Every datasheet number is either a measurement or a promise, and the promise--Analytical Testing Vials

A 0.45 µm PTFE syringe filter datasheet said “aqueous and organic compatible.” The lab ran 84 samples through a single lot before the third analyst noticed the recovery on polar pesticides had dropped 22% from the morning run. The datasheet never stated the filter housing was polypropylene, not nylon. The nominal pore size was real. The wetted material claim was not. If you cannot separate a tested number from a marketing number on an analytical testing datasheet, you will eventually build a method on a number that was never measured. This guide covers analytical testing datasheet specifications and what each value means, which lines are guaranteed by a test method, which are nominal design targets, and how to spot a sheet that hides its test method entirely.

Every datasheet number is either a measurement or a promise, and the promise lines are usually unlabeled

I have spent 15 years buying and selling chromatography consumables, and I can count on one hand the number of suppliers who mark which specifications are lot-tested and which are printed from the CAD file. A datasheet is a legal document only in the narrow sense that it commits the supplier to replace product that fails a stated guarantee. The problem is that most lines on the sheet are not guarantees. They are design targets, typical values, or historical averages. The supplier never measured them on your lot, and they never will.

The practical distinction is simple. A tested value is one where the supplier can show you the measurement method, the acceptance limit, and the lot-specific result. A nominal value is a dimension or property that comes from the tooling drawing, the resin grade, or the design specification. It was true once, on a golden batch or a mold validation run. It may still be true. It may not be.

When a lab manager tells me a lot of vials “suddenly” failed, my first question is always the same: which number on the sheet changed? The answer is almost always that no number changed. The sheet stayed identical because the supplier never re-tested the parameter that actually drifted. The mold wore. The resin lot changed. The annealing cycle shortened. The datasheet did not move because it was never tied to those variables in the first place.

This is not a supplier conspiracy. It is economics. Testing costs money. A supplier who tests every parameter on every lot would price themselves out of the market. The honest ones tell you which parameters they test and which they inherit from design. The dishonest ones print a wall of numbers with no method references and let you assume they are all equal. Your job as a buyer is to force that distinction into the open before you commit to a 500-unit order.

The first line to check is not the value, it is whether a test method exists at all

A specification without a method reference is a rumor. I do not care if the number is 0.22 µm or 6.5 mm or 350 °C. If there is no ASTM, ISO, USP, EPA, or internal method code attached to it, the value is a nominal claim. You cannot audit it, you cannot reproduce it, and you cannot hold the supplier to it in a dispute.

Here is the test I run on every new datasheet. I take a red pen and circle every line that has a number but no method. Then I circle every line that says “typical” or “nominal” or “reference value.” Then I circle every line that gives a range but no tolerance. The uncircled lines are the only ones I can actually rely on. On most consumable datasheets, that leaves me with fewer than half the printed specifications.

Look at a typical HPLC vial sheet. You will see a 12 × 32 mm dimension, a 9 mm screw thread, a borosilicate glass type, and a write-on patch description. The 12 × 32 mm dimension is often listed with a tolerance of ±0.25 mm, and the supplier can show you a gauge drawing. The 9 mm thread is a GPI finish, which is a dimensional standard, so it is testable with a go/no-go gauge. But the “Type I borosilicate” claim is only testable if the supplier runs a hydrolytic resistance test per USP <660> or ISO 4802. Most do not run that test on every lot. They buy certified glass tubing and assume the conversion preserved the chemistry. That assumption is usually right, but it is still an assumption.

Thermo Fisher publishes a lot of method-referenced data for their chromatography and sample handling lines, and Agilent does the same for GC columns and accessories. When you see a value on their sheets with an ASTM or USP method code, you can treat it as a measurement. When you see a value with no code, you should treat it as a design target until proven otherwise.

Which numbers on a vial datasheet are actually measured, and which are nominal

Let me walk through a real 2 mL HPLC vial datasheet line by line, because this is the product most labs buy in volume and the one where bad assumptions cost the most.

Vial outer diameter (12.0 ± 0.25 mm): This is a tested dimension, or it should be. The supplier checks it with a laser micrometer or a dial gauge on a sample from each cavity. The tolerance matters because the autosampler tray pocket is machined to a fixed diameter. A vial that runs 12.3 mm will jam in a tray that expects 12.2 mm. Ask the supplier how many cavities they sample per lot. If they sample one cavity out of 48, the tolerance on the sheet does not protect you from the other 47.

