2026: returning to the track

Updated 2 Nov 2026

Running notebook · Running record · Personal bests · Race history

2015–2019 · 2020 · 2021 · 2022 · 2023–2025 · 2026

On 19 September, three 1200 m repetitions exhausted me. On 8 October, the same workout was fourteen seconds quicker per repetition, with another repetition definitely available. Repeating the workout lets me see what changed. The notes keep the bike commute, pacing and felt effort beside the times.

On 15 October, the three 800s averaged 2:35.55, comfortably clearing the 2:40 gate. They were faster than the previous attempt and felt more composed.

On 16 October, the first three 1600s averaged 6:23.88, almost exactly 4:00/km. The early rhythm felt comfortable; direct sun and mounting strain made the third repetition survival work, before I recovered my posture for a strong finish.

On 19 October, the first three 900s averaged 2:57.00. A smooth opening rep gave way to heavy legs, a difficult headwind and a finish I could not accelerate.

On 21 October, the first mixed-rest 4+4×200 m averaged about 30.41 per 200 m. The opening repetitions felt controlled. I held almost the same speed through the second set, with slipping shoes and increasingly severe muscular strain.

On 23 October, the second three 1600s averaged 6:18.46, or 3:56.54/km. I ran progressively faster and kept my posture, still in heat and with some residual soreness from the 200s. Ankle tension returned afterwards and was easing by the discussion.

After three days without running while unwell, I returned on 27 October. The second three 900s averaged 2:58.80, or 3:18.67/km. I felt composed and finished strongly, but faster running felt unavailable from the opening rep.

On 28 October, the third three 1600s averaged 6:18.77, or 3:56.73/km. I paced conservatively after the difficult final 900s, kept good posture and composure, then trusted the legs more in the quicker final repetition. Light right inner-ankle tension during and after the run later settled; I decided to insert rest on October 29, then do 4+4×200 m on October 30 and shift the remaining sequence forward by one day.

On 30 October, I repeated the mixed-rest 4+4×200 m in the cycle’s X slot. Recorded set averages were 30.36 and 30.40, with late button presses in the first set. I reproduced the earlier speed with a shorter, 3:03.23 set break. The final repetition was all out and left me completely spent.

On 1 November, the third three 900s averaged 3:02.77, or 3:23.07/km. I felt awful after eating too much at lunch: nausea, bowel urgency and uncomfortable breathing accompanied a cold night on an unlit track. A toilet stop interrupted the session. I decided to repeat the 900s on November 2 and move LT to November 3. The repeat followed on November 2; LT remains planned.

On 2 November, the repeated 900s averaged 2:55.70, or 3:15.22/km, my quickest set so far. All three became faster. Crowds delayed the final start and required outside passes, but I felt good through the finish and afterwards. I then joined another group for three 200s with walk-back recovery, estimated around thirty seconds by feel. My ankles felt good.

This record includes completed training through 2 November 2026. The notebook explains the ambition and the choices now in force. The programme holds the progression gates and planned cycle. The sessions below preserve the completed work and the observations behind the next decisions.

At a glance · Protocol comparisons · Completed sessions · Recorded runs

Before the September programme

The return did not begin from an entirely empty year. The archive contains an April 3×1000 m in 3:56, 3:48, 3:43, and a 4×500 m in 1:35, 1:35, 1:34, 1:33. The 1000 m times are also their paces per kilometre; the 500s correspond to 3:10/km, 3:10/km, 3:08/km, 3:06/km. Their recovery was not recorded in the descriptions. These are earlier interval records, rather than results from the programme adopted in September. I also remembered self-timed 5Ks around 20:30 earlier in the year; no exact date for those efforts has been established.

Before the first track session, I also reported about 25 four-minute treadmill intervals over the preceding fifty days, at displayed speeds of 10–11.3 mph. That is a count of efforts, with no recovered breakdown into sessions or rests. It gives the track return some immediate training context; the treadmill display does not establish an equivalent outdoor pace.

