#     On serving as an Associate Editor
#                W.S. Harlan

In my first year as an associate editor for
Geophysics, I found that maybe one third of
all manuscripts were taking two thirds of my
time.  A few years later, I have been able to
distribute my time much more evenly with a
few preventive measures.  First, I make sure
that reviewers will be interested in the
paper and will not leave too many details to
me.  Second, I make sure authors understand
what is expected of all requested revisions.

Currently, only two reviewers are required to
examine a manuscript.  Two reviews are often
adequate, so this is a sensible minimum
number.  Nevertheless, I begin by asking four
reviewers.

The most conscientious reviewers are authors
who recently submitted a paper on a similar
subject.  Second best are those who have
published recently on related topics.  The
SEG's Digital Cumulative Index is a great way
to search title keywords.  GEOROM would be
even better.  I check the authors'
bibliography for related work but remember
that bibliographies may have important
omissions.  The SEG homepage finds member
addresses and even email addresses.  If not,
a recently published paper will have a
mailing address.  I avoid asking a given
reviewer more than once a year.

Most importantly, I try to be patient as I
look for candidates.  I prefer to avoid the
phone and ask by email or ordinary mail.  A
formal letter makes the potential reviewer
take the request more seriously.  On the
phone, the candidate might find it difficult
to say no, when in fact that candidate will
be too busy.  Some may instinctively say no
before they realize how interesting the paper
will be to them personally.

When I receive a new manuscript, I have the
abstract typed into a file.  This abstract is
incorporated into a form letter, which points
out that the work is "closely related" to the
potential reviewer's recent work.  I add a
few sentences describing how the work is
related.  I may emphasize that I want the
reviewer to address one aspect of the work
and other reviewers will address other
issues.  (Many papers combine diverse
specialties.)  The end of the letter says
that the manuscript will arrive in ten days
with at least three additional weeks for
review.  I ask the candidate to check the
mailing address and to respond by email, fax,
or my answering machine, if possible.  These
days, most respond by email.  Reviewer
addresses go into separate files that are
incorporated in all following letters, to
minimize typos.

If I have chosen the candidates well, then
two will allow me to forward their names to
SEG headquarters immediately.  Usually a
third will agree later.  I prefer to find a
third reviewer unless I am very familiar with
the subject.  If all four say yes, then so
much the better.  Later, if a reviewer takes
too long, I can rely on the two or three
reviews that have arrived.  If I end up with
just two perfunctory reviews, then it is my
punishment to go through every detail of the
paper carefully and do the reviewers' work
for them.  Next time, I ensure that I get
more reviews.  Extra reviews also make it
easier to arrive at a consensus.

For whatever reason, most papers submitted to
Geophysics seem to require significant
revision.  Sometimes it seems we fill a
publishing niche for those who have good
ideas but are not sure that anyone is
interested.  Not to exempt myself, I have
also submitted manuscripts that I knew needed
to be rewritten, just to see if I was wasting
my time.  I think we help such authors find
the essence of their papers so they can
finish.  Often the second draft of an
inconclusive paper improves enormously and
becomes an authoritative reference in the
field.  The author has finally found some
interested readers and writes directly for
those readers.  (Do we receive so many
unfinished papers because we are slow, or are
we slow because we receive so many unfinished
papers?)

My particular subject of inversion and
tomography may not be typical, but less than
a third of the manuscripts can be accepted or
rejected outright--it feels like less than
one-sixth either way.  When a paper requires
only minor revision, I simply express my
admiration.  When the reviewers and I agree
completely, a paper may be rejected as
inappropriate for Geophysics.  Occasionally,
I suggest that "we avoid the unnecessary
delay of a full review because the paper
could not possibly pass review in its current
form."  For example, a paper is entirely
theoretical and lacks any clear geophysical
application.  I might recommend another
journal like Inverse Theory.  Sometimes the
format of the paper does not remotely
resemble the description in the "Instructions
for Authors."  I may then ask that the paper
first be reformatted, or drastically
shortened, before review begins.  Reviewers
will feel their time is wasted if they are
forced to point out an obvious overwhelming
defect.

Perhaps two thirds of papers require
"significant revision."  About half of these
will be returned to one or two selected
reviewers after revision to see if their
"serious concerns" have been addressed.  Most
papers have at least one shortcoming that
could cause many readers to dismiss it
hastily.

All papers contain something of interest.  It
is not difficult to persuade authors to
restructure their papers and emphasize
different ideas, if they feel their ideas are
genuinely appreciated.  Reviews should stress
positive contributions of the paper so far as
possible and avoid dwelling on the negative.
There is no reason ever to say that certain
content is without value (it was of value to
the author).  Instead, we can stress that
other parts of the paper deserve a greater
proportion of space.  Almost all criticism
can be offered constructively, with a
realistic suggestion for improvement.  Never
should a comment appear to criticize the
author personally.

When the reviews arrive I read them all
first, then read through the paper, looking
for problems not mentioned by the reviewers,
particularly structural problems.  I then
collect my notes and write for the Editor a
summary review which has evolved into a
consistent format.  First I give the
consensus recommendation of the reviewers and
of myself for the level of revisions
necessary.