Vial height (32.0 ± 0.25 mm): Same story. Tested with a height gauge. The tolerance matters for needle penetration depth. A vial that is 0.5 mm tall will push the needle tip into the bottom and bend it. A vial that is 0.5 mm short will leave the needle 0.5 mm above the liquid level and draw air.

Glass type (Type I borosilicate): This is nominal unless the supplier runs USP <660> or ISO 4802 hydrolytic resistance testing on the finished vial. The distinction matters for pH-sensitive analytes. Type I glass leaches less alkali than soda-lime, but the leaching rate depends on the forming process, the annealing temperature, and the surface treatment. A vial blown too hot can leach more than a properly annealed soda-lime vial. The glass type on the sheet is a starting point, not a guarantee.

Septum material (PTFE/silicone, PTFE/butyl): This is a tested composition in the sense that the supplier knows what they laminated. But the functional properties you care about, resealability, coring resistance, extractables, are separate tests. A PTFE/silicone septum has a puncture count rating that depends on the needle gauge, the insertion angle, and the silicone durometer. If the datasheet says “multi-puncture” without specifying the needle gauge and the number of punctures, it is a nominal claim. A real septum datasheet says something like “reseals after 3 punctures with a 23G needle at 90° entry.”

Cap torque rating: If the datasheet lists a recommended torque, usually 8 to 12 inch-pounds for a 9 mm cap, that is a design value derived from the thread geometry and the septum compression set. It is not a tested value on your lot. The supplier cannot test torque on every cap. What they can test is the thread form with a go/no-go gauge. The torque value is a recommendation, not a specification.

Residual solvent or cleanliness level: If the datasheet says “certified low extractables” or “LC/MS grade,” you need to ask which test was run. A real claim references a specific protocol: total organic carbon after 24-hour water extraction, or GC/MS scan for specific compounds, or a certificate of analysis with lot-specific data. A claim with no protocol is a marketing phrase. I have seen “LC/MS clean” vials that leached enough phthalate to swamp a 10 ppb detection limit.

Filter datasheets are where the nominal numbers hide in plain sight

Syringe filters and membrane filters are worse than vials because the functional specification, the pore size, is a statistical distribution, not a single number. A 0.45 µm filter does not have pores that are all 0.45 µm. It has a pore size distribution with a mean or a maximum around 0.45 µm, depending on the membrane type and the manufacturer’s definition.

Nominal pore size vs. absolute pore size: This is the single most misunderstood line on any filter datasheet. A nominal 0.45 µm filter retains most particles at or above 0.45 µm, but it will pass some fraction of larger particles. An absolute 0.45 µm filter retains essentially all particles at or above that size, often with a 99.9% or 99.98% efficiency at a specified challenge particle size. Many suppliers print “0.45 µm” without saying which definition they use. That is a nominal claim hiding as a tested specification.

Bubble point: This is a real test. The supplier wets the membrane, applies air pressure, and records the pressure at which the first bubble appears. The bubble point correlates with the largest pore in the membrane. A 0.45 µm PTFE membrane typically has a bubble point around 0.9 to 1.2 bar in isopropanol. If the datasheet lists a bubble point with a pressure value and a wetting solvent, that is a tested number. If it lists only “0.45 µm” with no bubble point, you have no way to verify the pore size claim.

Membrane diameter and housing dimensions: These are tested with calipers or optical comparators. The tolerance matters because the filter must seal in the housing. A membrane that is 0.2 mm undersized will bypass fluid around the edge. A housing that is 0.1 mm over tolerance will leak at the luer fitting.

Housing material: This is where I have seen the most expensive failures. A datasheet may say “PTFE membrane” in large type, which is true, and “polypropylene housing” in small type, which the buyer misses. Polypropylene is not compatible with some solvents, and it can leach oligomers into the sample. The housing material is a nominal property from the resin grade, but its compatibility with your solvent is a test you need to run, not a number you can read off the sheet.

Extractable level: A real filter datasheet for LC/MS work lists the extractables in µg per filter or µg per cm² of membrane area, with the extraction solvent and time specified. A nominal datasheet says “low extractables” and stops. The difference is the difference between a method that passes an audit and one that fails it.