September–November at a glance

LT aims at sustainable high lactate turnover; VO₂ at time near maximal oxygen uptake; F at repeatable fast running; FA adds a shorter-recovery fast set. The labels describe the intended work.

Longer-effort paces below use the written times and intended work distances; they exclude recovery. The outer-loop paces instead use an estimated route length. F, VO₂ and LT name the intended work. I measured splits and reported how the sessions felt; I did not measure oxygen uptake, lactate flux or anaerobic reserve. Recovery is between repetitions unless specified. The September sessions precede the newly agreed 18-day cycle and should not be assigned to it after the fact.

Repeating a protocol

The first LT session followed a hard hilly bike ride. The first VO₂ session had pacing help; my perceived constraint changed on the second. The later F session used more deliberate effort and movement cues; October 12’s final two were deliberately more sprinty. Each comparison keeps those conditions with the change in time.

On 9 October, the last two full 800s were quicker again: 2:39.4 and 2:39.3. Phone failures left the first finish estimated and added a partial effort before restarting the second. The dated note keeps that sequence beside the faster times.

On 15 October, all three full 800s were reliably timed and below 2:40. Even the slowest, 2:36.27, was 3.03 seconds quicker than October 9’s fastest. Today’s mean was 3.80 seconds quicker than the mean of that session’s two recorded finishes, with a more composed ending. October 9’s interruptions and estimated first finish remain part of that comparison.

The first 3×1600 m on 16 October is a new reference, rather than another identical-protocol comparison. Each repetition was a third longer than the 1200s; mean pace was about 4.8 seconds/km slower than October 8, with longer walking recoveries and substantial reported heat strain.

On 19 October, the first 3×900 m added 12.5% to the work distance of October 15’s 800s. Mean pace was about 2.2 seconds/km slower, with shorter recorded rests and a strong reported headwind. This is the opening reference for a new format, with different conditions and execution.

The 21 October mixed-rest 200s add a different fast-running reference. The opening four averaged about 30.31, against 31.25 for October 12’s first four; the second four averaged 30.50 with shorter rests after a nearly five-minute set break. October 12’s 29.30 remains the faster single repetition. Today supplied more repetitions around thirty seconds, with a much harder finish and a distance correction kept below.

The second 3×1600 m on 23 October averaged 6:18.46 (3:56.54/km), improving on October 16 by 5.43 seconds per repetition. All three were quicker, and the third improved by 11.02 seconds. Recoveries were shorter too. I kept good posture and avoided the previous breakdown, with less severe heat and some residual muscle soreness. The progressively quicker finishes were part of the pacing, not metronomic laps.

The outer Kezar loops belong outside the 1200 m comparison: distance, surface and elevation all changed. The 10/10 attempt is an experiment, not a performance score. The 2021 record offers an older example of why pace, rest and the way a session ended need to remain together.

The completed sessions

19 September — LT, 3×1200 m / 1 min

4:53 · 4:58 · 4:58

Repetition pace: 4:04.2/km · 4:08.3/km · 4:08.3/km.

Average: 4:56.33, or 4:06.94/km. Total work: 3600 m in 14:49.

I was burned out after the third: stiffening legs and heavy breathing. An important condition emerged later in the notes: the hard, hilly bike commute had already left my legs burning before I began. This is the first completed track LT reference, with that complication attached.

22 September — VO₂, 3×800 m / 5:30

2:50 · 2:49 · 2:45

Repetition pace: 3:32.5/km · 3:31.3/km · 3:26.3/km.

Average: 2:48, or 3:30/km. Total work: 2400 m in 8:24.

Breathing was hard after about 200 m and felt maximal around 300 m. During recovery it settled after roughly 2½–3 minutes; the leg burn took longer, easing around four minutes. Another runner helped pace the third repetition. I did not add a finishing sprint.

I felt that the legs were constraining the session more than the aerobic ceiling. That was my experience of the effort, rather than a measured diagnosis. The 5:30 recovery was part of giving the next repetition a chance to be useful.

