30 But what constitutes one of those Venial sins of science? Perhaps hearing or reading unpublished data in an article you’re peer reviewing, which is related to your own work, and subsequently coming up with plans for your own research based off these ideas as soon as the original author publishes might be one such venial sin. Such behavior would give someone an undue advantage over others, obviously, but an ability to read good ideas and not connect them to your own research is not a power most people possess. It’s not right, but if one waits for the publication to arrive and suddenly begins doing the work, it doesn’t exactly fall within the cardinal ethics violations innumerate above. It is, I would say, a small sin and a thorny issue for many men and women who peer review articles.
Or, not publishing peaks below 1.2 ppm of your 1H NMR spectra because you know there is solvent (or, more likely, grease) in the spectra. Doing such a thing doesn’t change the data (assuming you’re confident that the peaks are indeed solvent) but it misleads people about the purity of the sample and, in any regard, it is just laziness not to purify a sample for NMR characterization in a paper. The same thing could be said about unfaithfully cutting a gel that you ran with some errant bands, or some of the innumerable shit you can do to microscopy data to make it look less “messy”… A number of reasonable excuses could exist to do this, the most likely being that the compound in question is no longer available and the only spectra that exists has solvent in it. Is this a good excuse? Does it matter that you’re omitting peaks which might suggest your sample isn’t totally pure? Does it even matter? Yes. Yes and, of course, yes. Grease doesn’t seem like a sufficient reason to resynthesizes a compound that takes about a month to make (though I have had to do it, and it fucking sucks). The idea, at least from the perspective of Organic Letters, in publishing NMR spectra is to show purity – omitting that is falsification, even if it is just ‘grease’. Where does it stop, then? Acetone? Ethyl acetate? Triphenylphosphine oxide? A few stray peaks that might be solvent but you’re not sure… behold the slippery slope theory.
The phrase “Typical Data” is sometimes used liberally to indicate “the best data”. While a well done study will include error in all calculations, it’s not uncommon for someone who does, say, microscopy in the study of cellular dynamics, to choose the best image from a series of images. Even though this image may be to the right of the bell curve in terms of illustrating a point, unless it’s a true outlier, is it a true ethical faux pas to do so? Arguably, yes, it is. On the other hand, it’s impossible to include copies of the several thousand cells and determining what is “typical” isn’t always obvious and thus lends itself to be a subjective process. Nevertheless, no one is provided with a copy of every cell in a field, so you have to take the author’s word.
Yields… lots of people fake them, I’m quite certain. I can, roughly, get the written yield of a reaction 80% of the time from the literature. Sometimes, I have to do the reaction five or so times and, on one of those occasions, my yield will represent what is in the lit. Other times, my yield never comes close to what is published. I’ve come to accept that some groups are notoriously bad about determining yield in reactions or just lie about them. Also, I consider anyone who doesn’t expressly state that a purification is challenging in the paper when, indeed, the purification is a real whore, to be a liar. Maybe not an awful liar, but a liar nevertheless.
These are annoying problems and pervasive. If you were to ask yourself if you’ve ever seen anyone do any of this shit, I would think most people would have to say “yes.” I’ve personally witnessed advisers suggest omitting parts of spectra because they’re ambiguous and irrelevant to the point. I’m a billion percent certain Tet Lett is a journal of Yield lies and certain groups (in certain countries) need to work especially hard at not fluffing that shit up. And, so far as I’m concerned, microscopy is a big black box where no real method to control “cherry picking” of images exist (that I know of.)
Back to top
Well if synthetics would report their average yield with associated standard deviation, instead of their best yield, you would eliminate one of those problems. Assuming the reaction was ran more than 1 time.
Assuming the reaction was ran more than 1 time.
So it would be prudent, then, to mention how many times the reaction was run. However, I don’t find a point in doing a reaction more than once if 1. you made a lot of it; 2. the focus of the paper is not on the synthesis itself, instead of the study of the molecule like many materials papers are. If you have *a* yield, and an NMR printout of the stuff, good enough for me.
