Paranoia

Lab books are important but my electronic data even more so. So have your data backed up in multiple places, using multiple methods! This is what I do:

1. A hourly backup, using Time Machine, to an attached hard drive. This covers the "oops, I need that file back" as well as the "crap, my hard drive crashed" scenarios.
2. A weekly bootable mirror, using SuperDuper, to a hard drive stored in a file cabinet. This covers the "need to get back up and running RIGHT NOW" scenario.
3. A twice-weekly backup, using JungleDisk, to Amazon S3. This covers the "need access to my files from Timbuktu" and "holy shit, the building burned down" scenarios. (JungleDisk also archives previous versions of files, in effect providing a partial backup of the Time Machine drive; it's the lack of a TM mirror that currently makes me most uncomfortable.)

Excessive? No way. If I could--easily and affordably--have hourly versioned backups to the Internet and five-minute backups locally, I'd do it. But then, I also photocopy my field data books.

Update (March 16): I've added DropBox to the mix. This may replace JungleDisk in the future, but for now they're both running. This gives me the "hourly [or better!] versioned backups to the Internet" capability noted above.

Quiddler

While playing Quiddler you've probably wondered about the highest-scoring word, highest-scoring hand, scoring distributions for English words, and things like that. Wonder no more! By my calculation (using a short program in R), here are the highest-scoring words by words length:

   len    value cards           word
1    1        0     0          
2    2       16     2             za
3    3       28     3            zax
4    4       44     4           zizz
5    5       47     5          jazzy
6    6       55     6         zaqqum
7    7       58     7        buzzwig
8    8       61     8       quizzify
9    9       65     9      unquizzed
10  10       69     9     quizzingly
11  11       69    10    buzzerphone
12  12       67    10   wickerworker
13  13       68    10  quinqueverbal
14  14       60    10 fingerprinting

The maximum-scoring words:

       word len value cards
buzzerphone  11    69    10
crackerjack  11    69    10
quirquincho  11    69    10
 quizzingly  10    69     9

(Here's hoping you draw lots of Z's.) So words from 2-14 letters are possible, using 2-10 cards. This boxplot show the distribution of words values by word length:

Not surprisingly, scores increase with increasing word length until you get to 11 letters, when most words have a zero score (they require more than 10 cards to form), but a few can still be made using Quiddler's two-letter cards...until 15 letters, when even this trick doesn't help any more.

Finally, there is a hand with 109 points (not including any bonuses): "zizz" (44), "zizz" (44), and "zer" (21). The base word list used was /usr/share/dict/words...don't ask me what a "zer" is.

Fire history and woody debris

This graph shows the contribution of decaying CWD generated each year by wildfire to an overall estimate of the fire-generated CWD decomposition flux in Canada. (That is, each color shows the carbon flux from a different year's cohort, and the top line shows the estimated total flux.) No pre-1959 national statistics are available, so I've assumed a constant fire rate before then, equal to the 1959-1999 mean.

You can see how the low-fire 1970s resulted in a decline in the overall flux, but then in the 1980s and 1990s some big fire years raised the overall annual flux to around 50 Tg.

From the department of unusual names

It's come to my attention that there's a tree canker named Neonectria fuckeliana. Question for my plant path friends: was this named by someone who was really pissed off at a woman named Eliana? Inquiring minds want to know!

Bohr

We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct.

Niels Bohr

I'd like to come up with a suitably-crazy theory some day.

Worldwide soil respiration

Here's a map (estimated from a model, natch, but a model based on more data points than anyone has ever put together before). I am getting quite tired of soil respiration, but this has been a good project: one year from starting to assemble the database to (almost) a manuscript.

Oops, no scale! Will fix that.

The reshape function in R is very slow with large data sets. That is all.