DnD female names

Thirty female DnD names, six per race, five races, seed 20261130, styles rotated. Below the list: what the Female setting actually does to a name, counted over 3,960 rows.

• Newene Moonleaf High elf
• Siadheriela Fernbough High elf
• Firathile Moonshadow High elf
• Gacewyne Moonwake High elf
• Gicaemir Silverbough High elf
• Gidor Starshadow High elf
• Dinir Duskspire Drow
• Moquiraele Nighthearth Drow
• Zuviriss Webshade Drow
• Lisaeseth Shadowhearth Drow
• Donalor Nightcrest Drow
• Maemare Ashhearth Drow
• Nenus Ashenwrithe Tiefling
• Maetherathe Embermark Tiefling
• Midesosor Palemark Tiefling
• Banele Ashenshade Tiefling
• Drezisir Nightmark Tiefling
• Drimaesa Ashenveil Tiefling
• Meni Honeybarrel Lightfoot halfling
• Petudy Burrowleaf Lightfoot halfling
• Norebo Teabough Lightfoot halfling
• Doso Rootpot Lightfoot halfling
• Demolo Amberwell Lightfoot halfling
• Weser Honeywell Lightfoot halfling
• Gronar Stonedelver Hill dwarf
• Gaithomunda Grimvein Hill dwarf
• Mogradine Stonebeard Hill dwarf
• Hakroror Stonefist Hill dwarf
• Homudur Copperbeard Hill dwarf
• Vairin Deepbeard Hill dwarf

Six per race, straight off the seed — not re-sorted, not trimmed for taste. If one of them invites a character, take it; if not, roll two hundred and set the gender to Female.

Here is the part name lists skip. Female is not a filter laid over a finished list: it rolls one extra die per name and then, sometimes, adds a single letter. Over 3,960 same-seed rows that letter was an a or an e 1,479 times, an n, an r or a th 195 times, and no name changed in any other way at all.

What the switch does to row one

Same seed, same race, one setting apart. The first row of each batch shows where the rule sits: at the end of the given-name builder, after the stem is written and before the surname is drawn.

Five races, five first rows, and the given name survives in all five — once with an e on the end. The surname never does: it is drawn after the extra die, so it lands somewhere else in the same culture's pool. From row two on, every draw is shifted and the names are new.

Two rules and a die each

The whole gender setting is one if/else on the last letter of the finished stem:

if (gender === "Female" && rand() < 0.45 && !isVowel(last)) out += pick(["a","e"]);
else if (gender === "Male" && rand() < 0.30 && isVowel(last)) out += pick(["n","r","th"]);

Read it left to right: the die is rolled before the last letter is looked at, which is why both settings re-draw the whole batch. The letter is only added when the test on the right holds. Neutral takes no branch at all.

44.3% against a constant of 45%, and 31.4% against 30%: with 3,338 and 622 eligible rows the sampling error is 0.9 and 1.8 points, so both sit inside one standard error. The a/e split is a coin — 739 and 740 — and n, r, th come out 58, 65, 72, which is what a three-way pick looks like at this sample size.

Why the list restarts

One extra random number per name is enough to shift every draw that follows it. That is the mechanical reason a gender toggle hands you a different roster instead of the same roster relabelled, and it shows up as overlap: same seed, 1,000 draws per race, the 36 female lists held 35,370 different given names, of which 1,705 (4.8%) also appear in the neutral list of the same seed. Male shares 2,736 of 34,750 (7.9%) — the gap is pool size per race, not a different rule.

Row two is the proof of the shift. Of 2,160 matched rows, the female row two carried the same given name as neutral's row two in 3 cases, and male's in 8. Both are rebuilt names, not edits.

Longer by four tenths of a letter

The two length shifts follow from the gate, not from a style setting:

• Female: 0.45 x 92.21% x 1 letter = 0.415 expected; measured 7.388 − 6.976 = +0.412.
• Male: 0.30 x 7.79% x 4/3 = 0.031 expected, where 4/3 is the mean length of the three picks (1, 1, 2); measured +0.030.

Female also eats into the n and r endings it can reach: a name that ended on n or r now ends on a or e 45% of the time, which is why that column drops from 29.95% to 16.28% while the total consonant share halves.

By race

Everything above pools 36 races. Split apart, the same rule reads as two lines: races whose names always end on a consonant gain the full 0.45 letters under Female and nothing under Male; races with vowel endings trade some of that for the male n, r, th.

Largest miss on the female side: 0.053 letters. On the male side: 0.054. With 2,000 draws per cell the standard error on a mean length is about 0.03 either way, so a gap of 0.06 is two of those and no more. 24 of the 36 races end every given name on a consonant, so Male cannot add a letter there at all; each of them still lands within 0.06 of neutral, which is what a re-draw with nothing appended should look like.

Where the ceiling sits

The builder drops a syllable until a name is 10 letters or shorter, and only then rolls the gender die; the 14-letter trim runs afterwards, so an appended letter is never the one that gets cut. Neutral therefore tops out at 10 letters, Female at 11, Male at 12 — and the 72,000-draw run found exactly that: longest neutral 10, longest female 11, longest male 12, with 8 names at 11 letters and 3 at 12 in the male column.

Practical read: if you want the a/e endings, Female is the setting, and expect about every other name to take one — not every name. On a name that already ends in a vowel it never happens: the rule reads the last letter first and skips the branch.

Questions

Reproduce this page

var s=document.createElement('script'); s.src='/assets/female-stats.js'; document.body.appendChild(s);

Run it on the homepage, not on this page: the generator only loads there. It prints the pooled totals, the gate table and the overlap counts as JSON, and every figure above can be matched line by line against it.

Generate a roster · Orc names · Character name generator · Last names