24 September — F, 6×200 m / 3 min

33.81 · 33.47 · 33.99 · 33.51 · 33.56 · 33.48

Average: 33.64. Spread: 0.52 seconds.

Relaxed, repeatable fast running. These were not all-out sprints. The consistency made this a useful initial reference for the F session. It did not tell me how long I could maintain that velocity continuously.

27 September — 2000 m time trial

7:16.67 — average pace 3:38.34/km.

The recorded laps sum to the finish time. I started quickly, backed off, and did not have a reliable feel for the right pace yet. Breathing was fully heavy after the second lap; by the fourth I felt spent. I deliberately changed gears through the final 600, 400, 200 and last 100 metres.

The fast last lap records that finish. It does not by itself establish a large unused reserve. I thought better pacing might have produced something around 7:13–7:15; the result I actually ran remains 7:16.67.

29 September — LT substitute on the outer Kezar loop

The track was closed, so I ran three repetitions of two asphalt loops, with one minute between repetitions. The activity timestamp is 29 September, while the conversation reported it just after midnight on the 30th. The activity record resolves the earlier uncertainty about which day the session belonged to.

Total work time: 15:06.84.

The first loop was too quick; the later laps were steadier. The third repetition was particularly hard, and a fourth was not available.

My refined Google Earth trace gave 631.89 m per loop. If my running line matched that trace, each repetition was about 1264 m and the average work pace was about 3:59/km. The three estimated repetition paces were 3:52/km, 4:03/km, 4:03/km. The trace is an estimate of the route, not a measurement of the exact line I ran. Its 7.42 m difference between minimum and maximum elevation is a vertical range, not total ascent.

The distance, surface and elevation differ from the track 1200s. I retain this as a completed session without inserting it into an identical-workout comparison.

1 October — VO₂, 3×800 m / 5:30

2:46.0 · 2:43.5 · 2:42.5

Repetition pace: 3:27.5/km · 3:24.4/km · 3:23.1/km.

Average: 2:44, or 3:25/km. Total work: 2400 m in 8:12.

The same session on 22 September averaged 2:48. This time it felt more limited by breathing and less by the muscles. The combination of faster repetitions and a different felt constraint mattered to me more than the four-second average alone. It still did not measure how much time I spent near maximal oxygen uptake.

The first graduation gate is a 2:40 average, without a major collapse in the final repetition. This session remained above it.

2–4 October — movement between track sessions

The proposed October 3 speed session moved to October 5. There had been plenty of activity between the track visits: on October 2 I ran to the gym and back, then out for food and back. On October 3 I spent much of the day armwrestling in Vallejo and made a barefoot grocery run that left small abrasions on my feet. I felt beaten up the next day, when I also ran about a mile with a backpack. By October 5 my feet felt much better, though my upper body was still sore.

These journeys and the armwrestling explain the delay. They also belong in the training context even when there is no separate upload for each one. The track sessions are only part of the load.

5 October — F, 6×200 m / 3 min

35.4 · 31.3 · 31.9 · 31.8 · 32.3 · 31.8

Average over all six: 32.42. Average of the last five: 31.82.

The first was too relaxed. After that I put more deliberate attention into knee lift, quick leg return and posture, trying to preserve the movement through each repetition. Maintaining the movement took more effort toward the end of each repetition, while the later repetitions stayed closely grouped.

This was quicker than the September session, but the effort and mechanical intent also changed. It was not a same-effort test of improved speed, nor an all-out 200 m trial.

6 October — interrupted 10 s hard / 10 s recovery experiment

Planned: 12 repetitions. Completed: 5, a 3–5-minute interruption, then 8. That is 13 ten-second efforts, or 130 seconds of nominal hard running across two sets.

I had made an audible timer for the session. The old phone shut down after the first five repetitions, breaking the intended continuous sequence. After restarting it, I did eight more.

The effort became brutal early. I could be trying very hard while feeling that little was happening. Recovery involved several seconds of deceleration, a few walking steps and then starting again.