I think, though, that such honesty would not go over well with reviewers, especially if the reviewers are hardcore organic chemists, and from the viewpoint of the authors, doing the reaction again would be a waste of time and resources.
Yields in the organic literature are often ill-defined. Is the yield calculated as % of theory assuming 100% starting material consumption (based on moles)? Or is it based on recovered starting material? Is it based on GC area? NMR integration? Post column mass?
Add to this the typical n=1 (time prep was run) and you quickly begin to realize that published yields are quite useless.
Any good author should of course mention what kind of calculation was used to determine “yield”. This is science after all, a certain level of exactness is required by researchers.
Mitch, I agree 100%…. the organic world has become quite complacent in certain areas.
Under-reporting of the synthetic procedure makes me want to poke other groups in the eye with a fork.
I love the recent example (in JACS) of “the preparation of 1-6 is described in the supplementary information”, and then when you look for the critical synthesis of 1, nothing!
Did you check the references in the SI?
Yes, they refer back to their original Science paper, but they very specifically discuss a “new method” for synthesizing 1. Then except for a brief description of some of the experimental details (what the heck is the five equivalents of sodium acetate for?) there is no other discussion.
*Sigh*
With all due respect to Mark Twain, being honest is a dangerous policy to have in a world full of liars, damn liars, and organic chemists.
Cultural attitudes have to shift at some point. We can’t just wait until they all die off.
“cultural” attitudes be damned- i’m afraid you’ll die unsatisfied if you think people will be more honest if the motive toward honesty is anything other than fear. Aside from orgsyn-style direct replication, it’s difficult to find markers of reproducibility in the organic literature- most of them are flawed in some way. I think one major issue is the method one’s work is quantiatively valued- in terms of literature citations. One can cite any sort of paper that has bullshit in it which may or may not be reproducible, so long as it had a particular impact to your research, whether it be in terms of inspiration, procedural precedence, or adaptation. In this way, bad science begets bad science- but bad science can beget good science, too.
But how often your literature is cited being the only real quantifier of a scientists’ impact in the field is silly, imo. And direct replication would be stupidly unnecessary and inefficient for things like total synthesis papers. What would be more useful is more objective standards- if your NMR has more than 5% solvent in it (and you will integrate those solvent peaks), no good. Want a more objective melting point? Take a DSC (talk to your friendly neighborhood polymer chemist- they have one). Elemental analysis is fucking stupid. And if you, as an editor, don’t think having common solvents in an NMR is particularly dangerous, then be more lax. Transparency in your data would be nice. If you don’t think all that hexane in your spectrum is a big deal, grow some and say it. It’s better than nothing.
I’d rather take a DSC than a melting point. So much more useful! It really helps to know where glass transitions are if you’re looking to do any device work.
I’ll believe a crystal structure over a solventless NMR any day.
(Some people might want to shoot me for this, but…some papers, I don’t consider worth reading unless I can scroll to the end and see a CIF listed in the available SI.)
To generalize your comment, I’d usually rather use a $50,000 instrument than a $200 one.
“the Lie, as a Virtue, A Principle, is eternal; the Lie, as a recreation, a solace, a refuge in time of need, the
fourth Grace, the tenth Muse, man’s best and surest friend, is immortal, and cannot perish from the earth………None of us could live with an habitual truth-teller; but thank goodness none of us has to. An habitual truth-teller is simply an impossible creature; he does not exist; he never has existed. Of course there are people who think they never lie, but it is not so–and this ignorance is one of the very things that shame our so-called civilization. Everybody lies–every day; every hour; awake; asleep; in his dreams; in his joy; in his mourning; if he keeps his tongue still, his hands, his feet, his eyes, his attitude, will convey deception–and purposely. Even in sermons–but that is a platitude.”
