NJ School Data
← Reports

South Orange-Maplewood · data report

Where did SOMSD's Black students go?

The district enrolled 2,979 Black students in 1998-99 and enrolls 1,444 today. This report puts every number we hold about that change in one place — what the schools count, what the census says lives in the two towns, where resident students actually go, and the two definitional breaks that make the plain comparison wrong.

Draft · evidence reportEnrollment figures are exact administrative counts. Every population figure is a survey estimate and is printed with its 90% margin of error; differences are called only when they exceed the margin of the difference. Two race-category changes — NJDOE's in 2006-07 and the Census Bureau's in 2020 — sit inside these series and are drawn, never spliced across.
21.0%Black share of SOMSD enrollment, 2025-26 (48.7% in 1998-99)
31.0%Black or Two-or-More share, the widest honest current definition
23.9%Black-alone share of the two towns' residents, 2020-2024 ACS
9.6%of resident K-12 students attend private school — flat for 15 years

The claim that occasioned this

Black student enrollment in the district has fallen to roughly 22% down from about 54%, but the overall Black population of South Orange-Maplewood has not dropped at such a significant rate. That means Black families that choose to call the SOMA community home, do not have faith [in] the dubious legacy of the school district and are sending their kids to private school in greater numbers year after year.James Davis, president, Black Parents Workshop, August 202616

That is a chain of three links, and they can be tested separately. The first is an enrollment fact, and the district's own records settle it. The second is a population fact, and the census can speak to it. The third is a causal claim about where children went, and it is the one the data can rule out at the size claimed.

What the district counts

“Roughly 22%” is right. SOMSD was 48.7% Black in 1998-99 and is 21.0% today.1 “About 54%” is not a district figure in any year we hold: the series tops out at 51.5% in 2003-04. It is very close to the Columbia High School share in the late 2000s, and it may be the district figure in the early 1990s, which this series does not reach.

The more important correction runs the other way, and it favours the letter. The usual rebuttal to a falling share is “the denominator grew.” Here it barely did. Decomposing the 27.7-point fall from 1998-99 to today: 25.1 points (91%) is 1,535 fewer Black students, and only 2.6 points is the district growing by 754 students. The share collapse is a collapse in count. Meanwhile total enrollment is up 12% over its own history — White enrollment +1,047, Hispanic +545, and Two or More Races +685 from a standing start. The premise that district enrollment is falling is wrong. One group's enrollment fell.

BlackBlack + Two or More Races (2006-07 onward)
Black students enrolled in SOMSD, 1998-99 to 2025-26Two lines. Black alone falls from about 3,000 to 1,444. Black plus Two or More Races, available only from 2006-07, ends at 2,129.01,0002,0003,000students1998-992001-022004-052007-082010-112013-142016-172019-202022-232025-262006-07: NJDOE adds “Two or More Races”1,4442,129
The count, not the share, is what moved. Black enrollment falls from 2,979 to 1,444 while the district grows. The dashed line adds Two or More Races, which NJDOE did not publish before 2006-07 — it starts at the seam because there is no earlier value to draw, not because it was zero.Source: NJDOE Fall Enrollment1. Exact administrative counts; no margin of error.

The definitional break the letter does not account for

NJDOE published no “Two or More Races” category before 2006-07, and the release records the category as structurally absent rather than zero in every earlier year.1 SOMSD reported 0 in 2006-07 and 685 in 2025-26 — 10.0% of the district. In a community with SOMA's rate of interracial households, an unknown but non-trivial share of those students would have been recorded as Black under the older form.

That puts a floor and a ceiling on the real decline, and nothing in any NJDOE file can locate the truth between them:

Measured from the 2001-02 peakThen2025-26Change
Black alone, as reported3,2901,444-1,846 (-56%)
Black + Two or More Races3,2902,129-1,161 (-35%)

The direction survives either way. The magnitude does not survive intact: the honest range is a 35% to 56% decline in Black students against a district that grew. Both anchors are published in the data file, because choosing the peak year rather than the first year of the series makes the decline look larger, and the reader should be able to see that choice.

