A stress result on a glass bottle only means something when four things are stated with it: the standard and grade scale, the viewing condition, the exact position on the bottle, and the wall thickness at that position. Residual stress is invisible, it causes breakage long after the bottle has left the plant, and it varies from bottle to bottle, so a batch has to be judged on its worst reading across a planned sample, never on one bottle or an average. The grade bands and patterns described here are orientation for writing a clause and reading a report; the figure that binds is the one in the standard your market names.
Why residual stress needs its own inspection
Residual stress is load the bottle places on itself. Part of it is left by forming, when a hot gob is pressed or blown and different areas of the wall stretch and cool unequally. More of it usually comes from cooling: the skin of the wall sets while the interior is still shrinking, leaving one layer in compression and the other in tension. The annealing lehr relaxes this by holding the bottle where the glass can still creep, then cooling it no faster than the glass can follow. Whatever remains afterwards stays in the bottle for good.
Nothing about it can be seen. Stones, blisters, checks, chips, a crooked finish or a thin side can each be caught by eye, gauge or camera at the cold end. A bottle with a heavy internal load can be clean, correctly dimensioned and free of every visible fault, because the load sits in the structure of a transparent material, not in a crack.
The damage arrives late. Glass breaks by slow crack growth: a tiny surface flaw lengthens under load, faster where moisture is present, until it is large enough to split the container. A permanent internal load keeps that growth going without pause, so a flaw that would sit harmlessly in a well annealed bottle keeps advancing in a poorly annealed one. The break can come minutes later on a conveyor, hours later in a warehouse or weeks later on a bathroom shelf, typically as a scatter of fragments with no bruise or contact mark. Cause and consequence are separated, which is how a batch that cleared every check at the plant can wreck a pallet somewhere else.
In service the internal load adds to whatever the outside world applies: the temperature step of washing or filling, carbonation pressure, vacuum under a lug lid, the weight of a stacked pallet, a knock in handling. The usable margin is the strength of the glass minus the sum of both. A bottle with hidden stress can therefore pass a thermal shock or impact test one day and fail it the next, and reading the residual state tells you how much margin every other test has to work with.
What a stress clause has to state
Stressed glass is birefringent. Light passing through it splits into two components that travel at slightly different speeds, and when they recombine they are out of step by a distance called the retardation. That distance is proportional to the stress summed along the path the light took. A stress measurement on a container is therefore a path-difference measurement, usually given in nanometres and often divided by wall thickness so that light and heavy bottles can be compared.
Turning what is seen into a figure that someone else can check requires four items on the record. Write them into a single sentence of the specification before the first bottle is read.
- Standard and scale. A grade is a band on a scale, and the scale belongs to whichever standard buyer and plant have agreed. Standards split the same physical range into different numbers of grades with different boundaries, and some grade body and edge separately while others grade the whole container. A report that says "grade 3" with no scale named proves nothing, and a maximum grade with no standard or method behind it cannot be enforced.
- Viewing condition. A plain dark field, crossed polarisers with a full-wave tint plate, and a single-wavelength reading with a compensator show different things. A colour in a white-light viewer is a band position; the same bottle under monochromatic light with a calibrated compensator gives nanometres. Record the configuration, the light source and whether a tint plate was in the path, or two honest laboratories will disagree about one bottle.
- Reading position. Because retardation accumulates along the light path, a reading through two walls cannot be set against a reading through one, and a heel reading cannot be set against a mid-body one. Name positions so another person can find them: mid-body on the side wall opposite the mould seam, the shoulder knuckle, through the base, the neck just under the finish.
- Wall thickness there. Retardation per unit of glass thickness is the figure that travels best between plants, since a thick base and a thin body wall cannot share one absolute limit. Where the chosen standard works in absolute retardation, note the thickness anyway so the reading can be reinterpreted later.
Where on the bottle to read
Capacity belongs to a bottle; a stress reading belongs to a position. One container gives a different answer at each of the points below, and accepting one of them as the whole story is the usual reason a tidy report sits alongside a batch that breaks.
| Position | Why it is read | What the reading tells you |
|---|---|---|
| Base and heel | The thickest glass, slowest to change temperature, so more of its temperature gradient survives into the annealing range and becomes permanent stress | Shows a badly set cooling curve first; the position that matters for stacked containers |
| Shoulder | The wall turns a corner, thickness is uneven, and glass inside the turn is treated differently from glass outside it | Often the highest reading on the bottle, and where a surface crack most often runs, which makes it a sensible place for a limit |
| Body side wall | The thin section that a thermal shock or pressure load actually tests | Normally the lowest reading; reported alone it answers a thermal question, not a delayed-fracture one |
| Neck and finish | The most complex geometry and thinnest glass, loaded by the capping head, the closure and the user's hand | Frequently specified apart from the body; in some markets edge stress is the regulated figure or the one a filler asks for |
| Mould seam | Glass can carry a local concentration where the mould halves meet | A pattern running down the seam from the shoulder says the two halves are not behaving alike |
Comparing a seam reading with one from the opposite side of the body is an inexpensive way to tell a whole-container annealing problem from a forming problem tied to one cavity.
