The Bortle dark-sky scale: nine classes, limiting magnitudes and what they mean for imaging

Where the scale comes from, and what we could read

John E. Bortle published his nine-class dark-sky scale in Sky & Telescope in February 2001. The magazine’s web version returned an HTTP 403 error to our checks, so we could not read the original. Everything below comes from two secondary copies: the Northern Virginia Astronomy Club, which reproduces the full class descriptions, and a table in Michael Covington’s December 2001 book manuscript, which abbreviates them.

The scale is visual. Each class is defined by what an observer sees: the zodiacal light, the structure of the Milky Way, whether particular galaxies show to the naked eye, how bright clouds look, and the faintest star visible, the naked-eye limiting magnitude (NELM).

The nine classes

Class	Name	NELM (NOVAC)	NELM (Covington)	32 cm reflector reaches (NOVAC)
1	Excellent dark-sky site	7.6–8.0, with effort	above 7.6	17.5 with effort
2	Typical truly dark site	7.1–7.5	7.1–7.5	16 or 17
3	Rural sky	6.6–7.0	6.6–7.0	16
4	Rural/suburban transition	6.1–6.5	6.1–6.5	15.5
5	Suburban sky	5.6–6.0	5.6–6.0	14.5–15
6	Bright suburban sky	about 5.5	about 5.5	14.0–14.5
7	Suburban/urban transition	5.0, “if you really try”	about 5.0	barely 14
8	City sky	4.5 at best	about 4.5	little better than 13
9	Inner-city sky	4.0 or less	4.0 or less	not given

Both copies give ranges for classes 1–5 and single values for classes 6–9. Tables that give a range for every class are not what these copies show. The two disagree only at class 1, where NOVAC’s fuller text has 7.6 to 8.0 “with effort” and Covington writes “above 7.6”. We follow NOVAC, because it reproduces the wording rather than summarizing it, but neither could be checked against the magazine.

What the classes look like

A few markers from the NOVAC text show how quickly things change. In class 1 the zodiacal light, gegenschein and zodiacal band are all visible, and the Triangulum Galaxy, M33, is an obvious naked-eye object. In class 3 M33 is easy with averted vision; in class 4 it is a difficult averted-vision object, only when higher than 50°. By class 6 M33 cannot be seen without binoculars and the Andromeda Galaxy is only modestly apparent. In class 7 the whole sky has a grayish white hue and the Milky Way is invisible or nearly so. In class 9 only the Moon, the planets and a few bright star clusters still give pleasing telescopic views.

The copies also carry material that may not be Bortle’s. NOVAC’s page adds one-line notes, such as light-pollution gradients in long exposures at class 3 and star counts above 30° elevation for classes 7 and 9, that are absent from Covington’s summary. Covington adds his own comments, for example that class 5 is still good enough for serious observing and astrophotography. We could not tell which, if any, appeared in the magazine.

What the scale does not give

Neither copy contains a measured sky brightness, for example in magnitudes per square arcsecond. We found no primary source for converting classes into sky-quality meter readings, so this page gives none. The limiting magnitudes also depend on the observer: the reproduced wording includes “with effort” and “if you really try”.

What sky brightness means for imaging

Sheffield’s notes on the CCD equation show the mechanism. When the sky background dominates the noise, S/N for a given target falls as the square root of the sky level, which the notes give as the reason to avoid light pollution and observe faint objects near new Moon. In that regime, S/N is proportional to the target signal times √t divided by √(sky), so holding S/N constant needs integration time in proportion to the sky brightness (derived):

time for equal S/N ∝ sky background per pixel   (background-limited case)

A sky twice as bright per pixel needs twice the time. Because the Bortle scale gives no sky brightness, it cannot turn a class into hours; the background in your own calibrated frames can. The integration time guide works through the equation.

Filters change the background term directly. Baader’s preshift article states that a sky brighter than Bortle 7 justifies its 3.5/4 nm filters, which it rates at about 1.4 times the contrast of its 6.5 nm filters. Its whitepaper derives the limit for strong light pollution: the S/N ratio of two filters tends to the square root of the inverse ratio of their bandwidths, √(6.5 ÷ 3.5) = 1.36. In time terms (computed), the 6.5 nm filter would need about 1.86 times the integration for the same S/N in that limit. Narrower filters bring their own constraint, a narrower range of usable focal ratios, covered in the narrowband filters guide.