BlackBlack + Two or More Races (2006-07 onward)
Black share of SOMSD enrollment, 1998-99 to 2025-26Black share falls from 48.7 percent to 21.0 percent. Adding Two or More Races from 2006-07 puts the current figure at 31.0 percent.0%15%30%45%60%share of enrollment1998-992001-022004-052007-082010-112013-142016-172019-202022-232025-262006-07: NJDOE adds “Two or More Races”21.0%31.0%
The same two definitions, as a share. The gap between the lines is the whole of the Two-or-More question: 21.0% of the district is recorded Black, 31.0% is recorded Black or multiracial, and where a given student would have landed under the pre-2006-07 form is not recoverable from any file.Source: NJDOE Fall Enrollment1. Shares of total district enrollment.

Where the change happens: at the kindergarten door

NJDOE does not publish race by grade. But SOMSD's schools are grade-banded, so school-level rows give it: six elementary schools, two middle schools, one high school.1 The gradient is monotone in every one of the 28 years — the younger the band, the less Black it is — and the elementary schools turn down roughly a decade before the high school does.

Columbia HS (9–12)Middle (6–8)Elementary (K–5)
Black share of SOMSD enrollment by grade band, 1998-99 to 2025-26Three lines. In every one of the 28 years the elementary schools are less Black than the middle schools, which are less Black than the high school. Elementary turns down about a decade before the high school.0%15%30%45%60%share of enrollment1998-992001-022004-052007-082010-112013-142016-172019-202022-232025-2626.6%20.8%17.9%
Today's elementary schools are the district's forecast. In 2010-11 the elementary schools were 30.5% Black while Columbia High was still 51.8%. Elementary enrollment overall is essentially flat across the whole span (2,996 then, 3,126 now) while its Black count fell from 1,342 to 559. That is the signature of a change in who enrolls at entry, not of students leaving. The high-school line is a lagging indicator.Source: NJDOE Fall Enrollment, school-level rows1. Montrose Early Childhood Center is pre-K and is counted in no band.

What the two towns look like

The district's ACS geography and the two municipalities are the same ground — Maplewood plus South Orange Village sum to 44,087 residents and the school-district geography reports exactly the same number, so the boundary mismatch that normally caveats this comparison does not arise here.2

South Orange + Maplewood, 2020-2024Estimate± 90% MOEShare
Black alone10,535± 1,42523.9%
Black alone or in combination12,093± 1,49927.4%
Residents aged 5–17, Black alone1,758± 406—

Set against a district that is 21.0% Black and 31.0% Black-or-multiracial, the school population and the resident population are close to the same community. On the headline numbers this is not a school system that a resident population has abandoned. But the age structure tells the other half, and it is the half that matters.

School age (5–17)Adults
Black residents of South Orange and Maplewood, per single year of ageResidents per single year of age. The school-age bands run about 95 to 145 per year, while the 35-44 and 55-64 bands run about 150 per year. The largest cohorts are adults.050100150200residents per year of age95Under 51455–912410–1413815–1715035–4410845–5415155–647365–74ages 18–34 not shown
The published age bands are unequal widths, so every bar here is residents per single year of age. Read raw, a ten-year adult band towers over a three-year teenage band for arithmetic reasons alone. Read per year of age, the two largest Black cohorts in the two towns are people aged 55–64 and 35–44. The cohort entering school is about a third smaller than the one now in elementary school. This is a population that aged in place.Source: ACS 2020-2024 5-year, table B01001B2. Estimates carry margins of error of ±177 to ±290 on the underlying bands4.

The comparison that matters, and the break inside it

Put the two sides on the same years and the same category. This is the chart the whole question turns on — and it is also where a second definitional break lands.

The Census Bureau changed how it codes race, starting with the 2020 survey year

Write-in capture went from 30 to 200 characters, and up to six detailed codes are now assigned per write-in area instead of prioritising multiple responses into two. The Bureau's own instruction is that “any data comparisons to prior race estimates should be made with caution, taking into account the improvements we have made to the race question.”67 A five-year vintage labelled 2020-2024 is entirely post-change; 2015-2019 is entirely pre-change; the four windows between them are mixtures. The change pushes the same way NJDOE's 2006-07 change did — out of a single race and into Two or More Races.