Two working rules come out of this. Choose a short list of named positions, put it in the clause and read that same list every time, so results months apart can be compared. And do not blend positions into one figure: the mean of a high base and a low body describes no part of the bottle and buries the spot that will fail. If a report must carry a single number, make it the worst position, with the position written beside it.
Polariscope or strain gauge
Stress reports rely on two kinds of instrument. They differ in scope, not in accuracy, and many arguments start from treating one as the other.
A polariscope, also called a polarised viewer or strain viewer, puts the bottle between two polarising filters whose axes are set at right angles. Unstressed glass stays dark. Stressed glass rotates the polarisation, light leaks through the second filter, and the region shows bright and coloured, with the colours following the order of the interference bands as retardation climbs. A trained reader places the bottle on a scale by that order, not by the hue alone. Adding a full-wave tint plate shifts the whole sequence, so retardations too small to register in a dark field appear as a colour change against an even background; this is the set-up for judging the low end of the scale instead of merely spotting a gross fault.
The viewer's strength is distribution. A band down one side, a hot spot at the heel, a seam pattern, or one bottle out of twelve that looks unlike the others are all obvious at a glance. It is quick and portable, which suits routine sorting and go or no-go decisions against a reference. Its output, though, is a picture and a judgement. A figure taken off a photograph through a viewer is an interpretation and will not resolve a disagreement between two plants.
A strain gauge, or bench polarimeter with a compensator, examines one small area and returns a number: retardation in nanometres through a defined path at a defined point, and with the thickness a normalised figure. It can be quoted, repeated by a third party and tracked against lehr records over time. It is also slow, needs the bottle positioned and the wall measured, and is silent about every position nobody chose to read. A gauge applied only to the body can certify the body while saying nothing about a base that will crack in storage.
Reading the pattern: what each condition means
The conditions below are the ones readers actually meet. They are described in words and in order of severity, without grade numbers, because the count of grades and their limits change from one standard to another. Treat the table as a guide to reading a report or briefing a laboratory; it does not stand in for the standard you invoke.
| Condition and view | What is seen and recorded | Meaning for annealing and where it peaks | Consequence for the buyer | Confirm with the plant |
|---|---|---|---|---|
| Lowest grade, dark field, no tint plate | An even dark field over the body, at most a faint, broken brightening on the side wall and near the heel. Record that no continuous band exists and which positions were examined. | The lehr worked; the bottle is near stress free. No peak anywhere, which is the point. | Widest margin for applied loads. Expect this of containers meant for hot filling, repeated washing or pressure duty. | Standard and scale, positions examined, light source, lehr record for the shift that made the sample. |
| Lowest grade, with full-wave tint plate | An even tinted background with a faint but continuous hue shift in one region. Record the shift and where it sits. | The low end has been judged, not assumed. Commonly the base and heel knuckle, where glass is thickest and cools last; sometimes a thin line on the shoulder where the wall changes section. | The right set-up for qualification, since it separates a well annealed bottle from a nearly well annealed one that a plain view cannot tell apart. | Whether a tint plate was used (reports with and without are not comparable) and whether a compensator figure was taken. |
| Intermediate grade, no tint plate | A continuous bright band along part of the side wall or shoulder, with a clear colour order. Record the highest band position and the length of the band. | Annealing incomplete without being a gross fault: hold or cooling path was insufficient for part of the bottle, or the lehr was heavily loaded there. Peaks at the shoulder and upper body; a band on one side only suggests uneven cooling, not the lehr overall. | Still serviceable for ordinary ambient-fill cosmetic or beverage use, but with less margin for heat or pressure. Do not qualify the batch against an endurance claim without further testing. | Band on one side or all round, mould cavity identity of the samples, belt speed and loading pattern for the run. |
| Intermediate grade, tint plate view of base and heel | A concentrated pattern in the base and heel knuckle, often with a sharp hue boundary where the heel radius begins. Record the position and extent of the zone, not a whole-bottle average. | The thickest glass cooled ahead of or behind the wall, so a steep gradient persisted across the base well into the annealing range. Peaks at the base, heel radius and the inside corner where body meets base. | The pattern most often linked to delayed breakage in storage, because stacking load lands on a base already carrying internal load. | Whether that cavity's mould differs in base thickness, the gob weight variation, and whether the first lehr zones are cooling the base faster than the wall. |