Restek publishes technical resources that explain how they test their chromatography consumables, and their approach to filter specifications is a good model for what a transparent datasheet looks like. They state the test method, the acceptance limit, and the lot release criterion. When you compare suppliers, compare those three lines, not the marketing copy above them.

The certificate of analysis is the only part of the datasheet that proves a number was measured

A datasheet is a general document. A certificate of analysis (CoA) is a lot-specific document. The difference is the entire game. The datasheet tells you what the product should be. The CoA tells you what the lot you actually received was, measured on that lot, by a specific method, with a specific result.

When I evaluate a new supplier, I ask for the CoA before I place the order. Not after. If the supplier cannot produce a CoA for the lot they are about to ship, I assume every number on the datasheet is nominal. That is not cynicism. It is the only rational position when you have no lot-specific data.

A real CoA for a chromatography consumable lists at least these fields: lot number, date of manufacture, test method reference, specification limit, and measured result. If the CoA lists only “conforms” without a numeric result, it is a certificate of conformance, not a certificate of analysis. The distinction matters. A certificate of conformance means someone checked a box. A certificate of analysis means someone ran a test and wrote down the number.

I once rejected a shipment of 10,000 HPLC vials because the CoA listed the outer diameter as 12.0 mm with a tolerance of ±0.25 mm, but the measured value was 12.28 mm on the sample they tested. That is within tolerance, but it is 0.28 mm from nominal, and my autosampler tray pockets were machined to 12.20 mm. The vials would have jammed intermittently. The CoA saved me from a week of troubleshooting a “random” autosampler fault. The datasheet alone would not have.

Two field cases where a nominal number masqueraded as a tested one

Case 1: The septum that resealed on paper but not in the autosampler. A contract lab running 400 injections per day on a single 2 mL vial position started seeing carryover spikes on the third injection from the same vial. The datasheet said the septum was “multi-puncture rated.” The CoA said nothing about puncture testing. When I called the supplier, the technical rep admitted the multi-puncture rating was based on a 22G needle at 90° entry, but the lab was using a 26G cone-tip needle at a 12° entry angle. The cone tip was tearing a crescent-shaped hole instead of a clean cylinder. The fix was not a new septum material. It was a new specification: the lab now requires the supplier to state the puncture test needle gauge, entry angle, and number of resealed punctures on the CoA. The supplier who would not provide that data lost the account.

Case 2: The filter that passed the bubble point test but failed the application. A pharmaceutical QC lab validated a dissolution method using a 0.45 µm PTFE syringe filter. The datasheet listed a bubble point of 1.0 bar in isopropanol, and the CoA confirmed it. The method ran fine for six months, then recovery on the active pharmaceutical ingredient dropped 15%. The filter lot had changed. The new lot still passed the bubble point test, but the pore size distribution had shifted toward the fine end, and the filter was now adsorbing the analyte. The datasheet had not changed. The CoA had not changed. The only thing that changed was the supplier’s membrane casting process, and no one had tested analyte recovery on the new lot. The fix was to add a recovery check to the method using a spiked standard through each new filter lot. The datasheet could not have predicted this failure because the datasheet never claimed to test analyte recovery.

Both cases share the same root cause. The buyer trusted a number that was never tied to the failure mode that actually mattered. The septum puncture rating was nominal. The filter pore size distribution was nominal. The datasheet looked authoritative because it had numbers with tolerances, but the tolerances applied to parameters that did not predict the failure.

A decision checklist for reading any analytical testing datasheet

Here is the checklist I use when I sit down with a new datasheet. It takes about ten minutes, and it has saved me more money than any negotiation tactic I have ever tried.