The rough displacements of the two sets were about 400 m and 600 m. Those include movement during recovery; they are not sprint distances. The sets also began under different conditions, so subtracting one from the other cannot reveal the distance of the final three efforts.

The session gave me a candidate to investigate, rather than a score to improve. It helped motivate the programme’s X slot: an experiment can earn a recurring place if it proves useful. This particular attempt did not establish the best format or its eventual training effect.

8 October — LT, 3×1200 m / 1 min

4:42.5 · 4:42.7 · 4:41.3

Repetition pace: 3:55.4/km · 3:55.6/km · 3:54.4/km.

Average: 4:42.17, or about 3:55.14/km. Total work: 3600 m in 14:06.5. The fastest and slowest repetitions were 1.4 seconds apart.

This was controlled. I could definitely have done a fourth repetition, and possibly a fifth. It passed the first LT gate: a 4:48 average with another repetition plausibly available. The next structure is therefore 3×1600 m; its eventual 6:08 graduation target does not have to be the pace of my first attempt at the longer workout.

Compared with 19 September, the average was 14.17 seconds faster per repetition, with much more reserve. That earlier session followed a hard bike ride, so the entire difference cannot be assigned to training adaptation.

I was still sore from the F and A combination, but 3:55/km felt relaxed. My sprinting also seemed to be returning: the stride opening, the knees going higher. At slower speeds I described needing very little conscious push, and thought a rhythm around 4:00–4:05/km could feel almost effortless. I thought the fast work had helped. These observations give me a question to pursue about movement and economy; I did not measure oxygen cost.

9 October — VO₂, 3×800 m with interrupted timing

Planned: 3×800 m / 5:30. Completed full repetitions:

~2:43 · 2:39.4 · 2:39.3

Full-repetition paces: ~3:23.8/km · 3:19.3/km · 3:19.1/km.

The first finish is a conservative estimate, and I think it may have been faster. I passed 200 m in about 41–42 seconds and 400 m in 1:21–1:22, then tried to pick up slightly toward the 2:40 target. The phone shut down just as I finished and was about to press lap. Restarting felt like two to three minutes; I then timed three more minutes of recovery, giving approximately five to six minutes in total.

The next attempt reached 200 m in about 40–41 seconds, but the phone died again at roughly 250–300 m. I was furious. After 60–90 seconds of walking back to the start, I restarted the full 800 and ran 2:39.4. I then rested 5:30 and ran 2:39.3. That is three completed 800s, plus the interrupted partial effort.

The second full repetition was unpleasant; the third was disgusting. I pushed at the end while keeping my composure, with breathing feeling maximal, and without a finishing sprint. Another 800 would have required an extreme effort. The final two nevertheless finished only 0.1 seconds apart.

I consider this a good session: all three repetitions done, quicker reported times, and a stable final repetition. I think the extra attempt made it harder. The phone failures also convinced me to replace phone timing with a watch.

Using the conservative first estimate gives an approximate set average of 2:40.6, or about 3:20.7/km; the missing finish may have been quicker. At that point, the 2:40 gate remained unconfirmed, with 3×800 m / 5:30 retained for the next VO₂ session. Next time, I want to cement it with reliable timing. The interruptions belong beside the result without taking away the work completed.

11 October — 5000 m time trial

19:59 — 3:59.8/km — 161 World Athletics points, using the reported finish.

The first four kilometres sum to 16:02, which was also my reported 4000 m split. Their times span only four seconds. *The final kilometre is the estimate implied by the accepted 19:59 finish.

This was one of the most unpleasant efforts of my life. My legs felt heavy, breathing felt completely spent, and after about 2500–3000 m I could no longer keep my posture. The knee lift was gone. During the last five to ten minutes, it felt as if I was crawling through the distance rather than running. The pace stayed broadly even while the movement became much harder to maintain.