| Intermediate grade, concentrated at finish and neck | A bright zone just under the finish and a haze or band across the neck, often reaching into the shoulder. Record neck position, band height and whether the finish itself shows a pattern. | The neck was already too cool on entering the lehr for relaxation to work, or the first zone pulled heat out faster than the neck could follow. Can be lopsided if the bottle leaned on a neighbour on the belt. | A capping and handling risk more than a thermal or pressure one: this is where concentrated loads land and where chips and neck fractures start. | The capping load expected, whether the target market specifies a separate finish grade, and how bottles were spaced on the belt. |
| Highest grade, strong high-order colours in a dark field | A dense pattern with repeating colour cycles over body and base that changes quickly as the bottle is moved slightly. Record the description, the number of colour orders, and that the reading is beyond the agreed scale. | Effectively unannealed: the bottle left the lehr too early or never went through it. Affects the whole container, most strongly the thickest section. | Unfit for any use involving thermal or mechanical load; a large share will fail by delayed fracture with no outside cause. | Lehr record and belt speed, whether the bottles were on the belt at all, and how the affected batch is segregated and disposed of. |
| Localised, along a mould seam or at a contact point | A narrow bright line following the seam downward from the shoulder, or a small bright spot at a fixed contact. Record seam position relative to the mould halves and the location of any spot. | Annealing may be broadly acceptable; forming has left a local concentration. Found on the body seam, base seam and wherever the hot bottle was held, pushed or rolled. | A forming and handling finding, so the remedy is at the machine, not the lehr. Start the investigation there. | Mould condition, swabbing or lubrication practice, take-out and transfer equipment, and whether the pattern moves when only the cavity changes. |
| Quantitative, monochromatic light with compensator or bench strain gauge | Nanometres at a named position and thickness, converted to a per-thickness figure. Record the number, wavelength, position, thickness and instrument. | The annealing state becomes a figure comparable across plants, batches and dates. Read wherever the instrument is set: usually the side wall for a body figure, base or heel for a critical one. | The only form of result that holds up in a commercial dispute, since both parties can repeat it on retained samples. | Instrument and calibration, wavelength, exact positions, thickness at each, and the scale the numbers refer to. |
Judging a batch by its distribution
No two bottles in a run are identical. Gob weight drifts shot to shot, cavities wear unevenly, and a bottle's place on the lehr belt governs how much heat it sheds at each stage, with the belt edges generally cooling differently from the centre. A batch has a spread of stress values, and the top of that spread is what fails.
One reading on one bottle is a spot check on that bottle. The odds of having picked the worst in the run are small, so the report flatters the batch. An average has the same weakness: the mean of ten bottles can sit comfortably within a limit while one of the ten is far outside, and that one, multiplied across a shipment, is the source of the breakages. For a batch, quote the worst reading and the spread.
A sound batch decision rests on a sampling plan with four parts:
- Sample size large enough that a single bad cavity has a realistic chance of turning up.
- Selection rule, the part most often absent: draw across the mould cavities, across the width of the belt, and from the start, middle and end of the run, not from one handy tray.
- Statistic: whether acceptance applies to every bottle tested, to the sample maximum, or to the share of the sample over a limit.
- Acceptance rule, best tied to a recognised sampling scheme with an acceptance quality limit matched to what a failure costs. A bottle breaking in a bathroom and one breaking in a filling hall are different commercial events.
Six ways a stress report becomes unusable
Each of these turns an honest measurement into one nobody can rely on.
- No position. A maximum with no location cannot be checked, repeated or compared with a specification, and suggests the spot was picked after the reading.
- No standard or scale. A bare grade number is a label. A clause built on it collapses the first time two laboratories compare results.
- No viewing condition. Dark field, tint plate and compensator readings differ, so a report that omits the set-up cannot be reproduced even by its author.
- Bottle still warm. Straight off the lehr the bottle is cooling, and the viewer is showing part of a temperature gradient, not the permanent state. Hot readings can look dramatic and mean nothing; a line assessed on them has been assessed on a transient.
- Only the easy position. The body wall gives the cleanest reading, which is why body-only reports miss the base pattern that carries the real risk. Fix the positions in advance.
