The B1 Years
Every minute of heart rate, skin temperature, galvanic skin response, step cadence, and calorie rate captured from one wrist between Sep 9, 2013 and Jul 15, 2015. A wearable that was recalled, replaced, and eventually shelved — but not before 848,752 minutes of continuous physiology were written to disk.
This started as a portrait of one wearable. It became a portrait of one life — and one ride down the Pacific coast that the wrist almost entirely missed.
Portland to San Francisco, 1,106 miles, 40 days on the saddle
Flew Nashville → Portland on Sep 10, 2013 with a Garmin eTrex 20 running. Pedaled west from Portland to the Oregon coast at Astoria, then south down US-101 through Oregon and the California redwoods, crossing Sonoma and Marin to arrive in San Francisco on Oct 24. Forty daily GPX tracks, joined minute-by-minute to the Basis data on the same wrist.
The 4-day cluster I flagged earlier (Sept 16–19) was only the opening push. The actual tour ran through Oct 24 — forty riding days, ~44 km/day average with five days over 75 km, and 12.4 km of cumulative climbing (40,750 ft — roughly 1.4× Mt. Everest from sea level, matching the well-known elevation toll of the Pacific Coast route). The hardest single day was Sept 25 (80.5 km with 1,066 m of smoothed gain — crossing the coastal range); the fastest flat day was Oct 10 (77 km at 16 km/h avg, the Crescent City push into California). Basis recorded 484 minutes total with HR > 120 bpm while on the bike, concentrated on the steepest climb days. The Nashville point on Sept 10 is the airport ground track from the outbound flight — the eTrex kept logging while boarding. Elevation-gain figures use an 11-point moving-average smoother on the raw GPX elevation (raw sums to 124,000 ft, a known artifact of Garmin barometric noise).
· Why this happened at all
· The 40 days in photos, weather, and own-words
Each ride day as a GoPro time-lapse frame (or an Instagram post on the three Portland prep days when the GoPro was off), joined to distance from the Garmin, Basis mean HR, Dark Sky temperature and precipitation from Tractor/Foursquare check-ins, a Foursquare shout from that day, and — on six days — a longer passage from the original 2013 blog post.
· Posted as it happened
Forty-seven Instagram posts from Sep 9 to Oct 27, 2013. The curated version of the ride I broadcast at the time. Eight of them are 8-second Instagram video clips (the format was new); shown here as first-frame stills.
· Voices from the road
Seventy-nine Foursquare check-in “shouts” written in real time during the 45-day window — the handwritten narrative of the ride. The voice is unmistakably yours.
Notice how the shout frequency collapses after leaving Astoria on Sep 25 (only 20 of 79 shouts come after that date). The Foursquare notes follow the social days — rest stops in coffee shops, brewpubs, bookstores — more than the riding days. Oct 18 is the last shout (“I destroyed the salad bar :)” at Round Table Pizza in Fort Bragg); the final six riding days from Mendocino down to San Francisco went unrecorded in words, the phone presumably buried in the panniers and the attention spent on the coast.
· Did the body adapt over 40 days?
Resting heart rate measured nightly by Basis during the ride itself. A fitter body would drift this line down; fatigue drifts it up.
RHR rose 2.8 bpm across the 47 nights of data during the ride (linear slope +0.06 bpm/day). This is not fitness gain — it's cumulative fatigue: the legs got strong but the body was running a sleep deficit on a moving campsite and never fully recovered. Notice the partial dip during the Sept 20–24 Astoria layover — one of the cleanest recovery signals in the whole dataset.
· The five hardest climb days, minute by minute
For each of the top-5 smoothed-climb days, the hour-by-hour heart-rate curve plotted against that day's wall-clock (local Pacific time). The Basis HR sensor gives us a sparse but honest ground-truth for how hard each climb really was.
