When you search for literature on any topic, you quickly notice something familiar: a handful of journals keep showing up again and again, while the rest of the relevant articles are scattered thinly across dozens or even hundreds of other periodicals. Samuel C. Bradford described this pattern back in the 1930s, and the most powerful way to understand it is not through its verbal statement but through a picture. The graphical form of Bradford’s Law, known as the Bradford curve or Bradford bibliograph, turns an abstract ratio into a visual story about how scholarly information clusters and disperses. This post unpacks that curve, section by section, so you can read it the way a scientometrician does.
Table of Contents
From a verbal law to a visual curve
Bradford’s Law began as an observation, not a graph. As Samuel C. Bradford, a librarian in London studying geophysics literature, noticed there is an exponentially diminishing return when you keep extending a search for references across more and more journals. He tested this by examining two bibliographies, the Current Bibliography of Applied Geophysics and the Quarterly Bibliography of Lubrication, ranking the journals in descending order of productivity and dividing the articles into three roughly equal zones.
The verbal form of the law states that if you arrange journals by decreasing productivity, you can split them into a nucleus of highly relevant periodicals and successive zones, each containing the same number of articles as the nucleus, with the journal counts following the ratio 1 : n : n². This is clean and quotable, but it hides something. It does not show you how the data actually behaves as you accumulate more journals. That is where the graph earns its keep.
What makes the graphical version interesting is a subtle historical point: Bradford expressed his law graphically using experimental data without realising the graphical expression was not mathematically identical to the verbal formulation. In other words, the two forms of the law describe the same phenomenon but are not perfectly interchangeable. Understanding the curve helps you see exactly where and why they diverge.
How the Bradford curve is plotted
The Bradford curve is built from a ranked table before it ever becomes a graph. First, every journal that has published on the subject is listed in descending order of productivity, meaning the journal with the most relevant articles sits at rank 1. Then a running, or cumulative, total of articles is calculated as you move down the list. The first journal contributes its articles, the second adds its share to the total, and so on, until the final journal accounts for the last few stray articles.
The graph itself uses a specific arrangement of axes. Bradford plotted R(n), the cumulative total of relevant papers, against the logarithm of n, the cumulative number of productive journals. The vertical axis (y-axis) shows the cumulative number of articles on a normal linear scale. The horizontal axis (x-axis) shows the cumulative number of journals, but on a logarithmic scale. This choice of a semi-logarithmic plot is not decorative. It is what gives the curve its distinctive and useful shape.
Why use a logarithmic x-axis at all? Because the number of journals grows so dramatically across the zones. If you plotted journals on a normal scale, the core would be squashed into a tiny sliver at the left edge while the long tail of single-article journals stretched far off to the right. The logarithmic scale compresses that enormous range into a readable picture, and crucially, it transforms the relationship in the middle of the data into a straight line, which is the heart of the law.
The overall shape: an elongated S
When the cumulative articles are plotted against the log of cumulative journals, the resulting graph is not a simple straight line. The data reveal an elongated S-shaped curve made of three distinct parts. Reading it from left to right, the curve rises steeply at first, then settles into a straight diagonal climb, and finally bends and flattens out at the far end.
Each of these three parts corresponds to one of Bradford’s zones. The steep beginning is the core. The straight middle is the moderately productive mid-section. The flattening end is the scattered tail. The beauty of the graph is that you can literally see the productivity of journals decreasing as your eye travels across it, because the curve grows less steep with every step to the right. Let us look at each zone in turn.
The core zone: a steep, concave climb
The leftmost part of the Bradford curve is the steepest. This initial concave portion represents the higher density of the nuclear zone. Here, each individual journal contributes a large number of articles, so the cumulative total shoots upward rapidly even though you have added only a few journals. This small, intensely productive cluster is the core, sometimes called the nucleus or nuclear zone.
The core matters enormously in practice. The core zone is a small group of journals that carries the most relevant and widely cited articles on a subject. For a researcher or a librarian, identifying the core is the whole point of the exercise. If you can subscribe to or monitor just the handful of journals in the nucleus, you capture a disproportionate share of the important literature on your topic. This is exactly why databases like the Web of Science have historically focused strictly on the core of international scientific journals while leaving the majority of publications in the outer zones aside.
To appreciate just how concentrated the core can be, consider a real dataset. In a citation study of a medical journal, the journals split into three zones in the proportion 43 : 210 : 1270, meaning only about 2.82% of all the journals accounted for a full third of the citations. That tiny percentage doing a third of the work is the core zone made visible.
