A Unisex Name Is a Moving Target in Historical Data

“Unisex” is often presented as a permanent label. In the SSA data it is better understood as a measurement taken over a particular period. A name can have substantial use for both recorded sexes for decades, move from one side toward the other, or qualify only because a small minority of births appears on the opposite side.

Those cases are not interchangeable. The useful question is not simply whether a name has ever been recorded for both boys and girls, but how large each side is and how that balance changes through time.

The source itself sets an important boundary

The Social Security files expose two sex markers, M and F. That is a property of the source schema. It is not a complete model of gender identity, and a percentage calculated from those columns should not be presented as one.

On this site, “unisex share” therefore means something narrow and auditable: the minority side’s share of the recorded M/F births for a name. It describes usage in this dataset, not what every person with the name considers the name to mean socially.

Why a yes/no badge throws away too much information

Imagine two names. One has an almost even historical split. Another has ninety-five percent of recorded births on one side and five percent on the other. A binary “unisex” tag can label both the same way while hiding a major practical difference.

Percentages preserve more of the evidence. They let a reader decide whether they care about any documented cross-sex use, a meaningful minority, or something close to balance. There is no universal threshold that turns those preferences into one objectively correct label.

The split changes over time

Lifetime totals can hide a crossover. If a name was strongly associated with boys in its early record and later became common for girls, adding the entire history together produces a blended percentage that describes neither period very well. The two-line chart is more informative because it shows when the relationship changed.

The reverse issue appears with old names that retain a small minority use. A lifetime badge can make the name look more balanced than its recent cohort actually is, or less balanced if recent use has moved toward parity. A current cultural impression and an all-time statistic are different measurements.

Direction matters, but it is not destiny

Historical US naming data contains well-known cases where a name’s dominant recorded sex changes over time. Many crossovers are directional rather than stable: the two series converge, cross or diverge again. That pattern is useful evidence of how use changed; it does not authorize a forecast about where the name will sit twenty years from now.

Naming conventions react to culture, media, migration and fashion. The historical series can measure those outcomes after they happen. It cannot know the next trigger in advance.

How to read a unisex name page

1. Start with the actual share rather than the label. Ask whether the minority side is small, substantial or close to even.
2. Look at both yearly lines. A lifetime 50/50 split created by one male era and one female era is not the same as two lines that run together for decades.
3. Check the most recent years separately. The all-time total can be dominated by an older cohort.
4. Check spelling variants. Different spellings of the same spoken name can develop different gender associations, so grouping changes what question the chart answers.
5. Remember the SSA suppression rule. Very small annual uses disappear from the source before percentages are calculated.

Why this site keeps the percentages visible

The unisex landing page is intended to surface names with meaningful recorded use on both sides, but the individual page preserves the underlying counts and series. That makes the category inspectable: a reader does not have to trust a vague label when the evidence can be shown directly.

It also prevents a common mistake in list-style content. A name appearing on a generic “unisex names” list can mean almost anything. Here, you can see the exact historical record that caused it to be included.

What the data cannot settle

National counts cannot tell you how a name will be perceived in a particular language, community, school or family. They also cannot tell you whether an individual bearer’s gender matches the source marker recorded at birth. Those are outside the dataset.

The right use of the statistic is descriptive: it shows how the name was recorded across the historical US data available to the project. The broader source limitations are listed in the methodology.

How many names are genuinely shared, counted

A percentage scale is more honest than a badge, but it only helps if you know how the scale is populated. Across the 3,631 published names, sorting by the minority share — the smaller of the two recorded sexes as a fraction of the name’s total — gives this:

Three quarters of published names are recorded almost entirely on one side. Only 315 names — 8.7% of the set — reach a 10% minority share, and only 54 are close to even. Whatever “unisex” means colloquially, in this record it describes a small, identifiable group rather than a broad middle ground.

That is also why a threshold badge is the wrong instrument. Any cutoff — 10%, 20%, 25% — lands inside a continuous distribution and separates names that differ by a percentage point, while a name at 5% and a name at 0.05% both fall on the same side of it despite differing by two orders of magnitude.

The most evenly split names in the record

Among names with a substantial history, these come closest to an even split: Harley at 48%, Jackie at 47%, Casey at 45%, Emery and Quinn at 43%, Riley at 42%, and Jaime at 41%.

The largest genuinely shared names by volume are a different list, because size and balance are separate properties: Kelly (734,746 recorded births, 12% minority share), Jordan (638,153, 33%), Willie (600,693, 24%), Tracy (493,249, 14%), Taylor (457,707, 25%) and Leslie (434,672, 27%).

Notice how little the two lists overlap. A name can be near-perfectly balanced and modest in size, or very large and lopsided. Both are described as unisex in ordinary use, and they create quite different experiences for the person carrying the name — which is the reason the percentage stays visible on the page instead of being compressed into a yes or no.

The shared names are recent, and evenly split by side

The 315 names at or above a 10% minority share are not spread evenly through history. Their peak decades cluster heavily in the recent record: 87 peak in the 2020s, 51 in the 2010s, 33 in the 2000s and 34 in the 1990s. Together that is roughly two thirds of them in the last thirty-five years. The largest earlier group is the 1920s with 32.

Some of that is the same censoring artefact that inflates every recent decade — a name still rising has its peak at the edge of the data. But the effect is too large to be only that, and it fits the ordinary observation that sharing a name across the recorded sexes is more common in modern naming than it was in the middle of the twentieth century.

One number that is genuinely even: of those 315 names, 161 are recorded predominantly female and 154 predominantly male. Shared names are not, in this record, a one-directional phenomenon in which names travel from one side to the other. Both directions are represented in almost exactly equal numbers, which is worth remembering before treating any individual name’s drift as an instance of a general rule.