Research Ethics for Physicists: Keep a Detailed Lab Notebook
Full Chapter List - So You Want To Be A Physicist... Series
Part I: Early Physics Education in High schools
Part II: Surviving the First Year of College
Part III: Mathematical Preparations
Part IV: The Life of a Physics Major
Part V: Applying for Graduate School
Part VI: What to Expect from Graduate School Before You Get There
Part VII: The US Graduate School System
Part VIII: Alternative Careers for a Physics Grad
Part VIIIa: Entering Physics Graduate School From Another Major
Part IX: First years of Graduate School from Being a TA to the Graduate Exams
Part X: Choosing a Research area and an advisor
Part XI: Initiating Research Work
Part XII: Research work and The Lab Book
Part XIII: Publishing in a Physics Journal
Part XIV: Oral Presentations
Part XIII: Publishing in a Physics Journal (Addendum)
Part XIV: Oral Presentations – Addendum
Part XV – Writing Your Doctoral Thesis/Dissertation
Part XVI – Your Thesis Defense
Part XVII – Getting a Job!
Part XVIII – Postdoctoral Position
Part XIX – Your Curriculum Vitae
Table of Contents
Starting Your Research and Keeping Good Records
Now, where were we? Oh yes — you have started your actual research work. You and your adviser have agreed on at least the general area you will be working in, and you have begun substantial literature searches: what is known, what is unknown, and what are the current hot topics. At this stage it is important NOT to be rigidly fixed on a single narrow topic, especially during the early years of your project. Often both you and your adviser are still exploring before you narrow down to the exact, thesis-level project.
The best thing you can do now is build as wide a knowledge base as possible. For example, if you work in tunneling spectroscopy, don’t restrict yourself to one family of materials — study a range if available. Something that seems unimportant at first can become crucial later. Trust me on this.
Research ethics and record-keeping
In this installment I want to stress the ethics of doing research. I’ll illustrate from the perspective of an experimentalist, though many elements also apply to theorists. Ethical research practice is universal across the sciences; treat what follows as a case study to adapt to your own field.
When I ran undergraduate physics labs, I emphasized that students must write everything they did and observed into a laboratory book. They needed to record what they were doing, why they were doing it, and their observations — even what they were thinking about the data they collected. Was something peculiar? Was equipment malfunctioning? Were the data inconsistent or making no sense? Write it down.
How to keep a proper lab notebook
I insisted they write in INK and prohibited erasing. If they made a mistake, they should cancel the entry so it remained legible (for example, cross it out with a single line and add the correction). Why be strict about this?
- Primary reason: it helps you later. A careful record lets you reconstruct what you did (parameters, procedures, settings) without relying on fallible memory.
- Secondary reason: the lab notebook can be official evidence. If disputes arise about who did what and when, a dated, inked notebook is often used as definitive documentation. If you work for an institution (governmental, academic, or commercial), the notebook is the property of that institution and may be required in investigations or disputes.
For these reasons, always date your entries and use ink when possible. A good notebook lets others reproduce your work exactly and protects you in case questions arise.
Real-world consequences: case studies
There are documented cases where poor record-keeping had serious consequences. The Schön debacle at Bell Labs is a well-known example: his inability to produce written records of his experiments raised serious doubts about his results. That case led to his dismissal, the retraction of many published papers, and the withdrawal of his Ph.D. degree.
Although events like that are rare, smaller-scale problems from poor documentation are common. It is never too early to be meticulous about record-keeping. Even theorists should keep a dated record of ideas, calculations, and drafts; later you’ll be glad you did. And if you ever become a world-famous physicist, people might want to preserve your notebooks in a museum!
Moral of the story: keep as complete a record of your research work as possible.
Addendum
I know my timing sometimes seems uncanny, but this one is striking. In the online Nature news edition (22 September 2005) there was a report about a Japanese researcher whose lack of laboratory notebooks cast doubt on his published results. The article says:
“A respected Japanese scientist who failed to produce laboratory notebooks confirming his published results now faces a furor over the credibility of his findings. On 13 September, the University of Tokyo’s School of Engineering held a press conference to say that Kazunari Taira, a professor at the school who specializes in RNA research, had not provided raw data to verify his team’s results. The RNA Society of Japan has also questioned some of Taira’s methods.”
So, I’m NOT making this up: start learning to write everything down on paper when your work relates to your studies or job. It can be tedious, but you’ll regret not doing it if problems arise.
PhD Physics
Accelerator physics, photocathodes, field-enhancement. tunneling spectroscopy, superconductivity
@Andy Resnick, how about just getting the raw data off onto another computer for analysis? Sure, there’s lots of options available – I’m sure some people do that.
