Filling temperature, the presence of a vacuum or pressure step, capping torque and the closure lot can all interact with the pack, and a change in the product’s own colour across a production run – which is common in natural or botanical liquids – will make the bottle appear to have changed even when it has not. Where a shade complaint arises after filling rather than before, the first question is whether the liquid changed rather than the glass.

The third is light exposure and time. Some coloured packs, particularly those with an organic coating or an organic decoration, are affected by prolonged exposure to light, and the effect is usually gradual. Where the product has a long shelf life and will be displayed under strong lighting, it is worth asking what the expected appearance is at the end of life rather than only at the point of delivery, and whether the retained reference should be complemented by an ageing assessment. The same applies to storage: a reference sample kept in daylight for a year may itself have moved, which is another reason the reference must be stored properly.

Failure Modes in Colour Acceptance, and the Order in Which a Dispute Is Handled

Disputes follow a small number of patterns, and each has a countermeasure that is cheaper than the dispute. The most common failure is the missing definition: no light source, no reference, no tolerance, no sampling rule, and therefore nothing to measure against. The second is the aged reference, where the sample that defines the target has itself changed and both parties are arguing about their own version of the truth. The third is the wrong route assumption, where a colour difference appears in a decorated or coated area but is investigated as a glass problem, or the reverse, so the correct process is never questioned. The fourth is mixed stock, where the physical evidence is destroyed by warehouse rotation before anyone examines it. The fifth is the measurement without method, where a numeric tolerance exists but no agreed way of measuring the article exists either, leaving the number unenforceable.

When a dispute is actually live, the order of handling matters more than the strength of either party’s position. The first step is to stop mixing the questioned stock and identify it by batch, so that the evidence still exists. The second is to establish which reference governs and confirm that both parties are using the same one. The third is to compare under the agreed light and viewing conditions, and to record what was seen rather than what was concluded – photographs taken under uncontrolled light are of limited value and can easily mislead both sides. The fourth is to determine which colour route is responsible, because that decides who can actually act on the finding. The fifth is to apply the written allowance rather than a newly invented one, and only then, if the specification itself proves unworkable, to renegotiate it deliberately for future orders with the new reference recorded. The final step is to close the loop: whatever was learned about the reference, the light, the sampling or the batch identity should be written into the specification so that the same dispute cannot be re-run next quarter.

Where This Page Stops and the Neighbouring Pages Take Over

This page deals only with acceptance: how to define what a shade has to match, how to judge it, how to retain and update the evidence, and how to resolve a disagreement. It deliberately does not cover how colour is produced. Buyers whose open question is which colouring route to use, how a sprayed or coated surface is built up, how adhesion and abrasion are tested, or how an exterior coating stands in relation to food contact requirements should read glass bottle spray coating first, since the process decides which variables can drift and therefore what your tolerance has to protect.

Buyers who have not yet chosen a shade should look at the colour ranges offered for specific container families, in particular amber Boston round bottles for apothecary, essential oil and tincture work and brown amber glass bottles for the wider amber range used across beverage, spirits and chemical products, because shade selection is a different decision from shade acceptance and is answered better against real containers than in the abstract. A buyer still orienting themselves across container types, neck finishes and decoration options will find the range overview on glass bottles a more useful starting point than this page, and can return here once a container and a colour route are fixed.

Frequently Asked Questions About Glass Bottle Colour Consistency

Why can two deliveries of the same coloured bottle look different?

Because a colour that sits in the glass and a colour that sits on the glass each drift for different reasons. In-body colour follows the furnace campaign, the raw material lots and the redox conditions in the melt, and it can drift slowly across months. A coating follows film thickness, spray parameters and curing conditions, and can vary within a single run as well as between runs. Wall thickness also changes what the eye sees, so a forming change can read as a colour change even when the colour itself has not moved. Most apparent differences trace back to one of those sources, and identifying which one is the first useful step.

What light source should be used to judge glass bottle colour?

