Why Measure Color
Roasters have been judging roast degree by color since coffee was first roasted over fire. The human eye is remarkably sensitive to color differences—we can distinguish thousands of brown shades—but it’s also remarkably unreliable. Lighting conditions change. Fatigue sets in. One roaster’s “medium” is another’s “medium-dark.” And the gap between what two people call “light roast” can span 15 to 20 Agtron points, representing a significant sensory difference in the cup.
Color measurement instruments solve the subjectivity problem by assigning a numeric value to the roast color of ground or whole-bean coffee. That number is repeatable, comparable across time and location, and independent of lighting conditions or human perception. It doesn’t tell you everything about a roast—color correlates with roast degree but doesn’t capture development time, rate of rise, or the internal chemistry of the bean. But it tells you one important thing very precisely, and in quality control, one precise measurement is worth more than a dozen subjective impressions.
The Agtron Scale
The Agtron Gourmet Scale is the industry reference standard for coffee color measurement. Developed by Agtron Inc. (now a division of SCA-affiliated technology providers), the Agtron spectrophotometer measures the amount of near-infrared light reflected from the surface of a ground coffee sample. Lighter coffees reflect more light; darker coffees absorb more. The instrument converts the reflectance measurement into a numeric value on a scale from 0 (absolute black, theoretical) to 100 (absolute white, theoretical).
In practical coffee terms, the scale works out roughly as follows:
Agtron 95-85 (very light): Extremely light roast, Nordic-style. The coffee appears tan to light brown, may show visible chaff residue, and often has a mottled or uneven surface appearance. Cup character is dominated by origin flavors—fruit, floral, herbal—with minimal roast character. These readings are found almost exclusively in Scandinavian specialty roasting.
Agtron 85-75 (light): Light specialty roast. Uniform light brown color, clean surface. High acidity, pronounced fruit and floral aromatics, light body. This is the working range for most light-roast specialty operations worldwide.
Agtron 75-65 (light-medium): The sweet spot for many specialty roasters targeting filter coffee. Origin character is still clearly present but balanced with emerging caramelization sweetness. Acidity is bright but not sharp. Many competition filter coffees land in this range.
Agtron 65-55 (medium): Traditional medium roast. Brown color with no visible oil on the surface. Origin character and roast character are roughly balanced. Chocolate, caramel, and nut notes begin to dominate over fruit and floral. This is the center of the American specialty range and the approximate target for most omni-roast profiles.
Agtron 55-45 (medium-dark): Visible oil migration beginning on the surface. Roast character is clearly dominant over origin character. Body is full, acidity is reduced, and flavor notes shift toward dark chocolate, brown sugar, and toasted grain. Traditional espresso profiles for Italian-style preparation often target this range.
Agtron 45-35 (dark): Oil clearly visible on the bean surface. Significant carbonization of sugars. Smoky, bittersweet, and ashy flavors dominate. Origin character is largely obscured. French roast and many commercial espresso blends fall here.
Agtron 35-25 (very dark): Heavy oil, near-black appearance. Carbonization is advanced. Bitter, smoky, and acrid flavors with little residual sweetness. Italian-style roast in the traditional European sense.
The SCA maintains a set of calibrated color reference discs (the Agtron/SCA Roast Classification Color Disc Set) that provide physical standards for eight roast levels from #95 (lightest) to #25 (darkest). These discs are used for visual reference, Agtron calibration, and establishing a common vocabulary between roasters, importers, and quality professionals.
Ground vs. Whole Bean Readings
A critical nuance of Agtron measurement is that ground and whole-bean readings differ significantly for the same coffee. The whole-bean surface is smoother and more reflective than the ground surface, and it represents only the outer layer of the bean—which is always more developed (darker) than the interior. Ground coffee exposes the interior, which is lighter, and the rough surface scatters light differently than the smooth bean surface.
The typical difference between whole-bean and ground readings is 8 to 15 Agtron points, with ground readings consistently higher (lighter) than whole-bean readings. A coffee that measures Agtron 60 whole bean will typically measure 70 to 75 ground. This difference is not an error—it’s a real measurement reflecting the real color difference between the bean surface and its interior.