Black aloneTwo or More RacesCombined90% margin of error
Resident students by race, ACS 5-year estimatesBlack alone falls from about 2,860 to 1,760 while Two or More Races rises from about 320 to 1,395. Added together the two are flat.09001,8002,7003,600resident students K–122008-20122010-20142012-20162014-20182016-20202018-20222020-2024ACS race coding changescombined 3,155Black alone 1,760Two or More 1,395
On the resident side, the fall in Black-alone students is matched almost one-for-one by the rise in Two or More Races. Black alone falls from about 2,860 to 1,760; Two or More rises from about 320 to 1,395. Added together the two are flat across thirteen windows — every value inside every other value's margin. The steepest part of the divergence sits exactly at the coding change.Source: ACS 5-year, tables B14007B and B14007G, residents enrolled K-122. Shaded bands are 90% margins of error. Consecutive 5-year windows share four survey years and are not independent observations.

That is the measure to carry across the seam. So here it is against the district's own count of the same combined category. The two sides are joined through the NCES district identifier — ACS geography 3415330 to NJDOE 13-4900 — parsed from an identifier and never matched on a name.3

Resident students (ACS estimate)Enrolled in SOMSD (NJDOE count)90% margin of error
Black and multiracial students: residents against district enrollmentThe ACS estimate of Black and multiracial resident students is flat near 3,100 across the window. The district's own count of Black and multiracial students falls from 2,730 to 2,194.01,0002,0003,0004,000students2008-20122010-20142012-20162014-20182016-20202018-20222020-2024ACS race coding changes (2020 survey year)3,155 ± 490 resident2,194 enrolled
The resident count is flat; the district's is not. ACS puts Black and multiracial resident students at 3,181 in the 2008-2012 window and 3,155 in 2020-2024 — a difference far inside the margin, so no change is demonstrated. Over the same years the district's own count of the same combined category fell from 2,730 to 2,194, 20%. Do not read the gap as private school — it also holds charter, out-of-district special-education placement, county vocational, homeschooling and survey non-response, and this data cannot split them.Sources: ACS 5-year B14007B + B14007G2; NJDOE Fall Enrollment1. The two race measurements are not the same instrument — see below.

The gap in the current window is 961 ± 490 students — large enough to clear zero. Its apparent growth from 451 does not clear the margin of the difference (± 643), so a widening gap is consistent with this data and is not demonstrated by it. That is the strongest statement the numbers support, and it is weaker than either side of this argument would like.

One caution that does not go away

ACS “Two or More Races” and NJDOE “Two or More Races” are not the same instrument. ACS is a householder's self-identification on a survey form; NJDOE is what a registrar recorded at enrollment. In the current window ACS reports roughly twice as many multiracial resident students (1,395) as the district records among its own (644). Combining the two categories makes the sides more comparable than the single-race lines are. It does not make them identical, and no join can.

The private-school question

This is the letter's one causal claim, and it is the one the data rules out at the size claimed. The entire private and parochial sector serving both towns, every race, kindergarten through grade 12, is 774 ± 156 students.2 Davis's account requires roughly 1,700 Black students to have gone somewhere other than SOMSD. Even if every private-school student resident in South Orange and Maplewood were Black, the sector is less than half the size the claim needs.

Private share of resident K-1290% margin of errorIndependent windowHollow = overlapping window
Private and parochial share of resident K-12 studentsThe private share falls from 15.9 percent in the 2005-2009 window to 9.2 percent in 2010-2014 and is flat afterwards at about one student in ten. Only the first fall clears its margin of error.0%5%10%15%20%share of resident students2005-20092007-20112009-20132011-20152013-20172015-20192017-20212019-20232020-202415.9%9.6%
The private share fell by a third and then held flat for fifteen years. The only movement in this series that clears its own margin is the first one: 15.9% in 2005-2009 to 9.2% in 2010-2014. Every step since is inside the margin. Filled dots mark the four windows that share no survey year with each other; hollow dots share four of five years with their neighbours and are not independent observations.Source: ACS 5-year, table B140022. Charter students count as public here, which is right — a charter family has not opted out of public education. B14002 has no race iteration, so this is all races.

This inverts the letter's causal claim on its own timeline. Davis says Black families are choosing private school “in greater numbers year after year.” Over exactly that span the private share of SOMA's resident students fell and then held flat. One honest limit: B14002 has no race iteration, so in principle a rising Black private share could hide inside a flat total — except that the whole sector is 774 students and the claim needs about 1,700.