The mid-section: the linear straight line
After the steep rise of the core, the curve straightens out into a long diagonal line. This linear portion, when the data are plotted on a semi-log scale, is equivalent to the Zipf distribution, which is why the relationship is often called the Bradford-Zipf distribution. This straight middle stretch is the mid-section, representing the moderately productive zone of journals.
The linear part is special because it is the section that actually corresponds to the verbal form of the law. As one teaching resource explains, in the mid-section the cumulative number of articles contributed by journals up to a given rank increases proportionately with the logarithm of the rank. In plain terms, every time you multiply the number of journals by a fixed factor, you add the same fixed number of articles. That constant multiplier is the Bradford multiplier, the n in the 1 : n : n² ratio.
This is where the geometry connects back to the zone idea. If you divide the straight-line portion into zones each containing the same number of articles, each zone corresponds to a set of journals, and the number of journals needed grows geometrically from one zone to the next. The straight line is therefore the visual proof of the law’s core promise: equal slices of literature require ever-larger groups of journals to supply them.
The tail zone: the Groos droop
At the far right of the graph, the straight line does not continue forever. It bends downward and flattens. This third part, often called the Groos droop, shows a departure from linearity for higher values of n, and the reason for it is not yet fully understood. This final section is the tail, also known as the peripheral zone, alien zone, or simply the periphery.
The tail is where relevance becomes thin. The area often called the Groos droop, or inflection point, represents the peripheral or outer zone, the least productive zone where relevant references are widely scattered among many journals. These are the periphery journals that sit far from the core subject and typically contribute just one or two topically relevant papers in a given period. You need to add a huge number of these journals to gain only a few more articles, which is why the curve barely rises even as it extends a long way to the right.
The droop is named after the researcher G. Groos, who documented this deviation. It is a reminder that Bradford’s Law is an empirical regularity, not a perfect mathematical rule. The verbal form predicts the straight line should continue, but real data consistently fall below that expectation at the extreme end. The graph honestly shows this gap, which is one more reason the visual form is so valuable for analysis.
Reading the curve as a whole
Put the three zones together and the Bradford curve tells a single coherent story about diminishing returns. If all references on a subject are divided into three equal groups, the first zone draws on a small core of journals, the second needs significantly more journals to supply the same number of citations, and the third needs many more still, producing fewer and fewer references per journal as you move outward. The slope of the curve at any point is a direct measure of journal productivity, steep where journals are rich in relevant articles and shallow where they are not.
This structure is not just a curiosity. It is the foundation of a practical technique called Bradfordizing, where a search result is re-ranked so that articles from core journals are placed ahead of those from the average Zone 2 journals and the sparse Zone 3 journals. By understanding the curve, information systems can push the most productive sources to the top of a result set, helping users find the central literature faster.
Why the graph beats the formula for understanding
The verbal and graphical forms of Bradford’s Law are two windows onto the same phenomenon, but they reveal different things. The verbal ratio of 1 : n : n² is compact and easy to test against a dataset. The graph, by contrast, shows the full behaviour of the distribution, including the steep core, the reliable straight mid-section, and the awkward droop at the tail that the formula glosses over. For a student trying to truly grasp scattering, the curve is indispensable because it makes the abstract idea of diminishing productivity something you can see at a glance rather than something you have to compute.
It is worth remembering that the law does not fit every dataset perfectly. In one scientometric study the three zones came out as 14 : 75 : 134 journals, which did not match the expected Bradford distribution well. Interdisciplinary topics, incomplete bibliographies, and fields that genuinely spread across four or five clusters can all distort the classic three-zone shape. The graph is honest about these deviations, which is precisely what makes it a better diagnostic tool than the tidy formula alone.
What do you think? If a database could only afford to index the journals in the core zone, what kinds of important research from the tail might quietly disappear from view? And when you analyse a subject you know well, would you expect its literature to form a clean three-zone curve, or do you think it would droop and scatter in ways Bradford never anticipated?
References
- https://link.springer.com/article/10.1007/s00381-014-2481-9
- https://www.researchgate.net/publication/271753203_Bradford's_law_of_Scattering_Revisited_A_study_based_on_the_References_in_Doctoral_theses_in_the_area_of_Physics
- https://www.researchgate.net/publication/280218558_Bradford's_Empirical_Law
- https://www.sciencedirect.com/science/article/abs/pii/S0022480418302336
- https://arxiv.org/pdf/1305.0357
- https://arxiv.org/pdf/1301.5380
- https://egyankosh.ac.in/bitstream/123456789/11366/1/Unit-6.pdf
- https://arxiv.org/pdf/1309.7949
- https://arxiv.org/pdf/2006.08486
Leave a Reply