How do you back up the data drives that are not on a network? That’s the main concern- long-term data storage. AFAIK, there are no ‘best practices’ for our kind of data (meaning we don’t generate classified data, patient data, student grades, things that require a certain type of secured, guaranteed, access). For those types of data, there are resources like Iron Mountain:
http://www.ironmountain.com/
In my lab, we keep experimental samples for at most 3 years past the date the data appeared in a publication. Raw digital data is just backed up to an external hard drive and when the hard drive is full, we buy another- and it’s worth mentioning we generate maybe a few GB of data (images, mostly) per month. 4TB drives are cheap.
I also encourage students to affix images/printouts/etc. directly into the lab notebook as needed. Even so, use of a ‘secure’ lab notebook is becoming secondary to the security of the raw data itself, especially when stored on network-accessible computers. Personally, I don’t store any raw data on any networked computer, and periodically back up the raw data onto removable hard drives. My lab computers that perform data acquisition are not on our network. How do you back up the data drives that are not on a network?
This is getting absurd, and it is as if you are just trying to have an argument with me. I’m an experimentalist. It is not as if I don’t know these things that you are telling me.
You publish what is necessary to make your point across, and include everything to be sufficiently convincing. If it means publishing every single piece of data or calculation, then so be it! But more often than not, this is NOT necessary! The LHC need not publish every gazillion bits of data that it used in its analysis. It was sufficient for it to represent them in the figures and analysis that they published in a mere few pages. As a referee, I don’t ever recall any paper that I’ve evaluated that had to include ALL the raw data as a necessary part of the paper. This is what I meant as it not being “practical”. It isn’t necessary in the majority of the cases. I have seen pedagogical papers in AJP and EJP that included raw data, but those are intrinsic to the papers, meaning when necessary, they are included!
My original response to strangerep was to ask when has it ever happened that a theoretical calculation is too long to get published? If it is a necessary part of the paper, then I do not see any reason for there to be any reason it will be denied for being “too long”. The Andrew Wiles paper is a prime example, and even then, he certainly didn’t have to include every single step in his derivation!
Very few people read every single word or line in a publication! In fact, the overwhelming majority of us go directly to the guts of the paper, and the main points. If you want to know more, then you dig deeper, contact the authors, etc.. etc. It is why PRL is limited to only 4 pages, so that you get to the main points right away and get it over with. You publish only what is necessary, for whatever length that is necessary.
I don’t care to continue with this line of “discussion” in here, because this is going into nitpicking the misinterpreted details of what I said that isn’t even part of the original post!
Zz.
The distinction was made in the following paragraph to that statement.
No, the difference was reiterated in the following paragraph. The fact that something is theoretically possible to be published does not mean it is not too long to publish.
Also, you missed a factor of 10. I said the appendix was 1000 pages long. You invited me to pay page charges in Phys Rev for a 100 page article, and tried to use that as evidence that there was no such thing as an article too long to publish.
In fact, background data and calculations being too long or too large to publish are a usual thing rather than being rare. They occur in many experimental research projects where voluminous data is produced. Or where there are complicated custom designs involved in the equipment. Or where there are detailed and tedious theoretical calculations. Or where there are custom computer programs used to produce the results.
The usual method of dealing with this involves some kind of archive someplace. On request, a reader of the article can be provided with a copy. Possibly under a non-disclosure agreement when there are proprietary details. And possibly requiring the reader to pay duplication costs.
But the mere fact that there is such a thing as the usual method of dealing with it shows that it is not rare.
Recently, the company I work for did some work for a research lab. They have something like 50 people working in a program of research attempting to solve a particular problem. This program involves a series of projects. Each project involves several major chunks of equipment, and several tests, or even several series of tests.
Keeping track of the large amount of information involved would be a challenge. Making it available to everybody who needed it would be another layer of challenge. Their solution was to use a wiki.
Each piece of lab equipment gets a wiki. This records a variety of things such as the current condition of the equipment, the manuals for various procedures, the schedule for various procedures and tests, and how this equipment relates to other equipment.
Each test gets a wiki page or pages. This records the details of planning for the test, who will be on the test and what their job will be on the test, the schedule, the required equipment, the date and location of pre-test briefings, and after the test, the results and analysis.
There is an overall wiki for the project giving the strategy, the major milestones, the problems that are known, etc.
The wiki records the time and date and person who makes every edit. And it allows reversion of any edit. And it provides a friendly environment for everybody who needs to know stuff to be able to get it. And it provides a basic security scheme that lets certain parts be hidden from general view.
A difference without a distinction. The distinction was made in the following paragraph to that statement.
Zz.
But was it really “too long to publish”, or is it just impractical?
[snips]A difference without a distinction.
This is regarding a theoretical calculation being too long to publish. I encountered an example during my grad work. A colleague shared the fact that his thesis was 110 pages (or whatever it was), exactly the maximum length allowed by the university he did his PhD at. But the appendix that showed the details of the calculation was 1000 pages. But was it really “too long to publish”, or is it just impractical?
