The Dissertation Guide lists the various rules for building and submitting a dissertation, so start there. The filing timeline is also very handy.

When you advance to candidacy after your qual you can drop one of your qual committee members (you only need 3 for your dissertation), but you cannot drop your outside member, or your adviser (who is your dissertation chair). Check the grad student handbook (above link) for details on who can and cannot serve on your committee. It is also relatively easy to change your dissertation committee at any point after your qual, just use this form.

If you have published papers, you can probably put together a “stapler dissertation”. The papers do not have to be on the same topic and you do not have to re-work them into a coherent whole. Just write a several-page introduction giving the appropriate background and references to demonstrate your understanding of how your papers fit into larger fields and then change the “title” of each paper into a “chapter”. That is a bit oversimplified, but really, you do not need a single, over-arching “thesis” with a big conclusion at the end. Published papers are prima facie evidence of doctoral-level competence, and if you have enough of them to make a dissertation, you are (nearly) finished. It also does not matter what you said you would do at your qual. Your committee will not care (or remember).

“Filing Fee Status” is a special kind of university registration which lets you file your dissertation sometime during a semester - but leaves you effectively unregistered in all other ways. Notably, you lose many regular student privileges such as Class Pass, use of the gym, and health insurance (but note that your spring SHIP insurance and Class Pass are good through most of the summer, until August 14 in 2010, even if you are approved for filing fee status before that date, so this is not such a big deal if you file before September). Then why would someone ever want to do filing fee? It's cheap ($185 as of 2007), far cheaper than registering regularly.

One case in which Filing Fee might be a good idea is if you already have a sweet postdoc lined up somewhere, but do not quite manage to get your dissertation done in time, so that you need to file in the first week of September, or early January or something like that. Rather than registering for an entire semester just so you can file your dissertation and then leave, you can pay filing fee, file your dissertation, immediately head off to postdoc-land.

When on filing fee status you can't be paid as a GSR, you need to get a special appointment like “junior specialist”. Once you actually file you will then have to change your appointment (a hassle) to get paid as a postdoc, if you are sticking around for a while. You also have to start paying off any undergrad student loans you've deferred.

As of summer 2007, university regulations have apparently been changed to require students to actually register for 3 units in the summer if they want to file their degrees in the summer. The web seems to indicate that you can't use Filing Fee to file in the summer, but that's actually actually not true: when you go on filing fee it's technically for the fall, but it retroactively applies to the previous summer.

Health insurance may be the last thing you want to worry about when trying to finish your dissertation. Luckily, you probably do not have to: Spring semester health insurance extends through August 14, regardless of when you file, so you are covered after you graduate. Fall term insurance extends through January 14. See this page at UHS.

If you want to buy SHIP insurance while on filing fee after SHIP expires, you can, but it's $1500 (as of 2006).

You will need the following files in your latex directory (on a Mac, it's often ~/Library/texmf/tex/latex/misc/ – if that directory doesn't exist, you can make it) or main thesis directory:

• ucthesis.cls
• tweak_ucthesis.sty
• bibsetup.sty
• aasmacros.tex
• the uct1x.clo and uct1x.sty files, where x={0,1,2}.

Chop off the “.txt” (and “.new”) at the ends of these filenames. Then, use this thesis.tex file as a template for the overall thesis document, which uses the UC Thesis environment and imports individual chapter tex documents from appropriate subfolders. Modify this frontmatter.tex file to make your appropriate preliminary pages. Use this topic.tex file as a template for individual chapter files (and put it in a folder called topic). Also, use this approvalpage.tex file as a template for your dissertation approval page, which your committee signs. These files should have enough stuff to make typesetting your thesis possible without having to learn too much of the guts of LaTeX.

If you have two co-chairs, then some of the files need to be altered to note this. In this case, use these modified versions of ucthesis.cls, tweak_ucthesis.sty, and frontmatter.tex. Just cut the “.cochairs.txt” off the filenames.

They will reject your dissertation if it does not have the right margins. The limits are 1 inch all around, and the page number must be no closer than 3/4 inch from the edge (and consistently placed and on every single page, even chapter pages). These are minimums, so you cannot go wrong if you give yourself some buffer room.

The margins set up in the above files should be correct (they have been accepted for at least two dissertations as of May 2010). If for some reason you want/need to tweak your margins, there is a command in ucthesis.cls, for instance, which reads “\topmargin xxxxin” where xxxx is some number of inches for the top column – change that until you have at least 1 inch. If you are using a Mac with Preview, you can measure the PDF margins by using the “select” tool in conjunction with “show inspector”.

The most common problems with margins come from formulae, tables, and figures. Tables and Figures are discussed below. Check all of your formulae (especially in-line formulae!) and make sure they conform! You may need to re-format some of them, or shrink the font.

