When to Visit a Neighborhood: 7am, 6pm, Saturday

At 6:00 p.m. on Thursday 5 November 2026, in Indianapolis, the sun has been down for twenty-two minutes. Civil twilight ends six minutes later. Run the same query for 17 June and sunset lands at 9:15 p.m., three hours and change after your evening slot opens. Those figures come from the U.S. Naval Observatory's one-day sun and moon service, queried 4 September 2026 for 39.7684, -86.1581.

So "6 p.m." is not a time of day. It is a variable that resolves differently depending on which side of the first Sunday in November you booked the flight. Under 15 U.S.C. 260a, daylight time runs from the second Sunday of March to the first Sunday of November, which in 2026 means 8 March to 1 November, and the week after that boundary your evening drive-around becomes a night drive with headlights, glare, and a completely different read on how busy a street is.

That is the small version of a larger problem. Three observation slots get onto almost every scouting itinerary — a weekday morning, a weekday evening, a Saturday — and they get treated as interchangeable blocks to be shuffled around flights and counter hours. They are not interchangeable. Each answers a question the other two cannot touch, and one of them is the only slot on the whole trip with no dataset standing behind it.

The only slot you cannot move

Start with what is fixed and what floats.

The Saturday slot is fixed by the calendar: there is one, it is the day it is, and no amount of juggling produces a second. The morning slot is fixed by arithmetic, because it has to fall on a weekday at the specific half-hour that town actually leaves the house. The evening slot floats. Which is exactly why it drifts into whatever gap the counter appointments leave behind, and why it usually gets run too early.

The order follows from what each reading feeds.

The morning run is a test of a prediction you already made. You arrive holding a depart-at estimate and a traffic count. You leave holding the delta between them, and that delta reshapes everything you ask for the rest of the trip. If the drive came in eleven minutes over the prediction, the evening slot stops being a general look around and becomes a hunt for where those eleven minutes live. If it came in on the nose, the evening is free for something else.

The evening run is a test of the direction nobody publishes. It has no prior to compare against, so it produces raw observations rather than a delta, and raw observations are worth more once you know what went wrong in the morning.

The Saturday run is a test of the town's other mode — different road volumes, different parking rules, a different set of doors open. It goes last because it is the reading most likely to charm you, and you want your questions sharpened before that happens.

Run Saturday first and you get a pleasant afternoon with nothing to compare it against. That is the whole argument for the sequence.

7 a.m. is checking a number you already wrote down

The morning slot is not exploratory. You should arrive at it holding two figures: a depart-at prediction for the actual route, and the local departure distribution.

The second one decides your alarm, and most people skip it. ACS detailed table B08302, Time of Departure to Go to Work, is the same table the commute measurement uses to see whether a town's workers crowd into one half-hour or spread across three. Here it does a narrower job: it picks the hour. Its universe, in the Census Bureau's own wording, is workers 16 years and over who did not work from home — precisely the population making the traffic you are trying to sit in. Pull it for the tract or county at data.census.gov, find the largest half-hour band between 5:00 a.m. and 8:59 a.m., and set the alarm from that rather than from habit. A town whose largest bucket is 6:30 to 6:59 will look deceptively empty to somebody who rolls out at 7:15.

What goes on the sheet during the run:

Note the third row. On 5 November 2026 in Indianapolis, sunrise is 7:18 and civil twilight begins at 6:50, so a 7:00 departure that week is made in the dark. On 17 June sunrise is 6:16, so the same departure is made in full daylight, three quarters of an hour behind the sun. Scout in June, move in November, and you measured the easy version.

The 6 p.m. reading has no table behind it

Here is the fact that should decide where this slot sits in the day. The American Community Survey asks when people leave home to go to work. It does not ask when they get home. B08302's fourteen categories are half-hour wide across the morning peak, but they open with a single overnight bucket, 12:00 a.m. to 4:59 a.m., and close with two more, 12:00 p.m. to 3:59 p.m. and 4:00 p.m. to 11:59 p.m. Those closing two are not the evening return at all — they are departures to work by people on later shifts, which is why they run four and eight hours wide while the morning bands are thirty minutes.

