Two grow-light listings sit side by side. Both say “full spectrum”, both show a colourful curve, and one has a much taller red peak. Which tells you more about the light you would actually receive?
This grow light spectrum guide helps South African buyers read those charts without confusing colour, photon output and canopy coverage. You will learn what the axes mean, why separately scaled graphs cannot show which lamp is stronger, and how to check a wavelength claim against the exact fixture being sold. There are worked examples, a local product-specification check and a short message you can send a supplier.
Start with this rule: compare the measurement behind the picture, not the picture alone. A useful graph identifies the model, units, wavelength range and operating configuration. Without them, it is an illustration rather than a dependable comparison.
First, identify what the chart is measuring
A spectral power distribution, usually shortened to SPD, describes a light source’s output across wavelengths. A graph of plant sensitivity describes something different: a biological response. Putting the two on one page does not make a plant-response curve a test of the lamp. The DesignLights Consortium glossary defines SPD as power from a source as a function of wavelength.
Look for a report title or caption that identifies the complete fixture. A diagram supplied for an LED chip is not automatically the spectrum of an assembled light with several kinds of diode. Nor does a generic sunlight illustration document a particular product.
For the broader buying checklist, start with our guide to checking full-spectrum LED grow lights. Here, we will stay with the chart itself.
Read both axes before following the curve
The horizontal axis normally gives wavelength in nanometres, abbreviated nm. The vertical axis needs closer attention:
| Vertical-axis label | What you are looking at | What to check next |
|---|---|---|
| Relative output, arbitrary units or a normalised scale | The distribution after an unspecified or stated rescaling | How the curve was normalised; absolute output is not supplied by the scale alone |
| Spectral radiant flux, such as W/nm | Radiant power per wavelength interval | Whether this is whole-fixture output and which interval is included |
| Spectral photon flux, such as µmol/s/nm | Photon output per wavelength interval | Whether the report supplies absolute values and the same measurement band |
| Spectral irradiance, such as W/m²/nm | Light arriving at a measurement surface, per wavelength interval | Measurement position, distance and geometry |
These labels are not interchangeable. The US Department of Energy’s colour and spectrum resource includes examples with wavelength on the horizontal axis and radiant power on the vertical axis. Check the actual labels on your report rather than assuming every rainbow-shaped diagram uses that convention.
An unlabeled vertical axis is a missing measurement detail. It is not evidence that the product is poor, but it is a reason to request the original report before comparing it with another light.
Why two identical-looking curves can represent different lights
Normalisation makes curves easy to display together. It can also hide the difference you are trying to measure.
Suppose each manufacturer divides its spectral values by that light’s highest value. Both graphs then peak at 1. The curve retains its shape, but its absolute scale has disappeared. A taller-looking peak on a differently formatted page tells you very little.
Consider this invented arithmetic example, not a test of TheOneGrow products or a recommended specification:
| Information supplied | Example A | Example B |
|---|---|---|
| Normalised curve | Same shape; peak at 1 | Same shape; peak at 1 |
| Complete-fixture input power | 300 W | 300 W |
| Reported PPF, both measured over 400–700 nm | 600 µmol/s | 900 µmol/s |
| PPF divided by input power | 2.0 µmol/J | 3.0 µmol/J |
In this example, the additional absolute-output figures distinguish the lights; the normalised pictures do not. The calculation uses photosynthetic photon efficacy, or PPE: PPF divided by electrical input power, as defined by DLC.
Even then, neither the curve nor that calculation shows how evenly either lamp lights a growing area. That needs separate spatial measurements. For that next step, use our guide to reading grow-light coverage maps.
A red peak is not a red-light percentage
Peak height describes a narrow part of a curve. A wavelength band covers an interval. Confusing those two can turn a plausible-looking graph into a misleading sales comparison.
For example, a narrow red peak may be taller than a broad green region, while the broad region still represents a substantial share of the total. To calculate a photon percentage, use photon-based data summed or integrated over explicitly defined bands—not the tallest pixel in a screenshot.
Here is a second illustrative calculation, using invented band totals from a fictional photon-output report:
| Non-overlapping reporting band | Integrated photon output | Share of the 400–700 nm total |
|---|---|---|
| 400 to below 500 nm | 90 µmol/s | 15% |
| 500 to below 600 nm | 150 µmol/s | 25% |
| 600 through 700 nm | 360 µmol/s | 60% |
| Total | 600 µmol/s | 100% |
The red-band share is 360 ÷ 600 × 100 = 60%. It is not a recommended spectral ratio. It simply shows what the calculation requires: defined band edges, integrated photon values and a declared denominator.
A radiant-power percentage and a photon percentage need not be equal. Photon energy varies with wavelength, so converting between them requires wavelength-resolved data; Apogee’s conversion notes explain why the source spectrum matters. Ask for the numerical export or a report containing the band totals; do not reconstruct precise percentages from a small marketing image.
What colour labels can—and cannot—tell you
Keep colour claims in proportion. A spectrum can describe a light; it cannot promise a harvest on its own.
Green wavelengths are not automatically wasted. In a controlled study of lettuce leaves, Liu and van Iersel found that the photosynthetic responses to blue, green and red light depended on intensity, with green light’s distribution within the leaf helping explain its usefulness. That is a reason to reject “plants cannot use green” as a blanket statement—not to prescribe one mixture for every crop. See the original research.
The same caution applies to a pronounced red peak. Its presence does not establish a yield increase, and its absence from a cropped or low-resolution image does not establish that the fixture emits no red light. Request the full report before drawing either conclusion.
Does 3000K or 5000K settle the comparison?
No. Correlated colour temperature, or CCT, characterises the appearance of white light. Different spectral distributions can have the same apparent colour; the DOE resource linked above demonstrates this explicitly.