Next, I explain what I really liked about the
paper.  For example, I liked the thoroughness
of the analysis, the numerical methods, or
the examples.  The paper will be very
influential, will inspire new directions in
research, will have immediate practical
application, etc.  Authors discourage easily
and may overestimate the difficulty of the
revisions, so it is important to emphasize
that you really do like their work.  The last
sentence of my letter is usually something
like "I look forward to seeing the revised
paper in print."

Next, my review for the Editor summarizes
those revisions that must not be overlooked.
To avoid unnecessary second revisions, it is
important that the author understand what
changes are expected, particularly major
changes.  These are issues that must be
resolved.  I recapitulate any of the
reviewers' comments that cannot be ignored.
The first few paragraphs summarize these
changes and leave details for later.  If the
abstract is defective, I ask that standards
for abstracts be forwarded.

After the summary, the remainder is marked as
"Additional comments for the authors."  Here,
I generally explain too much, rather than
risk having to explain again after the first
revision has returned.  A little extra time
here could save three or four months delay
later.  I start typing into this section
immediately, so as not to forget anything.
All suggestions are phrased as constructively
as possible.

Any paper that carefully analyzes an actual
recorded dataset has my respect.  Even if the
processing methods are conventional, the
application will demonstrate how well these
methods performed in practice.  A good case
study will also reveal the most important
issues to investigate next.  The most common
weakness of such papers is that they claim
too much unequivocal success for their
approach.  Perhaps a literature search is in
order to discover and compare different
approaches to the same problem.  The authors
can be reassured that their approach need not
be the final word on the subject.  They need
only identify the assumptions and
circumstances that are appropriate for using
this new approach.  How should the method be
modified in different circumstances?  An
original approach is always welcome because
it gives readers an alternative they did not
have before, even if they continue to favor
other methods.  Maybe readers will be
inspired to modify or combine other
approaches.

My least favorite papers are those with a
totally artificial synthetic experiment, such
as a survey with ten sources, ten receivers,
and a simplistic velocity model.  Such papers
are now requested to include a "useful
geophysical application to a realistic
dataset."  The theory should not be
disproportionately more complicated than the
application.

Often authors emphasize the material that is
newest to them and de-emphasize those ideas
that they actually understand best.  They
need to be persuaded to rely on references
for familiar material.  Alternatively, if
some unusual methods have not appeared before
in Geophysics, then I may encourage the
author to add short tutorial explanations,
even if a full explanation can be found in
non-geophysical journals.  Readers must be
convinced first that these ideas deserve
further investigation.

Sometimes the most original ideas are buried
late in the paper, perhaps out of fear that
decisions were too unconventional.  Such
authors need to be reassured that they can
emphasize their new approach in the Abstract,
Introduction, and Conclusions, with more
discussion and justification.  Again it is
important to say that the chosen method need
not be the best of all possible methods: it
is enough to provide a new alternative which
works in situations described in the paper.

Sometimes reviewers are excessively critical.
A weak or slightly oversimplified assumption
is declared to undermine the entire work.  In
such cases, I explain that the author can
more carefully explain the reasons for this
particular assumption and give some
alternatives, with their advantages and
disadvantages.  The author shows how the
methods in the paper can be modified to allow
different assumptions.  Thus, a well-written
paper is rarely invalidated by a single bad
assumption.

I may simply contradict an unconstructive or
excessively negative comment from a reviewer.
Sometimes, reviewers point out a problem and
then suggest a patently impractical revision,
such as a total rewrite of all algorithms.
(I prefer to emphasize stronger parts of the
paper.)  Less hardened authors might simply
throw up their hands and submit to another
journal.  I do not usually withhold a review
from the author, but I do say when I disagree
with the review, or with the tone of the
review.

I have taken to including a standard
sentence: "Please respond to all suggestions,
either by an appropriate revision of the
text, or by a separate explanation of why the
suggestion is inappropriate.  If a reviewer
has misunderstood the manuscript, then I
encourage you to clarify the text: other
readers may make similar mistakes."

Another common sentence: "Algorithms should
be described with enough detail and citations
that a diligent reader could conceivably
reproduce your results."

My standard review letter includes optional
sentences for common corrections such as
adding an informative abstract, labeling all
figure axes, marking vectors and tensors,
equation numbering, reference style, etc.  An
oversight here could delay the final
revision.

If a revision is returned, and a major
suggestion has been utterly ignored, then I
feel entitled to send the manuscript back
without further review and ask for an
explanation.  Most authors will
conscientiously try to answer all questions.
Many will even number the suggestions and
highlight the corresponding changes in the
text.  A small number will try minimal
changes, just in case we don't bother to
check.

I do not ask a reviewer to look at more than
one revision of a paper.  After that it is my
problem.  Perhaps I did not state clearly
enough what changes were wanted in the
revision.  Very rarely, I believe a revision
essential and the author disagrees.  I make
my case and let the Editor decide.  Never, as
I once feared, has there been an irresolvable
disagreement about the ideas in the paper.
If the author has given a full explanation of
the reasoning behind the work, with adequate
citations, and acknowledging alternative
approaches, then I think it safe to let
readers decide for themselves.

W.S. Harlan, May 13, 1996