There is no companion table for the trip home. That absence is the reason this slot exists.

So it carries the observations nothing else will supply: the reverse-direction backup on the road you drove at 7:00, how long the unprotected left onto the main road takes when the flow is against you, whether on-street parking near the address is full by the time residents are home, and how the block sounds once televisions and dishwashers start.

The obvious slot for the sound reading is 8:30 p.m. — dinner is over and you are still awake. That reading is worth measurably less than it looks. HUD's noise regulation defines the day-night average sound level as "the 24-hour average sound level, in decibels, obtained after addition of 10 decibels to sound levels in the night from 10 p.m. to 7 a.m." — 24 CFR 51.103(a), read through the eCFR versioner service on 4 September 2026 against title 24 as issued 2 September 2026. The same section sorts sites as Acceptable at or below DNL 65 dB, Normally Unacceptable above 65 up to 75, and Unacceptable above 75.

Ten decibels is not a rounding adjustment. In the average that regulation defines, a sound after 10 p.m. counts for ten times as much as the identical sound at 8:30. If the question is whether you will sleep here, the reading has to be taken after that boundary, on the block, with the engine off. That is also the hour when the freight and flight-path questions raised by the noise lookups stop being theoretical.

Push the evening slot late. It is the cheapest hour on the trip to give up.

Saturday is a different road, and Indiana measured how different

State transportation departments already quantify the weekday-to-Saturday gap, because they have to. Portable road-tube counters run 48 hours, typically Monday through Thursday, and the raw count has to be corrected before it means anything annual.

Indiana publishes the correction table. INDOT's Latest Traffic Adjustment Factors, effective for 2024, issued June 2025, carries a sheet headed Weekday Factors by Functional Classification 2024 giving a factor for each day of the week across five road-class groups — two urban and three rural, which is how INDOT itself splits them. The factor is a multiplier: the document states that the observed average daily volume is multiplied by the factor to obtain AADT, so a factor above 1.000 means that day carries less traffic than the annual daily average.

Read across the Saturday row in the average column and the five groups run from 1.049 to 1.144. Invert those and Saturday volume sits somewhere between roughly 95% and 87% of the annual daily average, depending entirely on which class of road you are standing next to. The Sunday row is far wider, 1.089 to 1.404, or about 92% down to 71%.

The flattest group on both days is rural interstates and rural freeways — 1.049 Saturday, 1.089 Sunday — which fits roads carrying through traffic that does not care what day it is. The steepest drop is the part worth slowing down for, because it is not the urban one. On Saturday it lands on rural major collectors, minor collectors and local streets at 1.144; on Sunday it moves to rural principal and minor arterials at 1.404. Both urban groups sit between the extremes on either day. And in every one of the five groups, Friday carries more traffic than any other day: factors of 0.859 to 0.892, meaning 112% to 116% of the annual average.

Worth sitting with. A road can be nearly unchanged on Saturday or down an eighth, and the difference is a property of the road class rather than of the town. Find your state's equivalent table before drawing any conclusion from one Saturday drive.

Three other things only Saturday can settle:

• Whether parking is actually free. Enforcement schedules routinely include Saturday and exclude Sunday. Saint Paul, Minnesota publishes downtown meters enforced Monday through Saturday, 8:00 a.m. to 10:00 p.m., except City holidays, with non-downtown meters on the same days until 6 p.m. — a page last edited 3 June 2026, read 4 September 2026. A Sunday visit tells you nothing about six of the seven days you would live there.
• Which storefronts are open rather than merely licensed. Many cities publish an active business license extract with addresses. Chicago's Business Licenses — Current Active returned 54,177 records when queried through its public API on 4 September 2026, with 1,170 of them inside ZIP 60640 alone and 11,331 carrying a Retail Food license description, each row holding an address, coordinates and a business activity. The extract contains only active licences, so it says what is permitted and never what has quietly closed. Walking a block on Saturday and counting dark doors against that list is the comparison.
• What transit actually runs. In the GTFS specification, revised 27 April 2026, calendar.txt carries required saturday and sunday fields where 1 means the service is available for all Saturdays in the date range and 0 means it is not. The file itself is only conditionally required, so a feed may omit it and state every service date explicitly in calendar_dates.txt instead — check both before concluding that a feed says nothing about weekends. Agencies publish these feeds openly. Weekend service is frequently a separate service_id with different headways, and the bus you rode on Wednesday is not evidence about Saturday.

