Are many figures in a paper considered bad practice?

Question

I am in the process of writing a paper for one of my PhD test series. As the test series was extensive, with many different materials and specimen, there are quite a lot of figures: images of the different specimen, graphs of the results for each type of material, images of failure analysis for each type of material etc.

A professor from another university, who was at our institute for a couple of days, came by my desk as I was writing the paper and (somewhat negatively) commented that it looked more like a comic book than a paper, due to the many images. Looking at other publications, my paper is definitely---at least so far---on the figure-heavy side, yet I feel omitting too many of them and instead putting them in a repository or supplemental material will impede the general understandability. Personally, I often would appreciate one or two more images that e.g. better explain the layout of a test specimen than a textual description.

Thus my question: is it generally considered bad practice to have many figures in a paper? I am in engineering, but am also interested from a more general viewpoint.

Answer 1

I feel omitting too many of them and instead putting them in a repository or supplemental material will impede the general understandability.

I'd urge you to reconsider whether this is the case; too much material will certainly impede general understandability because people will not finish your paper and will not find the important points in a sea of other stuff.

I think there might be an important missing step in your process which is distilling down the results to something that can be communicated to others in a form less verbose than the original raw data.

You can always provide additional details in another form: in supplemental material, in a repository like Zenodo, etc, so that people who do want all those details can access them while your more traditional reader can also take the main points from your paper. Remember that while you're putting a lot of effort into this paper, your readers will be limited in available time and will have many other papers to read in addition to yours.

Answer 2

there are quite a lot of figures: images of the different specimen, graphs of the results for each type of material, images of failure analysis for each type of material etc.

I am only guessing from your description, but this sounds like a case I have occasionally encountered in supervision, but also on conferences and similar: People are show everything they have done, and try to (consciously or subconsciously) impress by pure amount of work. They then run through their slides at an incomprehensible speed saying little more on each one than: “And here are the results for X.” The more interesting higher-level insights are then often omitted or lost to the audience who stopped listening.

The same thing can happen in a paper or thesis, and when it does, yes, this is bad. As you may find it difficult to realise this on account of familiarity, a litmus test would be to think about the respective talk and how much you have to say on an individual item. A good supervisor should also be able to spot this.

The remedy is usually some summarising figures, statistics, or principles, for example:

• I have created twenty types of specimen using adhering to principles X, Y, and Z. Here is an exemplary specimen to show how I implemented these principles. And here is another one to show an interesting special case. A list and description of all the specimens can be found in the supplement.

• Here is an exemplary failure analysis for material X. You can see Y and Z. What is most interesting to us is measure Ω, which we obtained by …. Here is an overview of measure Ω for all specimens and materials. As you can see, there is a considerable trend towards ….

• These examples suggest that there X begets Y. With this hypothesis in mind, we analysed umpteen further samples and could confirm this in all but one case. The exception is Z and can be explained by ….

Be aware that this kind of summary, generalisation, and abstraction may not be easy and you may invest a considerable amount of brainpower or require support with this. However, often it is also this kind of thing that makes for a good publication, thesis, or scientist.

Finally, mind that all of this is not to downplay the effort all the work you have done. But the point of analysing a lot of samples is most often to be able to make a more general statement that is not specific to a given sample – which is often comes with the applicability of reasonable statistics. So from “I have analysed umpteen specimen and material combinations” you shift to “I found a general principle through a systematic study of specimen and materials.”

Answer 3

As a statistical reviewer for a couple of medical journals, I like figures, and I like them in the main text. But journals may have limits. Unfortunately, this sometimes leads to authors making figures with many panels, all of which are then too small to see, and some of which aren't really related.

But I believe it is expensive to print figures (I could be wrong about that).

Answer 4

It's also worth considering the practical limitations of your target publication. Many journals have an upper limit on the number of figures/panels they'll allow. For digital publications there is basically zero cost to the publisher for adding more figures; the limit is there for the sake of readability. Depending on what you want to do with the work, "many" figures can indeed be "too many".

Answer 5

Ask your advisor, or other mentors or more experienced colleagues! Other answers (especially Wrzlprmft) give general principles on how to use figures, and how they’re sometimes overused. But a lot of the judgement on “how much is too much” is dependent on your field, subfield, and your specific work in particular. You’ll be able to get much better advice from someone who knows that context and your work, and can look at your paper draft.

To go with that, don’t put too much weight on one passing comment from a visitor. They may not have meant it very seriously, or thought very hard about it; and especially with the snarky phrasing you describe, they could just be someone with a critical personality, looking for faults in whatever they see. Critical comments like that (including from peer reviewers!) are always worth considering, thinking carefully whether you feel they’re right, and asking advice if necessary — but often after consideration, the right response can be “actually I like my paper how it is, and don’t agree with this criticism”.

Answer 6

I agree with you that in fact, many images that end up in the supplement would be fitting for the main text. The issue is that most journals have fixed formats for their papers - and this includes an upper limit for the number of figures. For student reports I often prefer to have many figures just in line with the text for readibility. So the question is: do you need to follow a specific journal format here or not? If it is the former, then consider this a good exercise in telling your story with space limits.

Answer 7

Figures are great!

I think that having lots of good figures can make a paper tremendously more readable. Figures are great for quickly conveying a general impression of a situation, and it sounds like that's how you're using them:

Personally, I often would appreciate one or two more images that e.g. better explain the layout of a test specimen than a textual description.

Figures can also make great landmarks: when I glance at a figure, I immediately know what part of the paper I'm in and what kind of stuff it's discussing.

Use figures thoughtfully

When I'm deciding whether to include a figure, my main two questions are:

1. Does the figure make the paper easier to read?
2. If so, is the gain in readability worth the work of drawing the figure?

It sounds like you're including the figures that you'd want to see if you were reading the paper, so your answer to the first question is yes. For figures that you've already made, the answer to the second question is also yes!

Since not everyone can see well, you should make sure that your paper is fully understandable with the figures removed. In my experience, it usually feels natural to do this, because because figures and text tend to complement each other: for example, a figure is a great way to describe an experimental setup quickly and qualitatively, and text is a great way to describe the same setup slowly and quantitatively.

Examples and context

Here are a few papers and books, grabbed off the top of my head, that I think use figures richly and effectively. I've tried to choose stuff that's relatively accessible.

• "Geometry of the Space of Phylogenetic Trees", by Louis Billera, Susan Holmes, and Karen Vogtmann.
• "Stripe Patterns on Surfaces", by Felix Knöppel, Keenan Crane, Ulrich Pinkall, and Peter Schröder
• The Road To Reality, by Roger Penrose.

My research background is in geometry, so maybe figures are especially effective in my area. However, I regularly read about physics, programming, and other topics, and I think that what I've said here applies broadly.

Answer 8

Using many figures is almost invariably better than cramming many subfigures into too few figures. That ends up with overly complicated captions and quite likely too-small text (some journals have a minimum font size in figures for this and related reasons).

The solution is to be selective. For that the input not just of more experience colleagues is useful, but less experienced ones. If they get lost in the mass of results, for example, you may have put too much in there.

Of course the comment that prompted this could have partly been to do with your writing style - if you put the figures in early in the drafting process, then write the text, the figures appear more concentrated than they will be at the end.