In developing this answer, and others, it seems fairly typical that I start out broad and then add more and more detail as necessary to answer the question and respond to comments. Sometimes the act of digging deeper itself leads to a need for more detail. This can be both in detail and quality of sources.

For example, in that answer I started with a broad sweep of the situation on the Eastern Front based on Wikipedia and my own past knowledge. In response to objections and assertions I dug up specific numbers and better sources to check and rebut them. Finally I went into details of individual divisions. Had I decided to go further, I would have started to break the numbers down by year, month, and sector.

In this case my original assertions were strengthened, but sometimes they are weakened by refinement. Or in a more collaborative tone, sometimes our understanding is confirmed, sometimes our understanding is shifted.

It feels a bit like the progress of science: continual refinement of understanding. It also feels like every level of detail acts pretty much the same whether it's at national, unit, or individual level. Like zooming in on a fractal.

Is this a pattern others have observed? Is it a useful way to think about historiography, particularly the casual historiography on this site?

  • 6
    Quite a good metaphor, but is this 'a question that can be answered, not just discussed'? I think that's not a great rule for a history site, so no down or close vote from me, but I think this is likely to attract both
    – Ne Mo
    Aug 3, 2018 at 19:46
  • This reminds me of Braudel's time divisions. A 'broad sweep of the situation' could be considered its conjuncture, while going specific makes us dwell into events. Of course there are more levels to it than Braudel's three divisions, but it is a similar concept, methinks.
    – James Cook
    Aug 3, 2018 at 21:38
  • @NeMo I considered whether it was more appropriate for History.meta, but ultimately decided it's about history, not History.SE.
    – Schwern
    Aug 4, 2018 at 3:23

4 Answers 4


The organic growth of inquiry is well-known. You ask whether it resembles a fractal, which is a self-similar shape with an infinite edge length.

According to Michel-Rolph Trouillot, the problem of historiography is defined not only by the evidence we have, but also by the silences around that evidence. The refinement of a historical thesis requires concern about the factors that aren't recorded. Fragments of evidence are escarpments into a sea of silences; despite their narrowness, closer inspection reveals yet more coastline.

Nonetheless, the fragmentary nature of historical evidence means the lay of the land isn't self-similar like a fractal. When a clue opens up new avenues of research, we can't expect something parallel to happen elsewhere in the problem space. In this regard, a growing space of inquiry might be better modeled as a connected network graph like the spread of an infection.

Historiography is like a fractal in that greater detail is continually available, but is unlike one in that it exhibits limited internal symmetry.

  • "escarpments into a sea of silences" :) The analogy to a graph is an interesting one. One cannot simply zoom and enhance, you need a contiguous connection to the neighboring level. The graph is undirected, details add up to broad trends and broad trends affect detail.
    – Schwern
    Aug 6, 2018 at 16:54
  • These are all very good answers, thank you. I'm choosing this one because looking at history as a network graph provides a more accurate analogy than my fractal one.
    – Schwern
    Aug 7, 2018 at 22:43

All models are wrong; some models are interesting George Box

Can it be said? Sure, absolutely. Is it useful to say? That is the question.

I think your observation resonates with me - research only expands the horizon of topics to which the research could connect.

But I'm not sure that the observation helps me in any way. I'm not sure that I'm a better researcher, or that my reputation will increase dramatically

It feels a bit like the progress of science: continual refinement of understanding.

My professional historian girlfriend points out that it feels like science because historiography is a science.

  • 2
    I would hate to cause a dispute, but you might diplomatically inform your girlfriend that historiography probably is a science, but histiography certainly isn't. ;)
    – Evargalo
    Aug 6, 2018 at 8:55
  • 5
    LOL - she puts up with me despite my spelling.
    – MCW
    Aug 6, 2018 at 9:02
  • 1
    I think of historians like cosmologists. We can never go to where we study. We can never perform experiments on our subjects. We can just observe the fragments we have and try to tease trends and isolate variables from them. Cosmologists have the distorted time machine of light. Historians have the distorted time machine of the historical record.
    – Schwern
    Aug 6, 2018 at 17:01

Yes historiography is discursive techniques all the way down, like turtles. The problem is that discourses aren't built out of knowable rules—to our knowledge—and we currently think that if they were we couldn't know them. It is the current belief that if structures of social and cultural discourse if they did infact exist would be unknowable. This is the post-structural critique of structuralist knowledge. But EP Thompson also said it in plainer English when attacking Althusser.

So while the metaphor is a potentially valid "lie for children," in terms of science education, eventually we need to push the metaphor until its capacity to explain breaks.

One other problem is that if similarity appears at a deeper level of inquiry, the similarity is all too often the historian, or group of historians', biases that are replicated across scales of analysis. I, for example, am always going to read for labouring class self activity, minority cultures, changes in methods of exploitation and resistance, changes in control over social wealth and "productive" social wealth. I have to read twice for "moral economies" as opposed to "political economies."

While my readings exist in the documentary records of the past it is brave today to suggest that my readings are the "best fit." This is true of all readings that don't abuse the documentary record, or which haven't been rendered perverse by new documents or readings.

This is the other broken aspect of the metaphor: the relationship between the documentary records of the past and the historian is interpenetrative and hermeneutic. I read into the text, and I read the text into my reading of the text. At the crudest scale: two agents with apparent intention-towards-meaning relate in history. Historiography is a human act.


A fractal is a geometric figure where each part has the same statistical character as the whole. Fractals are useful in modeling complex structures (such as eroded coastlines or snowflakes) in which similar patterns recur at progressively smaller scales, and in describing partly random or chaotic phenomena such as crystal growth, fluid turbulence, and galaxy formation.

Given that definition, One example where history could be modeled as a Fractal would be populations where the scale would be both #of people in the population and time. History is often studied at the civilization level, the collective experience, but also exists as a series of smaller experiences. Such as a battle's outcome is made up of division, platoon, company and individual soldiers experiences. We say Meade won the battle of Gettysburg in the U.S Civil War, but to study this we can also strive to understand the individual experiences which resulted in the collective experience.

  • The 20th Maine on Little Round Top
    • Col Chamberlaine
  • The 2th Regiment of the New York Militia on Cemetery Ridge
    • General Sickles
  • The Army of Northern Virginia and high water mark of the Confederacy
    • General George Pickett


That might be an example. Fractals like the Mendelbrote Set are infinitely detailed and reveal infinite granularity as one drills in from the larger scale. I guess if you think of individual groups as fractions of populations, using Geometric Series or Geometric Progression one could theoretically drill in infinitely, given an large enough population. Although ultimately one would be limited by the fact populations aren't infinite divisible unless time is part of the scale, because they are made up of distinct individuals which unlike mathematical theory are not infinitely divisible. My thought is this scale could be further increased, if it was compounded with time allowing for further expansion. Historic time is limited, although the limit goes back beyond what we can see, the future is perhaps unlimited (theoretically in a non Doctor Strange Love sense).

Doesn't fit perfectly though because fractals like the Mendelbrote Set are both infinitely expandable and infinitely contractable. Zoom both ways. But there are numerous examples of fractals being used for modeling where they are only zoomable, and not infinitely so. The pineapple, coast lines, or a lightning bolt.

I've heard taught history being more cyclical (although imperfectly so) or more specifically circular. History repeats itself, thus knowing where you've been, one can better understand where civilization might be heading. That also applies itself to fractals as fractals often at the same or even different levels on the given scale repeat patterns.

Examples of Cyclical Theories applied to History:

  • Dynastic cycle - dynasty rises to a political, cultural, and economic peak and then, because of moral corruption, declines, loses the Mandate of Heaven, and falls, only to be replaced by a new dynasty. The cycle then repeats under a surface pattern of repetitive motifs.
  • Kondratiev wave - history of civilizations are broken up into periodic waves. A wave ranges from forty to sixty years, the cycles consist of alternating intervals of high sectoral growth and intervals of relatively slow growth.
  • Social Cycle Theory - Social cycle theories are among the earliest social theories in sociology. Unlike the theory of social evolutionism, which views the evolution of society and human history as progressing in some new, unique direction(s), sociological cycle theory argues that events and stages of society and history are generally repeating themselves in cycles.
  • Tytler Cycle - "A democracy cannot exist as a permanent form of government. It can only exist until the majority discovers it can vote itself largess out of the public treasury. After that, the majority always votes for the candidate promising the most benefits with the result the democracy collapses because of the loose fiscal policy ensuing, always to be followed by a dictatorship, then a monarchy”.

Another example might be the relationship between the ascendence of pragmatic secular liberalism and messianic religious conservatism in the United States .

As detailed by former Kennedy aid and now historian Arthur M. Schlesinger Jr. in his book The Cycles of American History

Since the founding of the country the tension between these two often opposed philosophies have influenced public policy. When the national mood favors a focus on public purpose, liberals are ascendant.

  • Franklin Delano Roosevelt and his Great Depression and WW2 fighting administrations
  • Lyndon Baines Johnson and his war on poverty and space programs.

When citizens grow weary of activism, they trend to private interests and conservative rule.

  • Ronald Reagan and his administrations general theme of favoring private sector solutions and keeping the government out of peoples business. Smaller less obtrusive government.
  • I'm wary of cyclic theories because they're often presented as inevitable. I view them as consequences of ungoverned late stage systems and avoidable with good governance. And each cycle that governance improves. For example, a free market will boom and crash, but we know more about markets to buffer them... even if many are still seduced by the easy money of unbuffered boom times. Similarly, a democracy can be gamed by the oligarchy with demagoguery, but we now know better to recognize and stop that... even if many are still seduced by the false promises and easy excuses.
    – Schwern
    Aug 6, 2018 at 17:17
  • @Schwern: Yes I agree, when they occur they are obvious but predicting them is not as intuitive as 20/20 hind sight would suggest. As Mark Wallace quoted a statistician , all models are wrong, but some are interesting. If one is looking for a basis of creating a fractal model of historiography though, established cyclical models is a good place to start, controversial as they might be.
    – user27618
    Aug 6, 2018 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.