Treat a Kelvin label as one descriptor, not a substitute for the curve. A mixture of white LEDs and additional red LEDs needs a report for that mixture. Choosing between two fixtures still requires their output and coverage information, not a warmer-versus-cooler rule.
Check what happens outside the reported PAR band
The traditional photosynthetically active radiation band is 400–700 nm. Purdue Extension’s measurement guide explains this band and the distinction between light quality, intensity and duration. A report ending at 700 nm cannot, by itself, settle a claim about output beyond that limit.
Far-red research also needs careful interpretation. Zhen and Bugbee demonstrated useful interactions between 700–750 nm photons and shorter wavelengths in controlled lettuce experiments. Their published study does not make every far-red accessory necessary, or validate a retail product’s promised yield increase.
For comparison purposes, ask whether a photon figure covers 400–700 nm or an extended interval such as 400–750 nm. Apogee’s sensor documentation distinguishes instruments using these two bands. A higher number measured across a wider band is not automatically a like-for-like improvement.
Likewise, “UV” and “IR” are not sufficiently detailed specifications. Ask which wavelengths are claimed, whether they were measured, and which channels were operating. Do not infer exposure instructions or safe operating procedures from a coloured peak. Use the exact product’s instructions and safety documentation.
A South African listing example: checking the GLOC 720W
The TheOneGrow GLOC 720W product page provides a useful starting point for practising this distinction. At the time of this review, its published specifications list 720 W input power, 1,872 µmol/s PPF and 2.6 µmol/J efficacy. Its description also mentions 660 nm deep-red enhancement.
The arithmetic is consistent: 1,872 ÷ 720 = 2.6. That checks the relationship between the listed numbers. It is not an independent measurement of the fixture, and it does not reveal the proportion of photons in each wavelength band.
Before using the listing to make a detailed spectral comparison, request the whole-fixture report for the supplied version, with its measurement band, axes and operating settings. Keep the product page and report together. If a supplier changes the driver, diode configuration or controllable channels, establish whether the report still represents the quoted unit.
Turn a spectrum claim into a usable supplier response
For South African buyers comparing a local listing with an imported option, a report is useful only when it can be tied to the unit on the quotation. Keep the local model name, supplied configuration and supporting documents together rather than comparing whichever graph is easiest to find online.
Use this compact worksheet for each candidate:
| Record in your comparison | A useful answer | When the information is missing |
|---|---|---|
| Model and configuration | Same model/version as the quotation, with operating channels identified | Mark the curve as provisional; request the matching report |
| Axes and normalisation | Clear units and an explanation of any relative scaling | Do not rank total output from peak height |
| Wavelength interval | Stated limits for the curve and each integrated output figure | Do not compare totals reported across different bands |
| Photon proportions | Defined band boundaries and photon-based totals or numerical data | Do not invent percentages from the graphic |
| Test setting | Input power, dimmer/channel settings, date and measurement method | Do not assume a maximum-setting report describes another setting |
| Supporting measurements | A corresponding output report and, separately, a coverage map | Keep spectrum, output and coverage conclusions separate |
A useful request need not sound like a laboratory tender:
Please send the complete-fixture spectrum report for the model and version on this quotation. Please include the axis units, whether the curve is normalised, the wavelength range used for PPF, and the input-power and channel settings during measurement. Where available, please include the numerical spectral data and a corresponding PPFD map. Please identify any differences between the tested configuration and the unit supplied in South Africa.
This request asks for evidence, not a promise that a particular colour will solve every growing problem. Broader procurement questions belong in our professional LED grow-light buying guide.
Keep heat and running cost outside the spectrum verdict
A smooth spectrum does not make a light heat-free or tell you its electricity bill. Nelson and Bugbee’s LED and HPS leaf-temperature study shows why lighting technology should not be separated from the surrounding conditions when discussing heat.
For your own space, check the fixture’s stated operating limits and installation requirements. Work out electricity costs from actual input power, your intended operating time and the tariff on your bill, with other equipment accounted for separately. Do not convert a spectrum claim into an assumed cooling-cost saving or a South Africa-wide running-cost estimate.
Questions that often come up when reading a spectrum report
Can a normalised graph show which light produces more photons?
Not when each graph has been independently scaled to its own peak and no absolute scale is provided. Ask for comparable whole-fixture photon-output figures. If the reports use different wavelength intervals or settings, resolve those differences first.
Does a taller 660 nm peak mean more red light overall?
Not necessarily. Peak height is not the integrated photon output of the red band. Ask for photon-based band totals with the boundaries stated. Even a valid red-band percentage does not, on its own, demonstrate a better crop outcome.
Can I compare a W/nm graph with a photon-output graph directly?
Not by comparing peak heights or areas as though the units were identical. One reports radiant energy and the other photon output. Request the same representation from both suppliers, or obtain suitable numerical data for a wavelength-by-wavelength conversion.
Will an ordinary PAR meter show me the spectrum?
A band-integrating quantum meter gives a photon-density reading over its response band, not a wavelength-by-wavelength curve. Check its documented spectral response. A wavelength-resolved report requires an appropriately calibrated spectroradiometric measurement, not just a single PPFD reading.
Can the chart tell me how high to hang the fixture?
No. Wavelength distribution does not describe how the light spreads across your growing area. Use the fixture’s installation guidance and the relevant coverage information. Keep a spectral comparison separate from a hanging-height recommendation.
The useful conclusion is a traceable comparison
A good spectrum comparison ends with a clear statement of what the documents establish—and what they leave unanswered. You should be able to identify the fixture, explain the axes, recognise normalised data and compare photon figures over the same band. Then use coverage, installation and support information to finish the buying decision.
The most convincing picture is not necessarily the most informative one. Choose the evidence you can check.