Circle every number with no test method reference. Treat those as nominal design values. Do not build a method on them.
Circle every number with no tolerance or acceptance limit. A value with no tolerance is a target, not a specification.
Ask for the CoA for the specific lot before ordering. If the CoA says “conforms” without numeric results, downgrade the datasheet to marketing material.
For vials, confirm the dimensional tolerance is tested per cavity, not per lot. A 48-cavity mold can produce 47 bad cavities and one good one that passes the lot test.
For septa, require the puncture rating to state needle gauge, entry angle, and number of resealed punctures. A “multi-puncture” claim with no test conditions is worthless.
For filters, confirm whether the pore size is nominal or absolute. If the datasheet does not say, ask for the bubble point value and the wetting solvent.
For any “low extractables” or “LC/MS grade” claim, require the extraction protocol and the lot-specific result. A phrase is not a measurement.
Check the housing or body material on every filter and vial. The wetted material is what touches your sample, not the membrane or the glass type.
For any value you will use in a validated method, ask the supplier to put the test method and acceptance limit in writing on the CoA. If they will not, find another supplier.

What a transparent datasheet looks like, line by line

Let me show you the difference between a nominal datasheet and a tested one, using a 2 mL screw-top HPLC vial with a 9 mm cap and PTFE/silicone septum as the example.

Nominal datasheet:

12 × 32 mm clear glass vial
Type I borosilicate
9 mm screw cap with PTFE/silicone septum
Multi-puncture septum
Low extractables
Compatible with all major autosamplers

Tested datasheet:

Vial OD: 12.0 ± 0.25 mm, measured per ASTM E691 on 5 cavities per lot
Vial height: 32.0 ± 0.25 mm, measured with a calibrated height gauge
Glass type: Type I borosilicate, hydrolytic resistance per USP <660>, lot-specific result on CoA
Septum: 1.0 mm PTFE / 1.5 mm silicone, durometer 45 Shore A, reseals after 3 punctures with 23G needle at 90° entry, tested per lot
Extractables: <0.1 µg total organic carbon per vial after 24-hour water extraction at 60 °C, tested per lot
Autosampler compatibility: verified in Agilent 1100/1200/1260, Waters 2695, Shimadzu SIL-20A tray pockets, test report available

The second sheet tells you what was measured, how it was measured, and what the limit was. The first sheet tells you what the product is supposed to be. The price difference between the two suppliers is often less than 10%. The cost of a failed run is measured in hours of analyst time, re-injection, and audit exposure. Pay for the tested sheet.

When you are comparing suppliers across a lab vials category, do not compare the marketing copy. Compare the method references. A supplier who prints “ASTM E691” on their dimensional check is telling you they run a real measurement system. A supplier who prints “12 × 32 mm” with no tolerance is telling you they copied a drawing. The difference shows up in your autosampler, not in the catalog.

Method references on a datasheet are a proxy for the supplier’s quality system

A supplier who cites EPA SW-846 methods for their sample prep consumables, or USP <660> for their glass, or ISO 4802 for hydrolytic resistance, is running a quality system that generates data. A supplier who cites nothing is running a quality system that generates invoices. That is not a fair statement in every case, but it is fair enough to use as a screening rule.

The reason method references matter is not that you will ever run USP <660> yourself. You will not. The reason they matter is that a method reference forces the supplier to define the test, the equipment, the sample size, and the acceptance limit. That definition is what makes a number auditable. An unauditable number is a rumor, and rumors do not hold up in a regulatory inspection or a supplier dispute.

When I audit a supplier, I ask to see the raw data behind the CoA. Not the summary. The raw data. A supplier who can show me the bubble point curve, the dimensional histogram, and the puncture test log for a specific lot is a supplier I can trust with a validated method. A supplier who shows me a signed certificate with no numbers is a supplier I can replace.

The same logic applies to your own lab. When you write a method, cite the consumable specification that matters, not the one that is easy to copy. If your method requires a 0.45 µm filter, specify the bubble point range and the membrane material, not just the pore size. If your method requires a 9 mm cap, specify the torque range and the septum reseal rating, not just the thread size. The more specific your method, the harder it is for a supplier to substitute a nominal product for a tested one without you noticing.

The decision rule

Here is the rule I give every lab manager who asks me how to buy consumables without getting burned. A number on an analytical testing datasheet is only real if three things are true: the supplier states the test method, the supplier states the acceptance limit, and the CoA for your specific lot shows a numeric result within that limit. Every other number on the sheet is a design target, a typical value, or a hope. Buy the product that proves its numbers, not the one that prints the most of them. The datasheet with 40 tested lines beats the datasheet with 80 nominal lines every time.

References & further reading

Related resources from our lab consumables network:

External references and standards: