Chapter 8 Experiments & Experiences
Torgeir Holen, Hovedfagsoppgave, Bergen, 1998
8.1 The Negative Experiences
The negative experiences is a mostly neglected field in biochemistry laboratories. The
negative data are seldom reported in scientific journals (because of extreme shortage of
space) or even in this kind of novice thesis (because of the assumed lack of professional
style). This practice, I believe, results in novices and professionals alike having to deal
with their mistakes with no guidance from any theoretical framework.
To clarify matters, I do recognize that there exists a practical framework to convey these
negative experiences, for example by lab courses and consultations with tutors during the
laboratory work. As I understand this chapter is somewhat unusual, I will in the following
paragraphs try to illuminate my point of view, before I proceed to some of the
experiences from my experiments, as opposed to the results of the same experiments.
Thereafter I will discuss different ways to implement reports of negative experiences in
laboratories.
8.2 Negative, but Necessary
The negative experiences should not here be confused with some kind of bad experience.
We are talking about a state of confusion in connection with failed experiments that is
very familiar to everyone working in laboratories. Everyone speaks about the problems
one faces in an unfamiliar situation, but there seems to be no theorization around the
phenomenon. Now, some even claim that this kind of state is beneficiary because the
confused ideas might lead to new ideas, where an exact clarity is sterile for further
insight. I agree it is necessary for everyone to face this confusion during one’s training to
gain the insight of how to gain the information and clarity that is desired, but when one
has been a bit battered, so to speak, there should be a theoretical framework to be
consulted. It is said that cleverness is to learn from one’s own mistakes, but true
intelligence is to learn from others’. Then the information about others’ failures must be
available.
8.3 The Oral Tradition
There is no organized theoretical framework for these negative experiences that I am
aware of. That is not to say that the knowledge about the negative experiences is not
existing or even unavailable. Indirectly, the experiences are included in the extensive
method literature in molecular biology, but the statements only state what works. The
oral tradition in a lab, or a scientific community, contains these negative experiences and
there is much to be learned from consultation with colleagues. Still it is not structured or
organized in any theoretical way, except maybe in old peoples biographies an in tutors’
internal discussions, and as such the access to the knowledge therefore is random and
limited. Also, the knowledge in an oral tradition is very easily lost or distorted. This is a
very familiar fact to people working in the framework of any organization. It is probably
no coincidence that basic protocols vary so much between labs and that even high-tech
scientists can be superstitious.
8.4 The Question
Now assume that these negative experiences exist and are important to the everyday life
in the laboratory, and therefore should be theorized about. This is not an unproblematic
assumption as some will claim that everybody has to learn constantly. We are learning
creatures and we all meet new situations every day. Some will even say that writing about
these experiences would be stating something trivial. To this is to say firstly, that the
Emperor might in fact have no clothes, so learning about learning might be useful, in my
opinion valuable experiences are daily lost because there are no mechanisms of
preservation. Secondly, I am not writing a sterile, metaphysical treatise on learning in
general, but on how to master a biochemical laboratory situation as quickly and
painlessly as possible. If we assume that these negative experiences are factual, it is
interesting to note that older colleagues seem to master a new situation faster than
novices. My question is why this is so.
8.5 The Master’s Experience
The obvious, and therefore mistrusted, answer should be that these elders are more
experienced. Yes, but what is the nature of these experiences and how are they generally
acquired? Another explanation is that these are the people that are talented in this way of
making a living and as Masters of their trade they are naturally better at their work than
the through-going masses of novices. The ones that survive the lab-environment will tend
to stay and replace the old Masters. This Unnatural Selection, so to speak, is an attractive
mechanism and is probably part of the answer, but it avoids answering the real question.
What kind of acquired or innate talent made these Masters better than the average
novice? The third explanation is that these Masters have read and understood the
theoretical biochemical framework better than the average novice, and therefore is
superior in the lab. This explanation is missing my point entirely, as theoretical
knowledge as opposed to practical lab-knowledge are not necessarily overlapping sets. I
have no direct answers to the main question, but a suspicion that it is a mindset,
consisting of simple rules of thumb and recognition of certain situations, a recognition
that is acquired through many years of working in a lab. These simple rules and general
observations should be written down and be published for the greater good. Then again, it
is possible that the rules for describing a general, fruitful approach to laboratory life
might become so general as to be self-defeating. This is discussed in the implementation
part below.
8.6 This is not Science, is it?
I think most people would at this stage agree that there might be some philosophical or
sociological interest to these questions, but object that it is too vague and distant to have
anything to do with practical lab work for a novice like me. Furthermore, that Science is
something entirely different, and this chapter should not be in a thesis on molecular
biology. I beg to differ. Firstly, the everyday problems faced by a novice is not really
about hypothesis testing, it is not about whether a protein interact with DNA, and
certainly not whether the latest articles in our field have some relevant new discovery.
These are all very interesting theoretical problems, but the novice just want to know why
this protocol does not work, and has not worked for weeks.
This point has been fiercely disputed by my tutors, so I will try to clarify matters. Rein
Aasland and Litta Olsen, in their text “Experimental Design” (Rein Aasland, pers. com.)
model every activity in the laboratory as a micro hypothesis – an element of a larger
program – to reach the position where major hypotheses can be tested. This can be
viewed as some sort of Popperian position where science is modeled exclusively as
hypothesis testing. I do not dispute that it is possible to view science in this light. I even
tend to agree that it is desirable to do science this way. But I refuse to believe that this is
the whole story. Thomas Kuhn initiated the understanding that many activities in science
are too mundane to be considered hypothesis testing. I find the view that much of science
is problem solving (or puzzle solving) very attractive, and my view is that this is not in
direct conflict with the thesis that a central point of science is hypothesis testing. Aasland
& Olsen’s position is normative, mine is descriptive.
Usually the novice has done some faulty assumption, and it is the ability to make the
correct assumptions that is my central problem, and not the debugging of the protocol per
se. We do not have time to debug all the procedures and protocols all the time. The
Masters have somehow learned to avoid the major problems, while the novices are
spending huge parts of their time debugging apparently simple protocols or just keep on
pounding their heads against the wall repeating the experiments. The novices, and maybe
the ordinary scientist too, need some rules of thumb to recognize the kind of trouble they
are in and consult the right sources, something that the Masters probably do quite
subconsciously after years of practice.
The second objection, that these problems are not Science (with a capical S) is harder to
deal with. Of course one could ask what science really is, but then again one would have
to deal with Popper, Kuhn, Feyerabend, Latour and Merton, and very few of the world’s
hundreds of thousands of scientists do. They just know what Science is, because they
have been working in a laboratory and been reading Science and Nature for quite a while.
There has been no adequate general description of science. The theories from
philosophers of science are generally found too simple or distorted as seen by the
scientists themselves (Harald Trefall, lecture 1993). More recently and specifically, there
has until now been no working philosophy of biochemistry (or molecular biology), and
least not in the biochemistry itself (Roger Strand, 1997). So it might be that this is not
science, but I believe that these will make my experiments more reproducible. This point
alone probably justifies these considerations in a molecular biology thesis as some kind
of methodological discussion. I believe that true science is deeply immersed in these
practical problems all the time. I believe that is important for my lab, my fellow future
students and for my tutors to read about some of the experiences that I have gained in the
lab that are not producible in the table or figure of the usual experiments.
8.7 The assumptions
I have noticed that it is hard for skillful people to truly comprehend what novices in their
field are facing. The skillful have problems understanding why the central questions are
not formulated. The skillful are often not aware of the basic assumptions that have been
transcended on their behalf, like learning to ride a bicycle then never thinking of it again.
But the basic assumptions are the basis of their very action. Of course there is an
hermeneutical effect as one has to have some experience to gain an overview. Chess
instructors, for example, never answer questions in the start. They just tell the novices to
play. When some experience is gained, there is time to set these in perspective, a
theoretical framework. Then further experiences can be organized and further insight
gained more easily. In the laboratory there are so many assumptions as molecular biology
delves deeper into the mysteries of the cells, and all cannot be thought about at the same
time. In the laboratory, as in other walks of life, some experiences should be done alone,
but then one should consult some theoretical framework. I will in the continuation treat
three different episodes in my thesis work as examples in the light of our discussion of
subconscious assumptions.
8.8 Examples
Here follows three examples from my own experience with my thesis work, and what I
think I learnt from them. I think these experiences are soon forgotten and replaced by
other, revised memories (see section 8.5) so it is important to preserve them in this
detailed, original form.
8.8.1 Example One: The First Restriction Experiment
My first (very simple) autonomous experiment was to establish the size of the insert in
the EST plasmid (see Chapter Two), and as I at that moment was very naive – tabula rasa
in the ways of the lab – it might be a
good introduction example. I was given
instruction by my lab colleagues and
also consulted my journals from my
earlier lab courses. The techniques used
were the SRS web system, which gives
information about the EST plasmids
from the databases, Promega tables of
restriction enzymes and an agarose gel.
The specific restriction performed is
Fig.8.1. Restriction mapping of various
not important per se. Just have a look at
EST plasmids.
the gel picture (Fig.8.1).
Now I were to interpret this gel picture. What does the picture mean? At first I remember
running into all kinds of problems, because there were so many bands. Especially the
weak, lower bands were difficult to reconcile with some kind of cutting pattern. My
brand new tutor Rein Aasland threw away all my calculations and immediately pointed
out what bands were “real” and what bands were “ghost bands” or “artifacts”. The results
were tabulated (see Table 2.1, Chapter 2). My first assumptions about bands being bands
had been mistaken. This recognition of what is real and what is just artifacts is a typical
skill the Masters have.
Another assumption was that since according to the database-data the cDNA fragment
were originally cut and cloned into the EST-plasmid with EcoRI and HindIII, there were
not EcoRI sites in the fragment. This implies that the fragment coming out when cutting
with these restriction enzymes is full-length, that is, that my tabulated 0.7 kb value (Table
2.1) was correct (section 2.1.1).
Later I ran into problems, because the sequences pieces and sequences from the databases
made up a map that was larger than 0.7 kb. The inconsistency between the size of the
HuPcl1 cDNA and the EST-fasproject (see Chapter Two), remained not realized and not
looked into for a long time, due to the idea that size estimation on a gel can be somewhat
faulty. Maybe I should have realized the inconsistency earlier, but I think a more
fundamental problem is at large here. My conception of a HuPcl1 cDNA of 0.7 kb was
one of the ideas and concepts that are not daily criticized or thought of, but one which
consciously or unconsciously, one uses or bases ones work on, with the accompanying
faulty conclusions. The assumption is not questioned, because it can be defended by other
assumptions, as here about the limits to fragment size estimation.
Even this simple storytelling is becoming revisionistic as I also thought of the vector of
HuPcl1 as pBS-II-KS(+) instead of pBS-II-SK(-) (identical, but with the MCS reversed).
That is, the problems are multifactorial. Often, when none of several techniques are
familiar, the mistakes made in one technique lead to problems in the next. This leads to
some subtle inconsistencies that can be ignored for a long time.
8.8.2 Lesson 1
What do I feel that I learnt from this? First something about interpreting data: The gel,
even one as simple as this, was very difficult to understand unless one used the “conceptglasses” that let us distinguish between “real” and “ghost” bands. Secondly, I see that my
sequencing strategy was based upon a faulty assumption (the HuPcl1 fragment being
1000 bp, not 700 bp), which had long term implications for my thesis work. Thirdly,
there was enough data available to solve the problem at the time. They were just not used
(in fact I probably was unable to use them yet, and this efficient exploitation of all the
available data is certainly one of the Masters’ virtues). The sequence from the ESTdatabase included the EcoRI site. I just had not cared to learn about the EGCC (see
Methods 9.30) tools yet. The parallel gel show that the 0.7 kb fragment from the
XhoI/EcoRI and EcoRI lane are the same size, that is, an indication of an EcoRI site, but
this was not considered important. To cite Fleming on his discovery of the famous
penicillin colony on the bacteria plate: “Before you can notice any strange happening you
have to be a good workman, you have to be a master of your craft.” (Strand 1997, citing
Ludovici, 1952). The conclusion must be that it is not enough to have the data available.
One must also have training in interpreting them, and when to believe in them.
The fourth and last point there is something about self-esteem. One reason the problems
were not realized was because the inconsistencies (those that remained when one had
taken into account the limits of the method) were explained as non-orderly conduct my
myself. One is not so naive that one does not know one is a novice. Things do go wrong
in the lab, and one blames oneself, with very weighty reasons of course. But I think that
these phenomena, in company with the usual problems of working with biological
problems, lead to obscuring of the central problems with the method. Novices make
mistakes, but when the experiment has gone wrong several times, there are weighty
reasons of suspecting something more. My tutor Rein Aasland’s advice is that if
something does not work quite soon, one should introduce many controls and check
“everything”. I would like to add...then you check the system itself and your own
assumptions. Trust no one. Especially not yourself.
8.8.3. Example Two: The Probes
To investigate the expression of HuPcl1 and HuPcl2 radioactive probes were prepared to
be used on Håvard Valvatne’s Northern blots (see Chapter 4). Here first two RNA probes
were tried, without any result. Then a hot DNA probe approach was tried. The first time
there was only intense background signal
(Fig.8.2).
To illustrate that exposure to the film was
just the final step in a long series of steps, I
will describe them here: The plasmids were
isolated – cut – fragments isolated – probe
made – membranes incubated with probe –
membranes washed – membranes exposed to
film. This procedure could take 3-4 days (7 if Fig.8.2 A quite spotty-looking film.
What has caused it?
one include plasmid isolation and fragment
preparation). It is hard to see where
something goes wrong, especially if you are a novice. Also in this case I was working
with radioactivity, which tends to make me nervous.
Then I suddenly got a fine signal (see Chapter), but after this I was not able to repeat it.
The same probe on mouse-Northern blots turned out blank. New probes made with
HuPcl1 and then twice with HuPcl2 gave blank films, even after 4 days of exposure. The
questions were many. Was the washing procedure too stringent? Was something wrong
with the hybridizing buffers? Was there something wrong with the developing machine
again? Did I forget to heat the probe in my nervousness in the isotope room? Was there
something wrong with the DNA fragment? Was there something wrong with the enzyme
that should make the probe? I found out – doing
science on my lack of results, instead of on the
biological problem, a common lab situation – that
there was in fact no probe made, as most of the
radioactivity stayed in solution, and did not
precipitate out in ethanol.
When I returned to the making of the HuPcl1 probe
for use in the Lambda Phage System (see Chapter
5), the problems of the poor probes persisted. The
Fig.8.3. Control gel for DNA
problem was solved at a lunchtime chat, when Rune
fragment before probe synthesis. In
Male, and experienced colleague, pointed out that
hindsight the band has much less
my gelbands (the QIAGEX fragment control, see
than 100 ng DNA.
Fig.8.3) probably could not contain the 100 ng DNA
I had estimated. Thus the synthesis of the probes over the last four months appeared to
have been made difficult by a systematic mistake. The DNA concentration had been
based on the absorbance of 260 nm UV-light. When compared to DNA concentration as
estimated on the gel, there was an inconsistency of circa 3 to 10 times.
When the fragment concentration was increased 15 times, the probe synthesis was much
more efficient, as judged by the semi-quantitative ethanol precipitation method.
The (main) problem had been that I had used absorbance at 260 nm to estimate the DNA
concentration of my DNA preparations. This is an exact method as long as the DNA is
pure. If not, it is not trustworthy. I knew this. I knew that a ratio of 1.83 (absorbance at
260 nm versus 280 nm) was ideal, but that for example 1.89 and 1.76 was all right too.
My hypothesis at the time, because there is never time to solve any problem longer than
to get the method to work, is that RNA and protein (which raise and lower the ratio
respectively) cancel each other out and quite impure DNA can then have a ratio close to
1.83. * I had been using too little DNA by a factor of 5-10. But to estimate concentration
on gel, one also needs some experience, but that is another story.**
8.8.4 Lesson 2
What can be learnt here? I had been working in the lab for 7-10 months when I had these
problems. I had solved some major problems in sequencing and knew quite a bit about
that method. But my experience was not generalized, something I felt so had when I
entered the new field of making radioactive probes. This generalization-of-experience
idea is also one motivation for this chapter. There must be some way to convey these
experiences without having to use four months of your life and much research money.
Firstly, I learnt that a DNA concentration is not necessarily a hard fact. It is a
measurement-number that refers to a method. And sometimes different methods give
different facts, something that should make one suspect foul play.
Secondly, in hindsight, there were some hints in the laboratory practice that should have
warned me. Litta Olsen, the experienced lab technician, never used UV spectroscopy,
because she said she did not like the equipment (a Pharmacia Biotech, GeneQuant II
RNA/DNA Calculator), while Håvard Valvatne, the lab’s PhD student, used it and liked
it (he even calibrated it against pure oligonucleotide DNA). So in the lab there were in
fact two schools of thought about this particular machine. Some hated it, but I liked the
machine, so I never questioned the method behind my DNA estimations, especially since
the method was well described in our Bible, that is Maniatis. Yet again, the assumptions
were faulty. The problems different people encountered were explained by other
phenomena, for example that the machine was not easy to use, so any inconsistencies
were seen as mistakes or just a bad day for the machine.
Thirdly, I would recommend using a positive control for the first runs of any system,
even if it is a very system or a very expensive one. It is worth it in the long run.
8.8.5 Example Three: A plasmid was constructed...
I wished to construct two pair of yeast plasmids (pGBT-HuPcl1, pGBT-HuPcl2, pGADHuPcl1, and pGAD-HuPcl2) to express the cDNA fragments (especially the PHD
fingers) as hybrid proteins in the two-hybrid system, and thereby search for proteinprotein interactions (see Chapter 6).
The construction of the two pair of plasmids turned into a very lengthy affair. The origin
of the confusion was a combination of poor sequence data, difficult plasmids (pGBT9
and pGAD424), data errors and of course the learning process belonging to each battery
of techniques that had to be mastered (agarose gels, QIAEX, restriction enzymes,
Klenow, phosphatases, blun-end-ligation, electroporation, MiniPreps,
GeneConstructionKit, sequencing etc...).
In scientific literature one can frequently read, “A plasmid was constructed...”, if the
procedure is mentioned at all. The use of the passive voice hides the fact that the plasmid
must be made by someone. And that takes skill. When I constructed these two-hybrid
constructs I ran into all kind of problems. This was blamed upon old and impure plasmidpreparations, and in turn five different preps were used (two 1 year old MaxiPreps, one
MidiPrep, one MaxiPrep and one MiniPrep).
Restriction analysis with enzymes unique to the MCS, like EcoRI and BamHI, which was
essential to the cloning strategy, gave a fragment of 1000 bp where only the linearized
vector was expected.
This inconsistency can be caused by many things: The enzymes can be the wrong one
(switched by mistake). The buffer can be old (leading to unspecific cutting). The cutting
strategy can be wrong (one can overlook a restriction site). The maps (GCK-files) can be
wrong. The incubation can be too long, again leading to unspecific cutting. The plasmids
can be the wrong ones (we work with many plasmids).
I concluded (or assumed) however, after repeatedly getting the same results with many
different DNA-preps, that the 1000 bp band was caused by star activity of EcoRI or
BamHI, caused by some impurity in the DNA prep. Even our purest and quite trusted
MaxiPres had been thrown into doubt (we even speculated about trying out CsCl2
ultracentrifugation purification of DNA), because the some other problems with
MaxiPreps earlier (too dense or too much DNA from MaxiPreps can inhibit restriction
enzymes, and of course we had the fact from Example Two (section 8.8.3)), that our
DNA in the lab was not wholly pure).
But when different preps had been tried, and then plasmid had been fingerprinted, when
all the enzymes, buffers and incubations steps had been changed, one suddenly realized
(after the usual days of despair), that there was one assumption that had not been
questioned, namely the Promega Catalogue enzyme preferences. When I correlated the
Promega 1996 Catalogue and the New England Biolabs Catalogue on EcoRI, Promega
insisted that EcoRI is a medium ionic strength enzyme (100 % activity in the Buffer
Multicore, that is, 135 mM total salt, while in high salt (Buffer D, 150 mM NaCl) the
activity was reported to 50-75 %), while New England Biolabs preferred high salt (160
mM total salt.)
There must be a star-activity site in pGBT9 some 1000 bp from the MCS ***, which is
recognized by EcoRI under sub-optimal conditions. The fault is of course not Promega’s,
as there might be some difference between specificity and “percentage activity”. But
where New England Biolabs mentions that EcoRI does have potential star activity on low
ionic strength, Promega mentions that EcoRI has star activity in the absence of NaCl.
This seems strange in the light of the MultiCore-buffer containing only 50 mM NaCl.
When using high salt buffer (Promega buffer D), the problem disappeared.
In addition there were thee usual problems, for example, that at cloning strategy turned
out to be wrong (first discovered upon control sequencing afterwards), and that my
pGBT-sequencing file contained a vital error. These unrelated problems of course add to
the confused atmosphere when I was confronting the problem of the 1000-bp band. The
recognition of the essential problem among dozens of factors is always harder there and
then. In hind-sight, everything seems simple.
8.8.6 Lesson 3
A wise biochemist once said that two months in the laboratory often saves two hours in
the library. This joke is not as paradoxical as one should think. Long days in the lab with
endless protocols induce the wish to cut down on at least some of the steps. If can save
five minutes on each centrifuge step, days will be saved over a whole year. And this is
also a sound strategy as protocols sometimes have unnecessary steps (the protocols are
not perfect, they have the essential feature that they work, that is if you do them right). If
you can shave off the right parts of the protocol, then your understanding of the
procedure increase. But understanding is also needed to know where you must be
meticulous and where there is room for variation.
These changes to the protocols can also turn to sloppiness, especially if one is tired. And
in biochemistry the consequences of your detour form the protocol might be only be seen
three days later. And then, as our examples show, they can easily be interpreted as caused
by something else.
Thus, the same conclusions can be drawn: In the general confused atmosphere of the lab,
the problems might not be recognized as real, but thought to be a result of other factors,
or your own sloppiness. When you start debugging protocols, instead of doing real
science (as in working on the thesis hypothesis itself), some vital assumption is not
recognize before most other factors have been checked.
In this case the assumption was one seldom mentioned, namely that the Promega
Catalogue have their weaknesses too. Furthermore, the sequences from the databases can
contain errors (and when mistakes are transferred into protocols and lab-files, it takes real
effort to hunt them down – a longer story, not to be told here).
A vital understanding is that it is not hard to see the problems in hindsight (if you for
example look for salt concentrations and have the New England Biolab tables to compare
with), and that sequence mistakes are not to hard to spot, if you first suspect that they
might be there.
Another vital understanding is to recognize that these controls are easy (and fast) to do if
you just are aware that such problems might appear, and that you regard them as an
important, real problem. As I conclude this third example, I stress that this last point will
lead to a major topic in my discussion, namely that many of the problems encountered,
are easily avoidable if you just had heard about them previously.
8.9 Discussion
In the following I will discuss what can be learnt from these three examples in general,
and who this knowledge can be introduced into the lab life of a novice or a scientist.
8.9.1 The Recognition of Trouble
My original question was, how do the Masters tread to avoid all the traps? Do they have
some rules of thumb that they incessantly mumble beneath their breath as they work?
Probably not. They have some rules, but they are taken out of context. For example, “it is
important to have controls.” Sure... A seemingly trivial point. And that is sad, because
this rule is so true. However, one cannot have controls for everything, as there are so
many steps in the protocols and so many protocols. So it is important to know when,
where and against what does one need controls. In other words, the rule of thumb must be
set in a context. This context is the wordless understanding that I think the Masters have
gained through years of experience. They have seen (and heard about) so many situations
where it was necessary to use this or that rule of thumb. The important thing is in a way
not the rule itself, but more about what the rule is directed against, the misfortunes not to
be encountered. One need to know about what mistakes can be made, preferably in quite
some detail.
8.9.2 Wordless Experience is not Worthless Experience
My examples contain many words. I truly believe that experiences like these are very
common in laboratories around the world. Everybody has these experiences, but I am not
sure if they are remembered in words. They are remembered when needed, but not even
then in many words, but in a feeling of what would be the right thing to do, plus some
rule of thumb, often phrased almost like a cliché: “Run the experiment again!”, “Change
all the reagents!”, Check all your calculations!”, “Reread the protocol!”, “Go ask if
something is wrong with the machine these days!” etc...
8.9.3 Implementations
If I truly have a point in stating that the negative experiences from failed experiments
ought to be conserved for posterity and the greater good, who should this then be
implemented in practice? One way would be to have the ordinary scientist read more
biographies and philosophy of science, which do sometimes consider these or similar
problems, but this is quite unrealistic as the barriers between disciplines even inside the
sciences are remarkably high at time, not to speak of the “fuzzy stuff” in the humanities.
A local solution, however, here at the local Faculty of Science, would be to make the
compulsory Examen Philosophicum contain more of these everyday science experiences.
Still, these approaches tend to theorize too much, and one is left in practice with almost
sterile Popper-and-Kuhn storytelling.
8.9.4 The Sterile Articles
A major leap in the method of science was implemented when the first secretary of The
Royal Society of London for Promotion of Natural Knowledge, Henry Oldenburg, started
the journal Philosophical Transactions of the Royal Society of London in 1660 (Day,
1995, Online Britannica). IMRAD organization of articles (Introduction, Methods,
Results and Discussion) have because of large amount of information and a competitive
atmosphere, won support after the science boost succeeding World War II (Day, 1995).
The format will ease the burden for reviewers and readers, especially if one shall browse
many articles. Still, I think that although the format is great, it might be a bit sterile if one
is not only interested in the facts of the experiments.
Scientific papers that considered the basic assumptions, general confusion and human
fallibility, would be much more fruitful for the major goal of scientific papers, namely
reproducibility and further insight. This goal would be more easily attained if one
managed to convey some of the real, confusing atmosphere of the lab. But if one take one
look at the atrophied Methods & Materials parts of modern articles, one understands that
this is quite unrealistic.
8.9.5. The LabLore Site
As the understanding and experience we want to convey is hard to express in words (the
general rules seemingly become useless clichés), it might even not be desirable to include
a part in articles named, for example, “Major Problems Encountered”. So one other
solution is the way of conveying understanding by indirect storytelling. My main
suggestion is the initiation of a peer reviewed internet site, where stories, dialogues and
examples, like the three examples I gave, will be collected in a searchable form, that is,
on topics and disciplines.
Links to valuable information sources like PubMed, databases, discussion groups, library
services etc, should be collected. One example of an existing information collection site
is Rein Aaslands web site www.uib.no/aasland.
A new website, called LabLore, could be an accessible, easy way to other peoples
negative, and positive, experiences. This could be another crucial tool for the student of
molecular biology.
* Post script: Long afterwards I discovered that the MiniPreps of Aasland lab were
probably systemically contaminated with trace amounts of phenol – which do absorb 260
nm UV-light very strongly. Furthermore, my attempt at explanation is based on a
common lab myth – as written about extensively in the latest edition of Sambrook (aka.
Maniatis) lab manuals from Cold Springs Harbor Laboratories – it is true that DNA can
contaminate protein isolates and raise the absorbance ratio above 1.6, but the inverse is
simply not true, since proteins absorb so little UV light compared with DNA or RNA.
** PPS: Remarkably, the exact same story of faulty DNA quantification – people
claming the UV spectrometer being faulty, and the gel estimation of bands more reliable
– played out in the Seeberg laboratories in Oslo in 2003. What we learn from history is
that people never learn, even when told in very clear terms.
*** PPPS Years later, using plasmids from a later generation of yeast two-hybrid
plasmids and from Xenopus expression plasmids, the concept of difficult plasmids prop
up again and again. Star activity and correct restriction analysis is as important as ever, as
months are lost if everything is not perfect. And nobody ever tells you anything about the
plasmids FedEx’ed around.