- One pass treated as permanent. A new mould, a different gob weight or wall thickness, a changed belt speed or furnace load, or a change in decoration firing all alter the residual state. Name the triggers for a re-test in the specification.
Tracing an out-of-range reading back through the line
A high reading is a finding. The sequence in which causes are ruled out saves more time than any single technique.
Check the measurement first
It is the cheapest thing to get wrong and is wrong more often than expected. Read the bottle again at room temperature with the light source checked, the polarisers confirmed crossed, the correct tint plate for the question, and the position freshly marked. Ask a second person to read it blind. Agreement means the finding is real.
Map the distribution before changing any setting
Its shape points to the cause. High readings from a single cavity implicate that cavity or how its bottles are formed and transferred. High readings from one side of the belt, or one place along it, implicate the lehr and the spacing on it. Readings high across the whole run implicate lehr settings or furnace condition for the shift. Readings high at one position on every bottle, such as base or shoulder, implicate local thickness or geometry.
Cavity-specific findings
Work down the forming side: mould condition, including wear at base and seam; gob weight and its variation, since a heavier gob puts extra glass in the thick sections; parison and blank-side conditions, including blank mould and plunger temperature; swabbing and lubrication, which affect how the glass slides and so how thickness is distributed; then take-out and transfer equipment, where a fixed contact can print a local pattern. The quickest confirmation is to compare the suspect cavity with the same position on neighbouring cavities fed from the same gob and passed through the same lehr.
Run-wide findings
Work through the lehr: belt speed, then loading and spacing, then zone temperatures, then the shape of the cooling curve within the annealing range. The hold lets stress relax and the controlled cooling afterwards stops new stress forming, so a curve with correct temperatures but too fast a final stage produces bottles that looked fine at the entrance and are stressed at the exit. An overloaded belt has the same effect by another route: tightly packed bottles shield one another and alter local cooling, which is why one curve can give good bottles on a light load and poor ones on a heavy load.
What the buyer can require and what the plant controls
Inside the lehr the bottle is raised to a temperature at which glass deforms slowly, held until internal strains have largely relaxed, cooled through the range where the glass stiffens, and then cooled slowly enough that the gradient across the wall does not lock in new load. Annealing point, strain point, hold time and cooling rates are process variables, and the plant owns all of them. The stress reading is the symptom of how well that process worked on a given bottle on a given shift.
For a buyer this has practical effects. A grade written into a purchase order is a required outcome; the plant owns the curve that delivers it. A competent supplier responds to a stress question by asking about the lehr record, belt speed and load for the run, because those are the levers. Be wary of one who treats the reading as adjustable from the measuring side, by picking a kinder position or a more forgiving light source. Keep the reading and the process record together so a good result can be repeated.
Three pattern-to-process links are worth knowing by heart:
- A local pattern at base and heel usually means the base was cooled too quickly early in the lehr, before the wall had relaxed. Look at the early zones and belt load, not the final cooling stages.
- A broad band along body or shoulder usually means the hold was too short, or the bottle was already too far into the annealing range when it entered. Look at the hold, belt speed or loading.
- A pattern gathered at the neck usually means the neck went in cool, leaving the first zone to do too much on a part that cannot take it.
Records and supplier evidence
Routine records exist so a result can be repeated; in a claim they have to serve as proof, and a report good enough for the first job rarely suffices for the second. The extra items cost little when the reading is taken and cannot be rebuilt later.
A record that will stand up carries:
- the standard and grade scale invoked;
- the light source and whether a tint plate was used;
- the instrument or viewer and its calibration status;
- the positions read, by diagram or unambiguous wording, with wall thickness at each;
- the readings with date, shift, furnace and mould cavity identity;
- the sampling plan, sample size, acceptance rule and batch disposition.
File the lehr record for the same run with it, since without that the reading can be neither explained nor repeated. Mark any retained samples so they tie back to the record; an unlabelled bottle proves nothing.
When bottles come from an outside supplier, that list becomes what you request, ideally before a repeat order and not after a failure. Ask for a report naming standard and scale, the positions and the number of bottles read, whether the sample spanned the cavities and lehr positions, and whether a viewer or a quantitative instrument produced the result. Ask too for a retained sample or a few bottles from the same run so the reading can be repeated independently. A viewer photograph is worth accepting as supporting evidence, with the quantitative figure requested alongside, because no third party will take a photograph as a grade.
Two items belong in the commercial documents. One is change control: any change of mould, wall weight, finish, furnace or lehr settings triggers a re-qualification reading, stated in advance. The other is evidence format, meaning what the supplier must provide and in what form. A maximum grade tied to a named standard and named positions can be enforced; "good annealing" cannot. Where the destination market regulates finish stress and body stress separately, structure the clause the same way.
