My frustrations for the day – I’m co-author on a manuscript submitted to PLoS. We’re now trying to upload the final version but we’re hitting silly problems that are caused by PLoS seemingly being beholden to Microsoft.
The originate because I use Linux whilst my colleagues use Macs and Windows, so I use OpenOffice, rather than Word. There are minor problems caused by this: equations aren’t editable and references get screwed up (I assume because my colleagues are using EndNote: I haven’t tried it but I suspect we would have similar problems if I used Mendeley). Things would be much simpler if Word used an open format, but Word’s compatibility problems are so bad it’s not even compatible with different versions of itself.
The alternatives to Word docs are rtf and LaTeX. rtf is another format owned by Microsoft. At least it has a publicly available specification, even if this does keep on on changing. And fun as it would be for me, I can’t see me persuading my co-authors to start using LaTeX.
Anyway, we managed to work our way around those problems only to be hit with a new one. PLoS insist that figures are only submitted as tiffs or eps. Now, tiffs are big files and as the figures are all graphs, a vector-based format like eps is perfect: it doesn’t lose resolution as you change its size. And R, of course, happily makes postscript files.
So, I made the graphs and sent them off. But then we hit the next problem – PLoS insist that we use Arial as a font. I didn’t know this but Arial is a proprietary font developed for Microsoft, and which they only release for Windows and Mac. So, I can’t use the font because I prefer to use an open source OS, which PLoS doesn’t like.
One nice thing about Postscript is that it’s actually a text-based format. So I can open up the file in a text editor and see this:
%%DocumentNeededResources: font Helvetica
%%+ font Helvetica-Bold
%%+ font Helvetica-Oblique
%%+ font Helvetica-BoldOblique
%%+ font Symbol
%%+ font Helvetica
%%+ font Helvetica-Bold
%%+ font Helvetica-Oblique
%%+ font Helvetica-BoldOblique
%%+ font Symbol
%%DocumentMedia: special 504 360 0 () ()
%%Title: R Graphics Output
%%Creator: R Software
%%BoundingBox: 0 0 504 360
Simply replacing ‘Helvetica’ with ‘Arial’ might work (“might” because I don’t know if the fonts have to be embedded), so I suspect a small piece of code at PLoS’s end would fix it (in fairness it would probably have to be added to an already long to-do list). But should I really be having these problems? Shouldn’t PLoS be trying to help us? After all their core objectives are
- Provide ways to overcome unnecessary barriers to immediate availability, access, and use of research
- Pursue a publishing strategy that optimizes the openness, quality, and integrity of the publication process
- Develop innovative approaches to the assessment, organization, and reuse of ideas and data
Isn’t insisting on using proprietary formats that can’t be used on open source platforms providing “unnecessary barriers to immediate availability, access, and use of research”? It’s not difficult to fix – simply insist on openly available fonts like Helvetica.
There’s a lot of shouting in the Open Access community about how the big publishers have a monopoly (together the three biggest publishers own a bit less than 45% of journals – what does ‘monopoly’ mean, boys and girls?). And yet here is PLoS forcing us to use Microsoft products. Why isn’t anyone asking for the use of open standards in publishing?
Of course, the best solution is to persuade the academic world to use LaTeX. If you don’t fancy the typesetting equivalent of scrambling around on your hands and knees, you can always use LyX.
IIRC, I think I saw recently somewhere on PLoS instructions that fonts should be Arial (or at least sans serif). Try with Helvetica until they complain.
As for equations and (more particularly) references, just ask your coauthors to send you a MS version without Field Codes. If they refuse to use good editing software (I like OOo at the moment, as it works beautifully with Zotero)
But yes, it has also occurred to me that too many manuscript handling systems insist on crap software combinations. Word, RefMan, Endnote… they’re all so last century. And, more importantly, not freely available. Universities and research institutes could save a bloody fortune in licences if they switched to open source software.
If any given researcher really needs to use the fine points of Word (or Excel…) that are not available in OOo, they can pay for it out of their grants, as they would with any other specialist software.