What the district did about it

SOMSD is not a passive observer of its own kindergarten intake. The Intentional Integration Initiative, adopted in June 2020, assigns entering students to elementary schools using an “Adjusted Berkeley” algorithm built on socioeconomic tier — family income and adult educational attainment in the student's Census block group — rather than attendance zone, and balances each school to within ±5 percentage points of the district-wide kindergarten tier mix.12 The program's own records are the only race-by-entry-cohort data anyone publishes for this district. The district's own November 2025 integration forum presents the same history alongside its race-balance series back to 1998 and its free-lunch shares for 1998 against 2019.14

WhiteBlackMultiracialLatine/Hispanic
SOMSD kindergarten cohort placed through the Intentional Integration Initiative, by raceFour years of the district's own integration reporting. White is near 56 to 60 percent, Black 12.7 to 17.4 percent, multiracial 11.9 to 16.4 percent.0%20%40%60%share of the placed cohort2021-222022-232023-242024-25White 56.0%Black 17.4%Multiracial 13.0%Latine/Hispanic 10.3%
The placed kindergarten cohort, by race, as the district reports it. Note the grain: these are students newly placed through the III that year — the incoming cohort — not whole-school enrollment. Black share of the placed cohort runs 12.7% to 17.4% across four years, against a district that is 21.0% Black overall and elementary schools that are 17.9% Black.Source: SOMSD board presentations, Dec 2022 and Jan 202513, curated in the site's SOMSD integration source packet. Cells below 5% are suppressed by the district and are not plotted.

On its own stated metric the program worked, and quickly. Seth Boyden — the district's most economically concentrated school, 73% Low-SES in the pre-III baseline year, 43 points above the district average — was inside the ±5 band within one integrated cohort and has held there since.15

Seth BoydenClinton, Delia Bolden, Marshall, South Mountain, Tuscan±5 point target band
Low-SES kindergarten variance from the district average, by elementary schoolEach line is one elementary school's Low-SES kindergarten share minus the district average. Seth Boyden starts 43 points above and lands inside the plus or minus 5 point target band within one cohort.-30-150+15+30+45percentage points from district average2020-212021-222022-232023-242024-25±5 point target bandSeth Boyden +43+4the other five elementary schools
Every elementary school's Low-SES kindergarten variance from the district average. 2020-21 is the pre-III baseline, under attendance zones. The shaded band is the ±5-point target the assignment algorithm is written to hold. What this chart does not show is race: the III assigns on socioeconomic tier, not on race, and the two are not the same variable.Source: Alves Group III Year 1–4 reports12, transcribed into the site's integration source packet; the +43 → +4 arc is confirmed in the Sept 2025 memo15.

How much should you trust the survey half of this?

Every population number on this page is an estimate from a household survey, and at the scale of two New Jersey towns broken out by race it is a small sample. The Census Bureau publishes a margin of error with every figure precisely so that readers stop treating estimates as counts. Three things a reader should hold:

1. The margin is often the size of the finding

The estimate of Black residents aged 5-17 is 1,758 ± 406 — a coefficient of variation of 14%, at the edge of what the Bureau's own handbook examples treat as reliable.4 That single margin spans the entire question of whether there are more Black children in these towns than the district enrolls. Every difference on this page is therefore tested against the margin of the difference, not eyeballed5 — and several differences that look obvious on a chart do not survive that test, which is stated wherever it happens.

2. Race categories are not stable measurements

Two of the breaks are already described above. The next one is scheduled: OMB's March 2024 revision to Statistical Policy Directive 15 — the first since 1997 — moves federal collection to a single combined race-and-ethnicity question and adds a Middle Eastern or North African category.11 It will change both ACS and, in time, NJDOE reporting. Any series drawn straight through the next few years of these tables will be measuring the instrument as much as the population.

3. Coverage is not equal across groups

The 2020 Post-Enumeration Survey found a 3.30% net undercount of the Black or African American population, statistically significant, against a net overcount of the non-Hispanic White population.89 The ACS is a different instrument from the decennial census, but it is weighted to population controls that descend from it, and differential coverage by race is a measured property of the system rather than a suspicion. Separately, pandemic disruption forced the Bureau to release the 2020 ACS 1-year estimates only as an experimental product and to revise its weighting methodology before judging the 2016-2020 five-year file fit for release.10 Five of the sixteen windows on this page contain that survey year.

