Names are records.
Context gives them meaning.

What is in the data?

National and state coverage is reported by the loaded dataset.

These observations come from Social Security card applications. Earlier birth years are less complete, especially before 1937. Recorded sex describes the source category, not who may use a name. Different spellings stay separate; the source removes spaces and hyphens and may contain data-entry errors.

SSA background and qualifications · SSA downloadable data and privacy rules

A gap is not zero

SSA withholds small counts to protect privacy, including records with fewer than five occurrences. An absent name-year may mean no births or an unpublished small count; we cannot distinguish them. Charts break at gaps, correlations exclude missing pairs, and incomplete histories can make behavior measures unavailable.

National and state files are independently published. Adding state values may not reproduce the national total. Territory files exist separately; they are not imported here. The map uses modern reference boundaries.

Count, rate, and rank answer different questions

Published births

How many births are represented by the published count?

Rate per 10,000

Name count ÷ all published births in the same year and recorded sex × 10,000. This is a share of represented records, not an estimate of all actual births.

Rank

Names ordered by count within each year and recorded sex; alphabetical order breaks ties. Rank 1 is most used.

Period share

Published counts divided by all published counts in the selected years and sex selection. Spelling filters do not change the denominator.

See the population behind the names

How concentrated is naming?

Beyond one name

How do spelling patterns change?

Follow starting letters, endings or name length through time, then explore the names behind any year.

Reading a behavioral pattern

The examples below are schematic illustrations, not observations or predictions.

All / Combined population

All groups female and male published observations by spelling and year. Count is the sum of available published components. Rate is that sum divided by all published female plus male births in the same year (and state for maps), multiplied by 10,000. Separate normalized rates are never added or averaged.

A single published component remains a partial published aggregate: the other component is unknown. If neither is published, there is no observation. Combined rank is recomputed across combined spellings; ties are alphabetical. Coverage counts years with at least one published component. Trends and fingerprints describe the published aggregate, including partial years, not inferred total use. Suppression changes can affect these trends.

Behavioral fingerprint

Publication coverage is observed years divided by the window length. Longest run counts consecutive published years. Observed span is first-to-last recorded year, not proof of continuous use.

Momentum is the final three-year mean rate minus the preceding three-year mean. Acceleration is the difference between two such changes across three consecutive three-year blocks. Both require every observation in those blocks. Units are rate points per block, not percentages or predictions. Variation is population standard deviation divided by mean and requires ten complete years.

Profile fingerprints run from first publication through the latest dataset year. Reports recompute coverage for the selected analysis window.

Similarity is evidence of resemblance

Spelling uses letter edits, popularity uses the ratio of latest annual rates, and peak era uses years between lifetime peaks. Curves use Pearson correlation with at least 15 paired years and 80% coverage. Flat curves cannot be correlated.

Shifted comparisons search offsets from −60 to +60 years in five-year steps. A positive shift places the candidate later. Contrasting mode finds the most opposite eligible alignment; it does not show that every possible alignment differs. Resemblance does not establish cause or prediction.

Your local work and Alpha feedback

Shortlists and full-name previews stay in this browser. Two-person comparisons are tab-only by default; Remember on this device explicitly saves both people's notes and current selection here. Uncheck it to remove that stored copy. Export a backup before clearing browser data. Shared analysis links contain the selected public names and filters, not private notes. Reports are generated on this server from public observations.

Optional usage measurement is off until you opt in. It uses a session-only random identifier, action category, view, device class and coarse campaign label; it does not include searched names, notes, full URLs, IP addresses, or keystrokes. Feedback is an intentional submission; its form shows what context is included. Hosting and tunnel providers may retain their own operational logs.