Rounding Peptide Test Results: Significant Figures on a COA

A procurement lead lines up the paperwork for three lots of the same compound. One peptide certificate of analysis reports purity as 99.1%, another as 99.08%, and a third as 99%. Are these lots different? Almost certainly not in any way the numbers can show. The differences come from how each result was rounded and how many digits the laboratory chose to print, and those reporting decisions are rarely explained on the document. For teams that buy in volume and compare results across many lots and vials, knowing how to read those digits prevents a lot of unnecessary queries and a few wrong conclusions.

Every printed digit is a claim about the method

A number on a certificate is not just a value; it is a statement about how finely that value was measured. Writing 99.1% asserts that the tenths digit is meaningful. Writing 99.08% asserts that the hundredths digit is too. The second figure is not a more detailed description of the peptide. It is a stronger claim about the precision of the test.

How many digits a result deserves depends on the repeatability of the method, not on what the data system can display. Chromatography software will calculate an area percentage to four decimal places without complaint, but replicate injections of the same solution typically wander by more than that. The spread described in measurement uncertainty for peptide purity is the real limit on how many places a purity can honestly carry.

Carry extra digits, then round once

Most avoidable disagreements between calculations come from rounding too early. The sound practice is to keep full precision through every intermediate step and apply rounding a single time, to the value that will be reported.

A simplified, purely illustrative example shows why. Suppose a lot has an HPLC purity of 98.64%, a water content of 6.2% and a counter-ion content of 11.4%, and a net content estimate multiplies the purity by the fraction that is neither water nor counter-ion:

• Full precision: 98.64 × (1 − 0.062 − 0.114) = 98.64 × 0.824 = 81.28, reported as 81.3%.
• Rounded at each step: 98.6 × (1 − 0.06 − 0.11) = 98.6 × 0.83 = 81.84, reported as 81.8%.

Same lot, same data, half a percentage point apart, purely because of where the rounding happened. The fuller treatment of this calculation is in net peptide content explained.

Two ways to round a five

Rounding is only ambiguous when the digit being dropped is exactly 5 and no further digits follow. Two conventions dominate:

Round half up is the method most people learned in school. Over a large set of numbers it nudges totals slightly upward. Round half to even (sometimes called banker’s rounding) sends ties to the nearest even digit, so the upward and downward cases balance out. Many statistical packages use it by default.

Neither is wrong. The problem is inconsistency: a lab that rounds one lot one way and the next lot the other way has created an apparent difference that has nothing to do with the material. There is also a quieter trap. Software stores decimals in binary, so a value displayed as 97.45 may actually be held as 97.4499999, which then rounds down under either rule. A laboratory that cares about the last digit will specify how its software handles this.

Comparing a result with its specification limit

Here rounding stops being cosmetic and starts deciding outcomes. The widely used convention is to round the observed result to match the decimal places written in the limit, then compare.

• Limit “not less than 95.0%”, result 94.96%: rounds to 95.0%, complies.
• Limit “not less than 95.0%”, result 94.94%: rounds to 94.9%, does not comply.
• Limit “not less than 95%”, result 94.6%: rounds to 95%, complies.
• Limit “not less than 95.0%”, result 94.6%: stays 94.6%, does not comply.

The last two rows show why the way a limit is written matters. “95%” and “95.0%” look alike on a quote, but they are different acceptance criteria. When your team writes a purchase specification, write the limit with the number of decimals you actually mean. The reasoning behind limits themselves is covered in how peptide specification limits are set.

Reading a peptide certificate of analysis: where extra digits mislead

Some entries on a certificate invite over-reading more than others:

• Area-percent purity. Where the integrator draws a baseline or splits a shoulder peak shifts the result far more than any third decimal place. See peak integration and purity.
• Observed mass. A theoretical mass can be calculated to many decimals. An observed mass deserves only as many as the instrument’s accuracy supports, which is why results are often judged in parts per million, as explained in mass accuracy and peptide identity.
• Tiny impurity values. A figure such as 0.02% from a method that can only quantify down to 0.05% is not a supported number. It should read as below the reporting threshold.
• Unit conversions. Turning 10 mg into 10,000 µg is exact arithmetic. Turning a measured 10.0 mg into 10,000.0 µg invents precision that never existed.

Trailing zeros and label quantities

A trailing zero after the decimal point carries information. “96%” says the result was determined to the nearest whole percent. “96.0%” says the tenths digit was measured and came out as zero. The same logic applies to quantities: a vial described as 10 mg and one described as 10.00 mg make different statements about how tightly the fill was controlled.

Zeros get dropped when they were earned (a spreadsheet trimming 96.0 to 96) and added when they were not (a template forcing two decimals on everything). When comparing documents across suppliers or across lots, check whether a missing or extra zero is a formatting habit before reading it as a change in resolution.

A quick checklist for comparing lots

When results for several lots sit side by side in your inventory records, work through these before raising a query:

1. Bring every result to the same number of decimals, using the coarsest precision among them.
2. Ask whether the difference that remains is larger than the method’s repeatability.
3. Check whether the limits were written with the same precision on each document.
4. Confirm the same rounding convention was used, if the laboratory states it.
5. Only then treat the difference as a real lot-to-lot change, and look at the causes set out in why purity results differ between certificates.

A difference that lives only in a digit neither method can resolve is a formatting choice. When the gap sits in a digit that both methods genuinely resolve, it deserves a follow-up. Telling the two apart usually takes a few seconds and saves an email thread.

All Bulk Peptides products are sold for laboratory research and analytical testing only and are not for human or veterinary use.