Methodology: I counted the number of speaking characters per episode. Partly just as a matter of feasibility, I did not count: a) people who did not speak, b) people who spoke only in crowd scenes where it was difficult or impossible to determine who was speaking, and c) disembodied voices (which means that a couple of times I left Chevron Guy off the count for an episode he was in, which means in a few spots the stats should probably be worse). This also means that characters who appear at two different ages are counted as a single speaking character (characters who appear in two different universes/timelines/bodily forms are, however, counted twice).
Counting characters of color was, obviously, a judgment call. (I counted Sha're, Skaara/Klorel, all other Abydonians, Apophis, and Bra'tac, for instance, though I am not sure how all the relevant actors identify their own races.)
In addition to counting and calculating percentages, I tracked the Bechdel test (There are two women, who talk to each other, about something other than a man) and a race-Bechdel test (There are two characters of color, who talk to each other, about something other than a white person) for each episode. In a lot of cases this was also a judgment call, and scores range from 'Pass' to 'Pass?' to 'Fail' to 'FAIL'. Caps indicates that an episode included only one woman or one character of color who spoke.
I have my raw data sheet with comments on Bechdel and race-Bechdel pass/fails and lists of characters who spoke in the order they first spoke, if people want data to look at/quibble with/recalculate for their own purposes. I'm not posting it now because it's long and unwieldy, but I'm happy to share it if anyone wants it.
And with that: Results
|Episode title||Speaking Characters||Women||% Women||Bechdel||CoC||% CoC||race-Bechdel|
|Children of the Gods||26||3||11.5%||Fail||9||34.6%||Pass?|
|The Enemy Within||12||2||16.7%||Fail||2||16.7%||Fail|
|The Broca Divide||11||4||36.4%||Pass||2||18.1%||Pass|
|The First Commandment||10||1||10%||FAIL||3||30%||Pass|
|The Torment of Tantalus||10+2**||3||30%||Fail||1||10%||FAIL|
|Fire and Water||8+1*||2||25%||Pass||1||12.5%||FAIL|
|There But for the Grace of God||14||4||28.6%||Pass||3||21.4%||Fail|
|Within the Serpent's Grasp||12||1||8.3%||FAIL||3||25%||Pass|
* + Non-humanoid alien who could not be assigned a gender or race, excluded from percentage calculations.
** + Character(s) shown at two different ages, counted as one person for percentage calculations.
! There are two people of color but TEAL'C IS BOTH OF THEM, AND ONLY ONE IS ALIVE/CONSCIOUS/A BIOLOGICAL ENTITY AT A TIME.
ETA2: Adjusted the count for "There But for the Grace of God" to reflect the fact that Laara Sadiq, who plays "Technician," is Pakistani and therefore counted as a person of color. This relieves us of the !! level of FAIL. \o/
Conclusions: I am actively trying not to draw too many (or too sweeping) conclusions; in a lot of cases the numbers and passes/fails speak for themselves, I think.
One thing I did notice is that in the two episodes where Sam is the only woman to speak ("The First Commandment" and "Solitudes") the episode centers significantly on Sam. ETA: Just realized that's two of the three episodes in which Sam is the only woman to speak; she's also the only one in "Within the Serpent's Grasp" and has no special role there. In the
I also noticed that - while percentages of women and characters of color are mostly dismal across the board - episodes frequently manage to pass either the Bechdel or race-Bechdel test, at times alternating from one to the other over a string of episodes.
Also, I do not think I would have bet dirt, going in, that "The Broca Divide" would manage to be the best episode of the season in terms of the Bechdel and race-Bechdel tests. I guess it had to be good for something.
I briefly attempted to track the statistics on who died and who was shown in some degree of undress, but it became unwieldy and crazymaking, so I leave that to future inquiries by brave people who are not me, and move on to counting season two.