One source should be agreed for acceptance and written into the specification, typically a daylight-equivalent lamp with a stated colour temperature viewed against a stated background. A second check under the light in which the product is actually sold is also worth doing, because a drift that is invisible under a neutral lamp can be obvious under warm retail lighting. The important point is not which source is chosen, but that both sides judge under the same one, that it is recorded, and that it is not silently changed between deliveries.

Can a colour tolerance be expressed as a number?

Yes, and the usual approach is to measure the colour and express the allowed difference on a colour difference scale within the CIELAB colour space, with the measurement method named in the specification so that both sides report in the same way. The recognised ASTM colour measurement methods describe how samples are prepared, measured and reported rather than what allowance to apply, so the numerical tolerance remains a commercial decision for the buyer and the factory. A number is only enforceable if the measurement geometry suits a curved, translucent, reflective article and if both sides can actually run the same method.

How long should reference samples be kept and by whom?

A sealed reference should be kept by both the factory and the buyer for as long as the product is in production, and a retained sample from each approved run should be kept for the period in which a complaint could reasonably arise. One working duplicate per site is needed for daily comparison, and one sealed set is needed as the governing object that does not get handled. Where several sites or contract fillers judge colour, each needs its own working sample, and the identity of the reference held at each location should be recorded so that everyone is comparing against the same thing.

What should happen when a batch is rejected for colour?

First stop the questioned stock from being mixed with anything else and identify it by batch, so the evidence survives. Then confirm that both parties are looking at the same governing reference, compare under the agreed light and viewing conditions, and record what was observed rather than what was concluded. Establish which colour route is responsible, since that determines who can act, and apply the written allowance rather than a new one. If the allowance itself turns out to be unworkable, renegotiate it deliberately for future orders and record the new reference rather than arguing the point order by order.

Does a sprayed or coated colour need a different acceptance rule?

It needs an additional one. Because the colour sits on the surface, the average shade is only part of the question: uniformity across a single bottle also matters, since a gradient or a light patch is a defect in its own right rather than just a shade difference. Coatings also raise adhesion, abrasion and solvent resistance questions that have nothing to do with colour, and where the film is intended to remain on a food contact surface there is a separate food contact position to consider. The colour acceptance rule and the coating performance rule should therefore be written as two documents, even when they are checked in the same inspection.

How many bottles should be checked to judge a batch colour?

The number has to come from a written sampling plan rather than from convenience, because a lot with a small proportion of off-shade bottles will look perfect when two are examined and imperfect when fifty are. Public sampling schemes such as those described in ISO 2859-1, together with an agreed acceptance number, give both sides a defensible rule, and the inspection level and acceptance number should be agreed in writing. What matters in practice is that the same plan is applied to every delivery, so that a batch is judged on the bottle’s consistency rather than on the size of the sample someone happened to pull.

Can a colour difference be caused by something other than the glass?

Yes, and this is worth checking before blaming the bottle. A change in the liquid itself, which is common in natural or botanical products, will make the pack appear to have changed when the glass has not. A new closure lot, a different label substrate, a decoration ink batch or a curing shift can also alter the perceived colour of the finished pack. Where the complaint arises after filling rather than before, the liquid and the components should be checked alongside the glass, because a batch can be perfectly within tolerance and still look wrong once it is dressed and filled.

Send Your Light Source and Judgement Conditions, Colour Reference and Batch Size

The three inputs that make this page usable in your case are the viewing condition you intend to judge under, the colour reference you are working to, and the batch size you expect to run. Send those, and the reply can be specific rather than general: which route your colour is likely to follow and therefore which variables your tolerance has to protect, how a reference and a retained sample could be structured between your site and the factory, what the acceptance rule could look like in words and on a colour difference scale, and how a sampling plan would be applied to your production quantities.

A brand writing a colour specification for the first time should send the container reference, the shade they are aiming at, the light under which the product will actually be sold, and the filling and decoration arrangements, because those determine whether the colour is read in transmission, in reflection, or through a decorated layer. A buyer already in production and dealing with a disagreement should send the batch identity of the questioned delivery, a photograph taken under the agreed condition if one exists, the retained reference details from both sides, and the written allowance currently in force – which is usually enough to establish whether the problem is the bottles or the definition.