None of this makes the ACS unusable. It makes it a survey. The discipline this report applies is the one the Bureau asks for: print the margin, test differences against it, and never let a survey estimate and an administrative count share a line without saying which is which.

The verdict, as strong as the data supports

Confirmed“Black enrollment has fallen to roughly 22%”
21.0% in 2025-26, 22.5% two years ago. The district's own count.
Not supported“Down from about 54%”
The district series tops out at 51.5% in 2003-04. 54% is close to Columbia High School's share in the late 2000s, not the district's.
Wrong premise“Enrollment is down”
Total enrollment is up 12% since 1998-99. One group's enrollment fell; White, Hispanic and multiracial enrollment all rose.
Confirmed, and stronger than claimedThe fall is a loss, not a dilution
91% of the share decline is fewer Black students and 9% is the district growing. The “the district just grew” rebuttal is wrong.
Bounded, not exactHow large is the fall?
Between 35% and 56%, depending entirely on how mixed-race students who would once have been recorded as Black are treated. NJDOE published no Two-or-More category before 2006-07 and no file can locate the truth inside that range.
True of adults, false of children“The Black population has not dropped at such a rate”
10,535 Black-alone residents, 23.9% of the two towns, against a district that is 21.0% Black — the same community on the headline numbers. But only 1,758 of them are aged 5-17, the largest Black cohorts are adults aged 55-64 and 35-44, and the cohort entering school is about a third smaller than the one now in elementary school.
Refuted at the magnitude claimed“Sending their kids to private school in greater numbers”
Every private and parochial student resident in both towns, all races, K-12, is 774 ± 156. The claim needs roughly 1,700. Over the span in question the private share fell by a third and then held flat for fifteen years.
Consistent with, not demonstratedIs the district losing Black and multiracial students faster than the towns are?
Resident Black-and-multiracial students are flat across thirteen windows while the district's own count of the same category fell 20%. The current gap is 961 ± 490, which clears zero. Its growth from 451 does not clear the margin of the difference. This is the honest ceiling on the claim, and it is a real correction to the reading that the two populations simply track each other.
Found, and not in the letterThis is happening at the kindergarten door
The grade-band gradient is monotone in all 28 years and the elementary schools turned down a decade before the high school. Elementary enrollment overall is flat while its Black count fell from 1,342 to 559. Today's elementary schools at 17.9% are the district's forecast, and the trend has run continuously for 25 years.

What no dataset here will settle

Sources

Every figure above traces to one of these. The two datasets are read from pinned, immutable Aquifer releases through their own pointers, so re-running the build reproduces the page exactly; a changed number means a changed release.

  1. 1
    NJ DOE Fall Enrollment Reports, 1998-99 → 2025-26

    The annual fall headcount of every student in every New Jersey public school and district, by grade and race/ethnicity. An administrative census: exact counts, no sampling error. Read from Aquifer release nj-fall-enrollment-aq041-repair-v1 (pointer 11), district 13-4900, 28 unbroken years.

    How this dataset works →
  2. 2
    American Community Survey 5-year estimates, sixteen vintages 2005-2009 → 2020-2024

    Fourteen tables at the South Orange-Maplewood school-district geography (summary level 970) and both municipalities (level 060): B01001 and its race iterations for population by age, B14002 for the public/private split, the B14007 family for residents enrolled by grade, and B02001/B02009/B03002 for the alone and alone-or-in-combination race forms. A household survey: every figure carries a published 90% margin of error. Read from Aquifer release acs-5year-nj-series-2009-2024-v1, verified identical to the publisher's flat files everywhere the two overlap.

    How this dataset works →
  3. 3
    NCES Common Core of Data, LEA Universe Survey directory, 2014-15 → 2024-25

    Used only to join an ACS school-district geography to an NJDOE district code, through the NCES LEAID: ACS 3415330 → CCD 3415330 → NJDOE 13-4900. Parsed from an identifier, never matched on a name. Aquifer release nces-ccd-directory-2015-2025-bridge-v1.

    NCES CCD files →
  4. 4
    U.S. Census Bureau, Understanding and Using American Community Survey Data, ch. 7: Understanding Error and Determining Statistical Significance

    The source for the margin-of-error and coefficient-of-variation arithmetic used throughout: standard error is the margin divided by 1.645, and two estimates differ only if their difference exceeds the margin of that difference.