I had a lengthy correspondence with PLoS’s graphics editor a year or two back, which resulted in me doing a ton of work to make my figures vaguely presentable in their preferred format, and disabused me of the notion that an ‘author pays’ publication model might try to make submitting manuscripts easier for authors!
But I agree with Mike – use whatever font you want in your figures, until someone actually complains.
RE: TIFFs have large file size – of course you’re correct but PLoS Med and PLoS One provide some of the most detailed webpages about figure preparation of any journal:
On these pages they describe how to reduce TIFF file size by resaving with LZW compression using the open source program GIMP:
I find this compression can reduce a TIFF file size from several MB to a few hundred KB with not much loss in image quality.
Tom – You’re absolutely right, but the question is, is all the fiddling around with different graphics formats a good use of my time when I can produce a perfect PDF direct from R, to the satisfaction of every other journal I’ve submitted to? (I know, I can produce EPS or TIFF from R too, but it takes somewhat more fiddling and has a tendency to mess up all my carefully set graphics parameters too.)
They were already complaining about Helvetica, this is how this rant started. I can see why they might want to avoid comic sans in papers, but why not allow a range?
Thanks for that, Tom. You’re right, they are detailed. Ah, they even explain how to get Arial into R (I had to poke around a bit to install Arial, but it worked).
It still doesn’t explain why I should have to do all this, though.
bq. It still doesn’t explain why I should have to do all this, though.
Consistency of appearance in the journal. And to achieve that, they probably have to appeal to the lowest common denominator, which is, unfortunately in this case, the one that most of the trained IT chimps in most research institutions are familiar with.
I exclude the IT guys in my current institute, who have been by far and away the best I’ve ever come across, offering genuine support in almost all cases, and eagerly providing open source software support to all wherever possible.
I’ve dug out the details of figures for publication in Nature. At the submission stage, all we ask is that the figures are good enough for the editor and referees to read, details are here: http://www.nature.com/nature/authors/submissions/subs/#a5
Note that we accept JPEG, GIF, PostScript (PS, EPS or PRN), PDF, TIFF and PowerPoint, and minimum dpi should be 150.
In the happy event that your paper is accepted, you will then be requested to provide publication-quality figures, following these guidelines (achtung! PDF): http://www.nature.com/nature/authors/gta/3c_Final_artwork.pdf
The section on file formats reads: “Acceptable formats include: AI, Vector EPS, layered PSD, postscript, PDF, PowerPoint, Word, Excel and CorelDraw (up to version 8). We cannot use the following formats: Canvas, DeltaGraph, Tex, ChemDraw, SigmaPlot — convert these files to PDF, EPS or postscript formats before submission”. I know from personal experience that although CorelDraw files are accepted they will lead to much gnashing of teeth and swearing under breath in the Art corner.
Finally for the text we accept Microsoft Word (preferred, although we accept only .doc not .docx), PostScript (PS, EPS or PRN), PDF, WordPerfect, Rich Text Format (RTF), TeX and plain text (TXT).
Clearly it would be nice if we could accept more open formats, but I don’t know how much extra work that would be on the server to make it possible.
Thanks. I wonder how many people provide their final version as plain text.
I’m a bit surprised you don’t accept formats like jpg, tiff, bmp and gif, especially because I’m sure a lot of the Word and Powerpoint files contain them. Are these all rgb formats?
BTW, I’ve just discovered that tiff is proprietary, it’s owned by Adobe.
I did wonder when I found that out as well!
Maybe I should clarify a bit: the overall file could be .eps, but the photo of a gibbon in panel a could be from a tiff (or even a high-quality jpg). The problem with jpg, tiff, bmp and gif is that the text is not editable, it takes ages for the art team to retype it all and can introduce errors (that your friendly subeditors should catch, but the fewer errors to catch the better). It makes changing lettering on a complex background a right b*tch as well.
All those formats are pretty old so I guess they’ll fall in the public domain rather soon anyway.