The difference between ground and whole-bean readings is itself a meaningful data point. A large gap (15+ points) suggests uneven development: the exterior is significantly more roasted than the interior, which indicates either too-fast roasting (the outside cooked before heat penetrated to the center) or under-development (the roast ended before the interior caught up to the exterior). A small gap (8 to 10 points) suggests even development throughout the bean, which generally correlates with better cup quality—more uniform extraction, cleaner flavors, and fewer astringent or grassy notes from under-developed interiors.
The SCA cupping protocol specifies ground-sample Agtron readings for standardization purposes. Most commercial quality-control programs measure ground samples. Some roasters measure both whole-bean and ground as a routine development check.
Modern Color Measurement Instruments
While the Agtron spectrophotometer established the scale and methodology, several competing instruments are now used in the coffee industry, each with different price points, measurement principles, and practical considerations.
Agtron (M-Basic, E30-CP)
The original and still the reference standard. Agtron instruments use near-infrared reflectance and are factory-calibrated to the Agtron Gourmet Scale. The M-Basic model (designed for the coffee industry) uses a ceramic sample cup, requires approximately 50 grams of ground coffee per measurement, and provides a direct Agtron number readout. Accuracy is excellent, repeatability is plus or minus 0.5 Agtron points, and the instrument requires periodic recalibration with factory-supplied reference tiles. Price range: $3,000 to $5,000.
Lighttells (CM-100, CM-200)
Lighttells instruments are manufactured in Taiwan and have gained significant market share in the specialty coffee industry since the mid-2010s. They measure color using visible-light spectrophotometry rather than near-infrared, and output readings in both Agtron-equivalent units and their own proprietary scale. The CM-100 is a compact, relatively affordable benchtop instrument that requires smaller sample sizes than the Agtron (approximately 20 grams). The CM-200 adds additional measurement capabilities including moisture and whole-bean color mapping. Price range: $1,200 to $2,500.
Lighttells instruments correlate well with Agtron readings across the middle of the roast spectrum (Agtron 45-75) but show slightly larger divergence at the extremes—very light and very dark roasts. For most practical quality-control purposes, the correlation is sufficient. Some quality labs maintain both an Agtron and a Lighttells and use the Agtron as the reference for calibrating the Lighttells.
Tonino (ColorTest)
The Tonino is a small, portable, open-source color measurement device designed specifically for coffee. Originally developed as a DIY project in the specialty coffee community, it was commercialized as the Tonino ColorTest and gained a following among home roasters and small-batch professional roasters. The Tonino measures color using a visible-light LED and photodiode sensor, outputting readings on a 0-200 scale (the Tonino scale) that can be converted to approximate Agtron values using published correlation tables.
The Tonino is significantly less expensive than commercial instruments ($200 to $400) and is adequate for tracking roast-to-roast consistency within a single operation. Its absolute accuracy is lower than Agtron or Lighttells, and readings can vary with sample preparation (grind size, tamping pressure in the sample cup, sample temperature). For professional quality-control programs that need to communicate color values to external parties, the Tonino is best used as a screening tool rather than a reference instrument.
ColorTrack by RoastRite
ColorTrack is a handheld, battery-operated color meter designed for in-roastery use. It measures whole-bean color using a visible-light sensor and outputs readings on the Agtron scale. Its advantage is portability and speed—a measurement takes seconds and requires only a small handful of beans directly from the cooling tray. Its limitation is lower precision than benchtop instruments (repeatability of plus or minus 2 to 3 Agtron points) and sensitivity to bean surface condition (oily beans read differently than dry-surface beans at the same internal development).
ColorTrack is most useful as a production-line tool for catching outlier batches in real time rather than for precise quality-control documentation. A roaster can check color on the cooling tray immediately after dropping a batch and flag any reading that falls outside the target window before the batch moves to packaging.
Roast Degree vs. Development
This is the critical distinction that color measurement alone cannot make: a color reading tells you how far the roast progressed in total, but it doesn’t tell you how it got there. Two coffees with identical Agtron readings can have dramatically different flavor profiles if they arrived at that color through different time-temperature pathways.