Checks that stress inspection does not replace
Stress grading covers one failure. Thermal weakness, wrong dimensions and fragility under impact are three others, each with its own measurement, and a specification that folds them together ends up as one vague requirement.
- Annealing as a process. How the lehr curve is run and how an annealing verification is specified as a method is a separate subject. It explains how the check is set up; stress inspection grades the outcome. The two are meant to be used together.
- Thermal shock. The temperature difference a container survives has its own mechanism and apparatus. Residual stress limits the margin available for it without measuring it, and a well annealed bottle can still be wrong for hot fill. See the thermal shock test method for glass bottles.
- Dimensions. Capacity, height, diameter, finish dimensions, squareness and wall thickness each have tolerances and gauges of their own. A bottle can be dimensionally perfect and badly stressed, or the reverse. The framework is set out under tolerance standards for glass bottles.
- Impact. Behaviour when dropped or struck under a controlled method shares the theme of delayed fracture but not the measurement; it is covered in the glass bottle drop test guide.
- Fill volume. How much goes into the container and what tolerance applies to the labelled quantity is a metrology question about the fill. A perfectly filled bottle can be badly annealed, and a well annealed one can be overfilled. The two meet only in the quality file, each with its standard named.
- Automated empty-bottle inspection. Line equipment that hunts for checks, stones, blisters and foreign bodies works on visible and physical features, so it cannot report residual stress. What it does catch is described under glass bottle inspection machines.
Capacity and dispensing form a different decision chain altogether. Matching a volume band to a dispensing method is handled in our guide to body wash bottle sizes, and nothing there determines how a bottle should be inspected for stress.
What to send us to set up a stress requirement
Three items let us turn a stress requirement into something a plant and a laboratory can both work to: the bottle shape or drawing with capacity and finish, the finished weight of the empty container, and the wall thickness at the positions that concern you, or a thickness map if you have one. Add the standard and grade scale your market calls for, the positions you want read, and whether the bottle is single-trip or will be washed and refilled, since a returnable container needs a different margin.
Tell us also what is driving the request: a stacking load, a thermal step, chipping at the finish, or breakage after the goods have shipped. That decides which positions carry most weight. From there we can set out the grading convention for your market, the positions, the viewing condition and instrument that give a figure comparable with your specification, a sampling plan spanning cavities and lehr positions, the acceptance rule for a batch, and which changes to mould, weight or lehr call for a re-test.
Questions buyers ask about glass bottle stress inspection
What is residual stress in a glass bottle?
A permanent load built into the glass by forming and cooling, as opposed to one applied by the contents, the closure or the pallet. The lehr removes most of it; the remainder is carried for the life of the bottle and cannot be seen without polarised light.
Why does an over-stressed bottle break later and not straight away?
Surface flaws grow slowly under load, especially in damp conditions, and a permanent internal load never lets that growth stop. Failure comes when a flaw reaches critical size, which may be minutes, hours or weeks after manufacture, usually with no external trigger and no visible damage at the origin.
What is retardation and how does it become a grade?
It is the path difference between the two light components that stressed glass produces, proportional to stress along the light path and usually quoted in nanometres, often per unit of wall thickness. The grade is the band that figure falls into on the scale of the standard invoked, so standard, scale, position and viewing condition have to be named together.
Should I ask for a polariscope or a strain gauge reading?
A polariscope answers where the stress lies and whether this batch resembles the last. A strain gauge or calibrated polarimeter gives a number at a defined position that can go into a specification and be repeated by another party. Many buyers need both: the viewer to find the pattern, the gauge to quantify the positions that matter.
Which part of the bottle should be measured?
A fixed set agreed in advance: base and heel, shoulder, body side wall, neck and finish, plus the seam compared against the opposite side of the body. Report the worst of them with its position named.
How many bottles are needed to judge a batch?
More than one, and how they are chosen matters more than the count. Draw across cavities, across the belt width and through the run, then quote the worst reading and the spread, because the tail of the distribution is what fails in the field.
Does a good stress result mean the bottle will survive thermal shock?
No. Low residual stress leaves more margin for an applied load, but the temperature difference a bottle survives is measured in a separate test and also depends on wall thickness distribution and surface condition. Read a good grade as wider margin, not as a thermal rating.
What evidence should a supplier provide?
A test report naming the standard and scale, the positions, the wall thickness at each, the number of bottles read and the type of instrument, together with lehr and run details for the same production. A retained sample from that run and re-qualification triggers written into the specification complete the file.