On paper Sept 25 is the hardest day; in the Basis HR it reads as a sustained mid-80s with a 149 peak — more a long grind than a spike. Oct 17 (79 km, 769 m) and Oct 13 (67 km, 674 m) show the opposite pattern: sawtooth HR with several excursions into the 140s, consistent with rolling terrain and sprints. The gaps in each line aren't missing exertion — they're minutes the optical HR sensor couldn't get a clean read, typical during sustained cycling.
Ride days vs home life — where the Basis agrees and where it doesn't
Compare the 40 days on the Pacific Coast against the 258 days of settled Chicago life between Dec 2013 and Aug 2014. Same device, same wrist, same minute cadence — what shifts and what doesn't when you replace the commute with a bike tour?
Mean wrist HR on ride days: 59.2 bpm. On home days: 59.6 bpm. A 0.4 bpm difference across thousands of minutes — statistically nothing. The Basis wrist-based HR sensor barely registered a two-month cycling tour. Same for calories: ride-day median kcal was only +113 above home baseline (2,229 vs 2,116). Step count actually went down on ride days (median 4,072 vs 4,444) — cycling moves the legs without swinging the arms. This is the key lesson: the Basis wrist model systematically under-reports cycling effort, which is why you need Garmin + Basis together to reconstruct the ride honestly. The wrist heard nothing; the GPS saw everything.
Resting heart rate over 310 nights — pre-recall only
Basis computes resting heart rate from the prior night's sleep — pulling the nightly trough. Smoothed over a 7-night window, and restricted to the pre-recall window (Sep 2013 → Aug 2014) where the data is dense and trustworthy.
Across the clean 11 months of pre-recall data: first 30 nights median 48.5 bpm, last 30 nights median 50.0 bpm. A mild +1.5 bpm drift — flatter than you'd expect over 11 months of moderately increasing weight. An earlier view of the same trajectory read "+5.5 bpm" because it included sparse, unreliable post-recall nights; restricting to the dense pre-recall window gives the true, quieter arc.
The circadian envelope
Every minute of every day, stacked by hour-of-local-time. The 10th–90th percentile band shows the range of what happened each hour; the center line is the median.
HR troughs at 51 bpm between 3 and 5 AM, crests at ~70 bpm around 5–7 PM. Skin temperature runs inverse: warmest overnight (vasodilation → heat release during sleep), coolest in late morning when out in the world. The bands are tighter at night than during the day, because daytime was wildly variable — while night was night.
The device came back. The nighttime wearing didn't.
Intel recalled the Basis B1 on Aug 5, 2014 over skin-burn reports. The data shows a 40.6-day gap (Aug 21 → Sep 30, 2014) — presumably the repair/replacement window. What happens next is striking: the replacement unit lived on the wrist during the day but mostly left it at night. This is the only section leaning on post-recall data; every other chart stops at Aug 21, 2014 where the sparsity becomes untrustworthy.
Before recall: 89% of nights logged sleep (mean skin-temp 92°F — clearly on a wrist). After recall: 10% of nights logged sleep (mean skin-temp at 2–5 AM collapsed to 78°F — room temperature, the device had moved to a nightstand). The sensors still worked; the behavior changed.
The body didn't care if it was Monday
Average HR across all days of the week varies by less than 1 bpm (60.9 to 61.8). Step and calorie totals also cluster tightly around the mean — no Monday shock, no weekend crash. Rhythms in this dataset follow the hour of the day, not the day of the week.
Longest continuous active blocks
A contiguous run of minutes with HR > 110 or step cadence > 50/min, allowing brief ≤ 2-min lulls. Ranked by duration. Of 1,479 blocks detected over 22 months, these ten are the single longest sustained pushes.
The top block — Sept 12, 2013 evening, 75 minutes, mean HR 119, peak 149 — falls inside the Pacific window and very likely captures a late-day training ride before the multi-day Sept 16–19 tour. A second tier at mean HR ~124 bpm with only 400–500 steps in 57 minutes (May 13, 2014 · Jul 29, 2014) shows the classic cycling fingerprint: high cardiovascular load, little arm swing.