    PDF →
  5. 5
    U.S. Census Bureau, Understanding and Using American Community Survey Data, ch. 8: Calculating Measures of Error for Derived Estimates

    The published formulas for aggregating margins (in quadrature) and for the margin of a proportion whose numerator sits inside its denominator, including the ratio-formula fallback when the radicand goes negative.

    PDF →
  6. 6
    U.S. Census Bureau, ACS user note: Improvements to the Race Question (2021-03)

    The publisher's own notice that ACS race coding changed beginning with the 2020 survey year: write-in capture went from 30 to 200 characters, and up to six detailed codes are now assigned per write-in area instead of prioritising multiple responses into two. Its instruction: “any data comparisons to prior race estimates should be made with caution.”

    Census.gov →
  7. 7
    U.S. Census Bureau, “Improvements to the 2020 Census Race and Hispanic Origin Question Designs, Data Processing, and Coding Procedures”

    The same redesign on the decennial side, and the Bureau's statement that the resulting statistics “will not only reflect demographic changes, but also improvements in how we asked the questions and captured and coded the responses.” The Bureau does not quantify how much of the national Two-or-More-Races increase is procedural and how much is real.

    Census.gov →
  8. 8
    U.S. Census Bureau, “Census Bureau Releases Estimates of Undercount and Overcount in the 2020 Census” (March 2022)

    The 2020 Post-Enumeration Survey found a statistically significant 3.30% net undercount of the Black or African American population, against a net overcount of the non-Hispanic White population. Differential coverage by race is a measured property of the federal statistical system, not a hypothesis.

    Census.gov →
  9. 9
    U.S. Census Bureau, “Who Was Undercounted, Overcounted in the 2020 Census?”

    The Bureau's reader-facing account of the same coverage results.

    Census.gov →
  10. 10
    U.S. Census Bureau, “Our Commitment to Quality: A Revised American Community Survey Estimation Methodology in the Complexity of a Pandemic” (March 2022)

    Why the 2020 ACS 1-year estimates were released only as an experimental product, and what the Bureau changed in its weighting to make the 2016-2020 five-year estimates fit for release. Its instruction is to use caution when comparing 2016-2020 estimates to earlier ACS data.

    Census.gov →
  11. 11
    Office of Management and Budget, Revisions to Statistical Policy Directive No. 15 (March 2024)

    The first revision since 1997 to the federal standards for collecting race and ethnicity data: one combined race-and-ethnicity question, a new Middle Eastern or North African category, and detailed categories collected by default. It will change both ACS and NJDOE reporting; the next break in these series is already scheduled.

    Federal Register →
  12. 12
    Alves Group, Intentional Integration Initiative Year 1–4 reports (Jan 2022, Feb 2023, May 2024, May 2025)

    The district's own program evaluations: SES tiers defined from family income and adult educational attainment in the student's Census block group, the ±5-point target band, and each elementary school's variance from the district kindergarten average every year 2020-21 → 2024-25.

    District III site →
  13. 13
    SOMSD board presentations on Intentional Integration and advanced-course enrollment (15 Dec 2022; 30 Jan 2025)

    Per-school kindergarten placement by race for 2021-22 → 2024-25, as the district reported it, with cells below 5% suppressed.

    Jan 2025 deck (PDF) →
  14. 14
    SOMSD Integration Forum presentation, November 2025

    History timeline, free-lunch shares 1998 against 2019, the SES tier charts, and the race balance series 1998-2025 as the district presents it.

    PDF →
  15. 15
    Alves & McArdle, “Responses to Mary Mann's Questions” (26 Sep 2025)

    The memo confirming Seth Boyden's Low-SES kindergarten variance moved from 43 points above the district average in 2020-21 to 4 points in 2024-25, and describing the ±5 variance target with ±7.5 and ±10 sensitivity tests.

    PDF →
  16. 16
    Black Parents Workshop open letter, reported by Village Green (11 Aug 2026)

    BPW president James Davis's letter announcing the organisation will not seek an extension of its 2020 settlement with SOMSD, containing the enrollment and private-school claims this report tests.

    Village Green →

Built by notebooks/somsd-demographics/build_demographics.py and build_report.py from nj-fall-enrollment-aq041-repair-v1 and acs-5year-nj-series-2009-2024-v1. Corrections: /corrections/.