Two calendar variables that change all three slots

School in session, or not. Morning traffic, bus stops, the queue outside the elementary school and the noise of the street after 3 p.m. all switch on the district calendar. A June trip in a district that finished on 11 June is measuring a town with its schools emptied out. The district's board-adopted calendar PDF is the authority here. For why calendars within one state tend to cluster, NCES Table 1.1, Minimum number of instructional days and hours in the school year, carries a School start/finish column recording each state's statutory window: North Carolina at "Start on the Monday closest to 8/26. Finish no later than the Friday closest to 6/11," Iowa at "Start no sooner than 8/23 and no later than the first Monday in December," Arkansas keyed to the Monday of the week in which 8/19 falls. That table's vintage is 2020 and these statutes get amended, so treat it as background and the district calendar as the fact.

The month. In the 2024 row of INDOT's seasonal table, the rural interstate group runs 1.240 in January against 0.930 in July, which puts January volume near 81% of the annual daily average and July near 108%. Take two months somebody would actually fly in and the gap is still there: February is 1.101 and August 0.945, or 91% against 106%. About a sixth more traffic, on the same road, before anything else varies.

Holiday closures interact with all of this, but they are a different problem: a counter-hours problem rather than an observation problem, and one that the four-day trip design has to solve before this schedule ever gets built.

Three columns, filled in the car

The sheet that comes home is small. For each of the three slots: what you recorded, what it was supposed to match, and the size of the gap. Nothing more.

Fill it in the car, not in the hotel that night. The specific number — twenty-eight minutes, four cars deep at the left turn, eleven storefronts dark out of nineteen licensed — survives about ninety minutes in memory before it turns into an adjective, and adjectives are what you already had before you bought the ticket.

Then be honest about the size of the sample. One morning, one evening, one Saturday is three readings, in one month, in one week of one school year. What they can establish is whether the desk numbers were wrong and roughly by how much. What they cannot establish is the correct number, any more than a single walkthrough establishes that a landlord is who they say they are, which is why that verification runs on records instead of impressions. Three readings that disagree with the prediction are a finding worth acting on. Three readings that agree with it are a prediction you can now use with more confidence, and not a measurement.

Sources, read or queried 4 September 2026: 24 CFR Part 51 Subpart B through the eCFR versioner service, title 24 as issued 2 September 2026, for the DNL definition and the site acceptability table; 15 U.S.C. 260a, 2023 edition via GPO, for the daylight time period; the U.S. Naval Observatory Astronomical Applications Department one-day sun and moon service, queried for Indianapolis on 17 June and 5 November 2026; Census Bureau table B08302 category definitions pulled from the ACS 5-year API group listing; INDOT Latest Traffic Adjustment Factors, effective for 2024, published June 2025, for the seasonal and weekday tables; the City of Saint Paul metered parking page; the City of Chicago Business Licenses — Current Active dataset, queried through its public API; the GTFS Schedule Reference, revised 27 April 2026; and NCES Table 1.1.

Every figure above belongs to one state, one city or one date, and sits here to show the shape of the lookup rather than to describe your town. Indiana's factors do not apply to Ohio's roads, Saint Paul's meter hours apply nowhere else, and sunset in Indianapolis is not sunset where you are going. Run the same queries against your own coordinates, your own state DOT, your own city's open data portal and your own district's calendar, and the numbers that come back will be different. That is the point of running them.