Consider two batches roasted to Agtron 65 ground:
Batch A: Total roast time 9 minutes, first crack at 7:30, dropped at 8:30. Development time: 60 seconds, DTR 11%. The roast is fast, with a short development phase. The color reads Agtron 65 because sufficient total thermal energy was applied, but the brief development time means Maillard reactions were not given enough time to complete. The cup may show grainy, cereal-like, or raw-dough flavors alongside the origin character—markers of under-development despite adequate color.
Batch B: Total roast time 12 minutes, first crack at 9:00, dropped at 11:30. Development time: 150 seconds, DTR 21%. The roast is slower, with an extended development phase. The same Agtron 65 reading, but the longer development time allowed Maillard and Strecker reactions to progress further, producing smoother caramelization, fuller body, and more integrated roast character. The cup is balanced and sweet with no raw or grainy notes.
Same color. Different coffees. This is why experienced roasters treat Agtron as one data point among many—useful for batch consistency and for communicating roast level to external parties, but insufficient as a standalone quality metric. Development time, rate of rise profile, bean temperature at first crack, and total roast time all contribute to the cup character in ways that color cannot capture.
QC Applications and Protocols
Color measurement serves several practical quality-control functions in a roasting operation:
Batch consistency monitoring: The most common application. A roaster establishes a target Agtron value (typically plus or minus 1 to 2 points) for each coffee on their menu and measures every production batch against that target. Batches outside the window are flagged for cupping evaluation before release. This simple protocol catches the most common production errors: charges that were too hot, batches that ran long, and environmental-temperature effects on the roaster’s thermal dynamics.
Blend component targeting: For blended coffees, color measurement ensures that each component is roasted to its specified level before blending. A blend recipe might call for a Brazilian component at Agtron 55 and an Ethiopian component at Agtron 70—measuring each component before blending ensures the blend profile is hit consistently across production runs.
Grind calibration checks: Color measurement can be used as a proxy for grind consistency. If the same coffee, roasted in the same batch, produces different Agtron readings when ground on different grinders (or after a grinder burr change), it indicates a difference in particle size distribution. Finer grinds read lighter (more surface area exposing lighter interior) while coarser grinds read darker (less interior exposure, more surface reflection from the darker exterior).
Shelf-life monitoring: Green coffee color changes measurably during storage as chlorogenic acids degrade and the bean surface fades from green to yellow-brown. Measuring green coffee color at intake and at intervals during storage provides a quantitative indicator of aging that supplements cupping evaluation.
Communication with buyers: Agtron values provide an objective reference point when discussing roast preferences with wholesale accounts. Telling a cafe that their espresso blend is roasted to Agtron 52 ground is more precise and reproducible than describing it as “medium-dark.”
The Limitations of Color
Color measurement is a powerful tool, but treating it as a comprehensive quality metric leads to bad decisions. The things color cannot tell you include:
Development adequacy: As discussed above, color measures outcome but not pathway. A fast, under-developed roast and a slow, fully-developed roast can produce identical color readings.
Defect presence: A roast with quakers (under-ripe beans that fail to brown normally), tipping (charred edges from excessive conductive heat), or scorching (contact burns on the bean surface from a too-hot drum) will produce a color reading that averages across defective and non-defective beans. The average may look normal while the actual batch contains significant defects.
Moisture content at roast start: The same roast profile applied to beans at 10% and 12% moisture will produce different development outcomes at the same color endpoint. Higher-moisture beans absorb more energy before Maillard reactions begin, potentially under-developing relative to drier beans at the same color reading.
Cultivar variation: Different cultivars respond differently to the same thermal input. A Geisha and a Catimor roasted to identical Agtron readings will have arrived there through different internal chemistry, and the cup quality will reflect those different pathways regardless of the identical color.
The most sophisticated quality-control programs in the industry use color measurement as one element of a multi-variable monitoring system that also includes time-temperature profiling (via Cropster or similar software), regular cupping, and physical defect assessment. Color is the most easily measured variable, which makes it the most commonly used—but it’s most valuable when the roaster understands exactly what it does and doesn’t tell them.