Remember that a thesis is supposed to have all the painful details. A journal publication should not! A thesis is also formatted differently than a journal, so the same content is often compressed in a journal format. I’m sure a Phys. Rev. Journal would be happy to publish a 100-page paper if you are willing to pay for it and if the reviewers deem it to be necessary and appropriate. So there is no barrier here in publishing something long. Look at Andrew Wiles’s paper related to “Fermat’s last theorem”. That isn’t short either!
Please note that this is a different topic than the original article, which is on keeping proper notes and lab book, not on physics journals. I’d rather not stray from that topic in this thread, because there is already another article on journal publications.
Zz.
When has this ever happened?
This is regarding a theoretical calculation being too long to publish. I encountered an example during my grad work. A colleague shared the fact that his thesis was 110 pages (or whatever it was), exactly the maximum length allowed by the university he did his PhD at. But the appendix that showed the details of the calculation was 1000 pages.
Well, one might ask “if I feed a dog chocolate, will something bad happen?”. I do not know of a specific case when this has been done, so the question would be hypothetical, at least for me. Nevertheless, the question is hardly pointless, since one is asking whether anyone else knows the answer.
But OK, I won’t bother you with further questions. The point is that sometime, somewhere, some person recorded a test of feeding chocolate to dogs. The events were almost certainly recorded in a laboratory notebook.
Not a good comparison, because that has been done before, and your vet CAN tell you easily the answer, even based on our understanding of the biology of a canine. The same can’t be said about your question, which has a closer association with a unicorn than a dog. Actually, all it says is that you are not the right kind of “vet” to answer this question. OK, that’s fine, you are an experimentalist.
But I just realized that there is a related real-world case: the calculation of Einstein, Infeld & Hoffmann many decades ago when they installed a copy of their tediously extensive calculations in a public library in case anyone wanted to check. So never mind, — I withdraw the second part of my question.
Bye.
Well, one might ask “if I feed a dog chocolate, will something bad happen?”. I do not know of a specific case when this has been done, so the question would be hypothetical, at least for me. Nevertheless, the question is hardly pointless, since one is asking whether anyone else knows the answer. Not a good comparison, because that has been done before, and your vet CAN tell you easily the answer, even based on our understanding of the biology of a canine. The same can’t be said about your question, which has a closer association with a unicorn than a dog.
Zz.
But if it is something that has never happened, then hypothetical or not, it is pointless. It is why we don’t discuss what, hypothetically, a unicorn prefers to eat.
I deal with real, actual issues, not non-existent hypothetical situations.
Zz. Well, one might ask “if I feed a dog chocolate, will something bad happen?”. I do not know of a specific case when this has been done, so the question would be hypothetical, at least for me. Nevertheless, the question is hardly pointless, since one is asking whether anyone else knows the answer.
But OK, I won’t bother you with further questions.
My question was hypothetical. But if it is something that has never happened, then hypothetical or not, it is pointless. It is why we don’t discuss what, hypothetically, a unicorn prefers to eat.
I deal with real, actual issues, not non-existent hypothetical situations.
Zz.
My question was hypothetical.
Yes, that’s all as I suspected — thanks.
What about in the case of theoretical research? E.g., if a calculation is too long to publish in a peer-reviewed journal? Would uploading a copy of the full calculation onto the arXiv or some other public server be sufficient to qualify as a legally acceptable “timestamp”? When has this ever happened?
Zz.
Yes, that's all as I suspected — thanks.What about in the case of theoretical research? E.g., if a calculation is too long to publish in a peer-reviewed journal? Would uploading a copy of the full calculation onto the arXiv or some other public server be sufficient to qualify as a legally acceptable "timestamp"?
From the legalistic viewpoint, must the records be ink-on-paper? What if one enters notes, etc, directly onto one’s laptop? If those notes were to be summoned into a court of law, would it be believable that it was in its original form and no one could have tampered with it?
The medium here is irrelevant. It is the ability to “time-stamp” and not have it tampered with is the crucial part. There are many projects now where there is an online, electronic lab book, where participants make entries that are permanent and time-stamped. So this is not an issue.
However, if you simply opened a Wordpad document, wrote notes in it, and then expect that to be a valid “lab book”, then you’d be mistaken. Furthermore, if this is done on your personal computer, this is also not acceptable unless this is your own personal research project funded by you. The person or institution that paid for you to do your work OWNS the work that you do. You cannot keep information and parameters of your work on your own personal device without a duplicate record of it that belongs to that person or institution.
Furthermore, and this is relevant to experimental work, a lab book is usually attached and goes with the experiment or equipment. There has to be a proper record and documentation on what was done on that system and when. This is crucial especially when there is a revolving door of people using that equipment. When something goes wrong, one of the first things people often do is look at what was done before and by whom. Keeping records for your own use is not only detrimental to diagnosing future possible issues, it is also selfish.
Zz.
From the legalistic viewpoint, must the records be ink-on-paper? What if one enters notes, etc, directly onto one's laptop?