The phone shut down in the last 50 m, leaving the finish unrecorded. I had run a 96-second lap with three laps to go, then gave everything in the last two, with an attempted sprint over the final 100–150 m. I estimated those last two laps at about 1:35 and 1:33, with roughly four seconds of uncertainty. I put the finish at a conservative 19:59; I thought even a pessimistic estimate would be around 20:02. The reported result here is 19:59, with that timing history beside it.

When discussing the 16 October workout, I clarified that I also brought my posture and knee lift back together over the final two laps of this trial, especially the last. The miserable middle and the strong finish both belong in the account.

In the men’s outdoor 2025 World Athletics table, 19:59 gives 161 points; 20:00 gives 160. The small difference at twenty minutes changes little about what this effort tells me.

October 8’s controlled LT repetitions averaged 3:55.1/km, with one-minute rests and 3600 m of work. Today I sustained 3:59.8/km over 5000 m continuously, at a very different felt cost. The LT workout had earned 3×1600 m. This trial adds a continuous benchmark and a question about keeping the movement available as the effort goes on; it does not identify which physiological change would answer that question.

12 October — F, 6×200 m / 3 min

31.73 · 31.21 · 31.43 · 30.64 · 30.22 · 29.30

Average over all six: 30.76. First four: 31.25 average. Last two: 29.76 average. Recovery remained three minutes between repetitions.

I pushed the last two to be more sprinty. It felt good. The final 29.30 brings one repetition below thirty seconds in this current 6×200 series.

October 5 averaged 32.42, with an overly relaxed opener and a 31.82 average over the last five. This set’s overall mean was 1.66 seconds quicker, while the effort also changed. The first four and the faster final two keep that distinction visible. The benchmark stays at six 200s with three-minute rests; there is no new F graduation gate.

15 October — VO₂, 3×800 m / 5:30

2:35.26 · 2:35.11 · 2:36.27

Repetition pace: 3:14.1/km · 3:13.9/km · 3:15.3/km.

Average: 2:35.55, or 3:14.43/km. Total work: 2400 m in 7:46.64. The fastest and slowest repetitions were 1.16 seconds apart. COROS lap timing confirms all three finishes; the recorded recoveries were 5:25.67 and 5:20.82, against the intended 5:30.

This was still extremely hard, but easier than October 9. It was hard rather than disgusting. In the last 200 m, breathing felt maximal while I stayed composed and felt that a sprint was available. I deliberately held the pace and finished strongly without using that final acceleration. The finish felt natural, deliberate, and much like home.

The early 200 m checks were around 38–39 seconds. I contained the first 50 m, found the rhythm through the opening curve, then used the splits to check it. The second lap took increasing attention to stride, knee lift, chest position and arm movement. The intention was to keep the speed through the finish. I thought I could have restarted after five minutes, although the recoveries I actually took are the ones recorded above.

There was one brief disorganised patch on the first-lap backstraight of the third repetition: the legs felt absent too early. I found the rhythm again by the end of that straight. Later, a damp-looking inside section made me hesitate between my usual close line on the curve and the visibly dry middle of the lane. The second repetition also had an unusual distraction: my left shoe was fully untied from about 300 m onward.

I reported hotter conditions than last time. I had skipped the preceding track session; the feverish feeling had cleared before bed the previous night. The left Achilles was quiet during the repetitions and afterward, with only a subtle left–right difference if I concentrated on it.

This clears the 2:40 average gate without a major final-repetition collapse and earns 3×900 m. Its 2:55.5 target is the next graduation gate; 5:30 remains the nominal recovery. The already-earned LT move to 3×1600 m is independent of this decision.

16 October — LT, 3×1600 m / 100 m walks

The first longer LT session, on day two of the 18-day cycle that began on 15 October. COROS recorded the three work repetitions as follows; paces use the intended 1600 m track distance.

Mean: 6:23.88, or 3:59.93/km. Total work: 4800 m in 19:11.65. The two approximately 100 m walking recoveries took 1:11.93 and 1:15.99. Elapsed time through the third finish was 21:39.57, over approximately 5 km of work and walking. A further 58.34-second walk followed in the recording. The work pace almost exactly matches the reported 19:59 continuous 5000 m; today’s elapsed route time includes the recoveries and is not another time trial.

