Two spectrophotometers can list the same 200–900 nm scan range and look interchangeable on a quote sheet, yet produce data that refuse to agree when a method is transferred from one lab to another. The difference often hides in two unglamorous lines of the specification table: wavelength accuracy and spectral bandwidth. Neither number makes a headline, and neither is easy to interpret without knowing how it was determined. Together, though, they govern where a peak actually lands, how tall it appears, and how confidently you can compare results across instruments, sites, and years. Before you place spec sheets side by side, it pays to know which questions separate a meaningful number from a decorative one.
What Wavelength Accuracy Really Measures
Wavelength accuracy describes how close the wavelength an instrument reports is to the wavelength of light that actually reaches the sample. It is usually quoted as a tolerance — for example, ± 0.75 nm, which is the figure Implen publishes for the NanoPhotometer® NP80, alongside a wavelength reproducibility of ± 0.2 nm. Those two numbers answer different questions. Reproducibility tells you how tightly repeated scans agree with one another, while accuracy tells you whether the entire set of readings sits in the right place. An instrument with excellent reproducibility and mediocre accuracy will give you beautifully consistent data at the wrong wavelength, which is far more dangerous than noisy data because nothing looks obviously wrong.
The next question is how the accuracy figure was established. In UV-Vis practice, wavelength accuracy is normally verified against known references: the sharp emission lines of a low-pressure mercury lamp (253.7, 365.0, 435.8 and 546.1 nm) or a deuterium lamp (486.0 and 656.1 nm), or the broad absorption bands of holmium oxide or didymium filters that span the ultraviolet and visible regions. Ask which references were used, how many points were checked, and whether the tolerance applies across the full range or only at the wavelengths where the instrument happens to be strongest. A number verified at 656 nm says little about behaviour at 260 nm, and vice versa.
It is also worth asking whether the quoted ± value represents a worst case across the scan range, a typical value, or a single best measurement. And at what bandwidth was it determined? Accuracy and bandwidth are entangled: if you measure a sharp emission line with a wide passband, the apparent peak position can shift simply because the line is being averaged with its surroundings. Finally, ask about stability. Accuracy at installation is one thing; accuracy after years of lamp pulses, temperature swings, and moving between benches is another. Implen claims guaranteed lifetime accuracy for its instruments, and states that the NP80 requires no routine recalibration checks — claims worth testing against your own verification schedule rather than accepting at face value. A quick holmium filter check in your lab settles the question faster than any brochure.
Bandwidth: The Number That Shapes Every Peak You Read
Spectral bandwidth, sometimes called spectral resolution or spectral slit width, is the width of the wavelength interval the instrument passes to the detector at any given setting, usually expressed as full width at half maximum. A 1.5 nm bandwidth means that a nominally monochromatic 260 nm reading is actually an average of light spanning roughly 259 to 261 nm. For a broad absorption feature such as the 280 nm protein peak, that averaging barely matters. For a narrow peak, a shoulder, or two overlapping chromophores, it matters a great deal: the measured absorbance is flattened, the peak appears broader than it is, and the wavelength of maximum absorbance can move.
The practical consequences show up in places people do not always expect. Ratio-based quality metrics such as 260/280 and 260/230 depend on the shape of two peaks with very different widths, so bandwidth can bias them in ways that look like sample contamination. Dye-labeled oligonucleotides, where the dye absorbance sits close to the nucleic acid signal, are another case. So are small molecules with fine vibrational structure in organic solvents, and any wavelength scan used to identify an unknown contaminant. Bandwidth also interacts with stray light to set the upper end of the linear range, which is why a stray-light figure such as < 0.5% at 240 nm using NaI belongs in the same conversation as the bandwidth specification.
When reading a data sheet, ask whether the bandwidth is a measured value or a nominal design value carried over from a slit width or an array’s pixel dispersion. Ask whether it is constant across 200–900 nm or varies, and whether the phrase used is “better than 1.5 nm,” “< 1.5 nm," or something vaguer. The NanoPhotometer NP80 illustrates why precision in wording matters: Implen's feature bullet lists bandwidth as "better than 1.5 nm," the optical specification table states "< 1.5 nm," and the cuvette option is described as offering spectral resolution < 1.5 nm. Those are consistent in spirit but not identical statements, and a careful buyer notes the difference rather than averaging them into one number.
Questions to Ask Before You Compare Two Spec Sheets
Start from your own assays rather than from the instrument list. Do you need accuracy at 260 nm for nucleic acid quantification, at 280 nm for protein UV, at 595 nm for Bradford, at 600 nm for cell density, or across 400–700 nm for dye chemistry? A core facility supporting many users needs a specification that holds across the whole working range; a method developer transferring a single validated assay needs it to hold at the few wavelengths that assay actually uses. Once you know which wavelengths carry your result, the comparison becomes concrete.
Match the specification to the assay, not the headline
If your workflow involves sharp features — wavelength scans, contaminant identification, dye-labeled oligos — push hard on bandwidth. If it involves ratio-based QC, ask how the instrument’s bandwidth affects those ratios at the concentration range you work in. If your samples are turbid or highly concentrated, pair the bandwidth question with photometric range and stray light. For absorbance accuracy, the NP80’s published figure is < 1.75% at 0.7 A at 280 nm of the reading, which is a perfectly reasonable way to state the spec — but it is a single wavelength and a single absorbance level, so ask whether the vendor can show you performance at the wavelengths and absorbance values that matter to you.
Ask how the figure is verified and maintained over time
Every specification is a snapshot of a measurement procedure. Ask what reference materials the vendor uses, whether the procedure is documented, and what your own incoming verification should look like. When you are evaluating a UV-Vis spectrophotometer such as the NanoPhotometer NP80, which combines microvolume and cuvette functionality with a 200–900 nm scan range, the sensible next step is to request the wavelength accuracy and bandwidth data as measured, then reproduce a subset of it with your own standards. If a vendor’s claim rests on design — as with Implen’s True Path Technology and its two fixed path-length anchor points — ask what evidence supports long-term stability and how you would detect drift if it occurred. Specifications you can verify are worth more than specifications you can only admire.
Porto Alegre jazz trumpeter turned Shenzhen hardware reviewer. Lucas reviews FPGA dev boards, Cantonese street noodles, and modal jazz chord progressions. He busks outside electronics megamalls and samples every new bubble-tea topping.