As a co-founder of PLoS, I want to weigh in here and say….
I COMPLETELY AGREE. From day one I have been frustrated by the reliance on proprietary data formats, and (another pet peeve) the conversion of vector graphics to bitmaps (something PLoS has finally fixed).
There’s all sorts of explanations for why PLoS hasn’t done a better job of this – a lot of it comes down to options offered by the software vendors who handle submissions. Our production people have been working hard over the years to lessen our reliance on these formats. But there’s ultimately really no justification for doing things this way – open science should use open formats.
Nicolas – ah, it all becomes clear. I forgot you have lots of photos with labels: I’m more use to putting up graphs, and thanks to R it’s usually straightforward to shuffle them around a bit.
Michael – thanks. The problem I had with Arial specifically was that there was no option but to use it, and there’s a simple alternative (Helvetica, which is also the default in R). it would be great if you supported .odt’s but I appreciate that it’s not trivial to implement.
Bob, this is indeed a strange problem. I don’t think authors should be worrying about fonts. And if you pick a font, at least use one that has no licensing restrictions. STIX fonts were created specifically for scholarly articles using a royalty-free license, but it seems as if almost nobody is using them.
There is another area where the PLoS journal submission system could do better: the author instructions should offer a download link to citation styles in CSL format (Zotero, Mendeley, etc.). Currently there is only Endnote and BibTeX.
Thanks, Martin. I guess PLoS know you’ll be wanting to push things like this now you’re working for them.
Not that I wan’t to contribute to a list of favourite fonts, but Red Hat released the Liberation font set under an open licence some years back, specifically as replacements for the main proprietary fonts shipped with other OSes. Liberation Sans aims to be metric compatible with Arial for example so even though they are not the same, they do look similar and the letters occupy the same space.
Whilst in an ideal world it wouldn’t matter what font was used, if one is stipulated, making sure it is one of these widely available, easily installable fonts would be of great benefit.
After a day tearing my hair out, it is a small consolation to find this blog. Today I have uploaded numerous figures to the PLOS submission site, TIFFs rejected because files too large, or in subsequent efforts, not of high enough resolution. I also submitted the figures in EPS (not a format I would usually use), which were rejected because of “missing fonts” (the Ariel issue I assume. If anyone can describe how I might overcome this issue in R (for Mac) I would appreciate it. I’ve been in contact with PLOS, but as yet, no reply. This is causing a real bottleneck in getting this paper published. Which is pretty ridiculous.
Ah, you bring back memories. Yep, pretty annoying. I assume this is somewhere low on the priority list for PLOS. And it’ll turn out to be more complicated than we’d guess to get it fixed.
Somewhere, deep in the PLOS instructions, they explain how to enable the use of Ariel, if you’re using *nix (which includes Macs). You might need to download the Ariel files, but if you google the filename (arial.ttf) and your OS, you should find the instructions.
I couldn’t agree more. I’ve just been through this process and it’s a complete fiasco. They offer LaTeX as a choice, provided you use their (poor) template. However it turns out this isn’t supported at all. I attempted to upload my final version just in latex and they said they needed the PDFs for proofs. Proofs? It looks *nothing like* the PDFs on PLOS’s website as their own template is single column with narrow margins amongst other things. In the end I had to hack their template myself to get something sensible looking to act as a proof.
Then it gets worse. They took the PDF and converted it to MS Word(!). During this process all sorts of bugs in the conversion cropped up, with table numbers changing and miscalculations of what is part of a URL and what is part of the surrounding text. Why on earth would you go from LaTex -> PDF -> Word in any publishing system?
It is really ironic, that they insist on using the Arial font, because most people will not be able to tell the subtle differences between Arial and Helvectica anyway. The 1982 Arial designed by Robin Nicholas and Patricia Saunders was “inspired” quite strongly by Helvetica (1957, Max Miedinger). The major ancestor of both fonts was likely the Akzidenz Grotesk BQ from 1896 designed by the Hermann Berthold foundry.