If you use deluxetable because you have a huge table than runs over many pages, it won't play nice with the ucthesis class file. The biggest problem is that it creates a new table on each page, so the list of tables after the table of contents will list the table multiple times. It also has a habit of putting the last page of the table a page late, leaving text interrupting the table. To fix this, use longtable, a package included with latex. The linked website has all of the details.

If your tables run wide (which happens a lot with these fat margins) there are a few tricks you can try to get them to fit. The first is to make your columns closer together with this command:

\setlength{\tabcolsep}{0.05in}

Adjust to taste. If that's not enough, you can use the lscape package and invoke the landscape environment to make your tables landscaped (don't use the \rotate command; it's flaky). Here's how a table might begin:

\begin{landscape}
\tiny
\begin{center}
\begin{longtable}{l@{ }rccccccccccr}

It's all on the longtable website. A nice PDF with a description of the longtable package is at this link.

The downside to using the landscape environment is that you have to put it outside of the floating table environment. This results in LaTeX putting the table in that exact place relative to the text, meaning you could end up with half the page before the table being empty. Another alternative, at least for tables that are wide, but short, is to use the sidewaystable environment in the “rotating” package. Whether or not this would suffer the same problems as deluxetable when also dealing with very long tables is unknown.

For figures, make sure your figures are not too big. If you have played with axes and such you might find the figures are pushing your captions out beyond the margins. If this happens, open up the actual eps file in a text editor and play with the line that reads “BoundingBox:” and play with the numbers there. These numbers tell LaTeX where the edges of the figure are, so by tweaking them you can trick LaTeX into making the figure take up less room.

The default LaTeX effect on figures will be to put large-ish figures on their own page (even when text could easily fit around them on the page) and smaller figures on the top of pages, with text underneath. Feel free to have all figures take up their own page by making them fill the page:

 \includegraphics[width=13cm]{figure.eps}

though this is by no means a requirement. You can also force your figures to be on a “page of floats” (= no text) in the figure command:

 \begin{figure}[p]

Another useful command is in the “afterpage” package; once you have added this package to \usepackage in thesis.tex, you can then use the following command to force a figure to be on its own page:

 \afterpage{\clearpage}
 \begin{figure}
 ...
 \end{figure}

though again, there is no formal rule requiring a figure to be on its own page.

If you have wide figures that you want to have rotated sideways, you could use the landscape environment, but it may not be ideal. The problem is that you have to put it outside of the floating figure environment. This results in LaTeX putting the figure in that exact place relative to the text, meaning you could end up with half the page before the figure being empty. Another alternative is to use the sidewaysfigure environment in the “rotating” package.

The captions of your tables and figures will appear in the lists of figures and tables after the table of contents. If you'd like shorter table and figure (or even chapter!) names, use the optional argument in the caption (or chapter!) command:

\caption[x vs. y]{x, corrected for nonlinear charge 
dispersion as a function of redshift and partial 
ionization vs. y sinh (gamma) (uncorrected for GR 
effects past 4th order)}

If you're going the stapler route, one of your tasks will be to convert the latex files for your manuscripts into individual chapters of your dissertation. One potential pitfall is you are likely to have latex labels multiply defined in each of your documents, e.g. having “\section{Introduction}\label{sec:intro}” in more than one chapter. These multiple definitions can cause the references to be confused when processing your master latex file. To avoid this problem, Mike Fitzgerald wrote a Python script (convert_chap.py) to add a prefix to the labels and references in a chapter (e.g. “\label{sec:intro}” → “\label{vega_sec:intro}”).

The thesis.tex file gives you the option of including with each chapter an abstract and acknowledgments. You should not include a list of co-authors or a publication reference; it is your dissertation. (But see below for the forms to fill out for previously published material).

Dissertations must now be filed electronically, thus “saving of approximately half a million pages of paper per year, make Berkeley doctoral dissertations more widely accessible, and spare doctoral candidates the cost of purchasing archival-quality paper.”

The filing must be done online four business days before the end of the semester. Once you receive a confirmation email from the Degrees Office, you will need to bring them hard copies of the required documentation: two surveys, a release form, and the approval page signed by your committee. None of these documents needs to be on archival paper.

If your dissertation contains previously published material, you will need to obtain a signed letter from your committee chair (adviser) stating that this constitutes dissertation-level work. You will also need a letter from every co-author stating that it is okay for you to reproduce this material in your dissertation. Alternatively, a printed email from each co-author stating the same thing can be included with the application, sent to Graduate Services along with copies of the published material. If you have a very large number of co-authors, a letter from your adviser to the dean stating that this requirement is a hardship for you since all of the many co-authors can not be reached and that your adviser speaks for all of them in approving the use of this material may be sufficient. This has worked in at least one case (Dexter has a copy of the letter that worked in her records), but it has also been known to fail, so if you want to try this, start several months in advance.