-Mark Twain, “On the Decay of the Art of Lying”
I did a similar post (although not quite as technically proficient) on seven deadly sins of science. Basically, I listed six – Plagiarism, Lying, Self-plagiarism, Vanity, Vituperation, and Procrastination, and left the seventh open for discussion. Having greasy NMR could be the missing sin…
db
I do apologize, I should have googled before I wrote. Kudos on beating me to the punch.
Since we’re low on comments on this post, I thought I’d throw in a totally random suggestion:
Someone should make chemistry t-shirts that say “Gotta have my JACS” with a picture of a box of Corn Pops but the lettering on the box says JACS instead of POPS. You could sell them on your Cafe Press if you still do that.
That might rub the ACS lawyers the wrong way…
Not if you write JCAS.
I love this post. Especially the BS that people publish when it comes to yields. Oh yeah, we got 100 mg of pure product from this reaction in 10 minutes…brsm. We won’t say that we had to dump 10 grams of our starting material in. I will say that when they publish the reaction procedure you can see it for what it is, but not every journal makes it easy to find procedures for all the important steps of a synthesis. Furthermore, people who report yields “by NMR” should be shot. Comparing two peaks together and getting a ratio does not a yield make. I understand that for combinatorial groups that make a bajillion compounds don’t have time to workup and purify all of their compounds, but at least you can give a representative yield after work up for one or a few of the compounds. damn.
I like the “enantiomeric ratio”
e.r. = 92:8 sure looks soo better than reporting your lousy 84% ee
There’s no problem with determining yields “by NMR” – just so long as you use an internal standard and do the math accordingly. Yeah, comparing the ratio of starting material and product signals just gives you a % conversion, not yield. Throw in a standard and you can calculate yields quite accurately. Unfortunately, you don’t often see the authors mention an internal standard (or what they used), so “NMR yield” usually means conversion.
Actually it does – unless one extends the interval inbetween the pulses, the signal integration is not very precise because some protons love to resonate little longer than the interval so they will come out smaller on integration. I frequently get ratios like 0.88 : 1 on unrelated protons, in spectra of a pure compound, on a 400MHz Bruker, with a clean baseline and a nice signal resolution.
Also, 3% of impurity is quire easy to miss in 1H spectra, depending on how sharp the impurity signals are
d1=10
problem solved. Yields aren’t calculated to any precision or accuracy anyway since, it would seem to be the consensus, that n always equals 1.
There are two answers to this –
1. It is good enough because we know NMR is an insensitive technique and the yields are an approximation when people were too lazy/sample not important enough/isolation too hard etc. Sometimes signal is lost in the dead-time between pulsing and acquisition or people don’t account for noise error/sidebands/signal overlap/quadrature detection etc. Round it down to the nearest 5% and its a good guess.
2. Its not good enough because people don’t quote the T1 relaxation time of the peak to be sure that all of the signal is being acquired. Particularly a problem in viscous solvent or a protic solvent when exchange is making some environments relax faster than others. If you do the right experiments with an appropriate added standard it is possible to get to below 1% accuracy relatively quickly compared to isolation.
hmmm. I don’t know how I feel about avatars. Avatars, consequently, was an unintended consequence of upgrading my threaded comments plugin.
Hey Kyle, no apology accepted. I think there was some prior art when I wrote about sins already
While we’re on the topic of venial sins, I have a question about the witholding of published data that someone on here might be able to answer. The crystal structure for an antibiotic was recently reported in J.Med.Chem., but if you go to the PDB, the coordinates (3CPW) are not set to be released until next April. I can understand holding back before you publish, but once the article’s out there, is it OK to hold back on the data? Is this just something they do with newly released PDB files?
Actually with the PDB you are supposed to publish the coordinates BEFORE you publish.
I am not sure if those sins are venial if the dispute is so eager. Unexact, forced or weak conclusions are simply bad. One should say only what there can be said about the obtained data – and no boosting it!