The pace felt comfortable in the early two or three laps. I was thinking that I would probably speed up in the later repetitions. Instead, I got cooked under the sun. It felt at least as hot as the previous day’s 800s, but those had two laps of running followed by five minutes in shade. Today there were four laps, then a short walk still in direct sun. I drank water in the breaks, without getting relief from the heat.

The last lap of repetition two was already very hard. I felt burned during the walk and overheated within the opening 200 m of repetition three. Its second and third laps brought the same loss of posture and knee lift I had felt during the 5000 m trial. This repetition was survival work.

My approximate checkpoints were 1:37 at 400 m, 3:14 at 800 m and 4:52 at 1200 m: roughly 97, 97 and 98 seconds per lap. I locked in for the final lap, went into the red and finished in 6:27. That implies approximately 95 seconds for the last lap, using the rounded checkpoints. I recovered good posture and knee lift as well as speed. This was a harder finish than I had intended before starting. I felt normal again within 10–15 minutes.

The difficult laps reminded me specifically of laps 7–10 of the 5000 m. I had also recovered my form for that trial’s finish, although it felt much more miserable. Today made the question of preserving good running through the difficult middle tangible again. There is a time to push and a time to back off; learning to stay composed through hard running is part of what I want to develop. The first 1600 session is complete. 6:08 average remains a later graduation target, and no cooler-weather time is inferred.

19 October — VO₂, 3×900 m / nominal 5:30

The first session at the longer distance. COROS recorded:

Mean: 2:57.00, or 3:16.67/km. Total work: 2700 m in 8:51.01. Actual recoveries were 5:03.34 and 5:14.31, shorter than the nominal 5:30. The set spans 7.31 seconds. Compared with October 15, I ran 12.5% more work distance at a mean pace about 2.2 seconds/km slower. The strong headwind, shorter rests and progressive slowing belong beside that comparison.

The inner ankle caused no problems before, during or after the session. The first rep felt smooth, easy and controlled. Breathing was barely noticeable through 400 m, elevated but manageable at 800 m, and I finished strongly. The legs felt heavy immediately afterward. At three minutes my breathing was under control, but the heaviness remained when I started again around five.

By the first lap of rep two I was breathing heavily. A strong headwind on the backstretch made me push hard with my legs to hold the pace. Later in the rep I felt at maximum capacity and began losing posture, pushing through the last 100 m on the curve to finish. The legs were cooked. Again, breathing settled after roughly three minutes, while my legs still felt heavy at the next start, around 5:15.

In rep three I was already breathing heavily on the first backstretch, and breathing felt maximal from 400 m onward. My approximate checkpoints were 39 seconds at 200 m, 1:18 at 400 m and 2:40 at 800 m. In the second lap’s headwind I pushed hard, but posture and coordination kept deteriorating. I concentrated on finishing strongly through the last 300 m and tried to change gear after 800 m, aiming to get under three minutes. This time the extra gear was not available. The rounded checkpoints imply roughly 78 seconds for the first 400, 82 for the next 400 and 21 for the final 100.

It felt more like muscular failure than anything else: heavy legs, lost composure and less effective running. That describes my experience, not a measured mechanism. I had approached the 900s much as I approached the 800s, expecting to work out the extra 100 m at the end. The longer effort may have had its say. I am happy with this first session; next time I want a lower average and more even pacing. 3×900 m stays the current format, with 2:55.5 average still the later gate.

The planned 4×200 m / 3 min + 4×200 m / 90 seconds has a specific connection for me. The later 200s resemble those final metres: out of breath, legs heavy, yet needing to gather myself, recover posture, produce fast running and finish strongly. I want to see whether practising that response in the 200s helps me access it late in longer efforts. That is a training hypothesis to follow.

21 October — FA, 4×200 m / 3 min + 4×200 m / 90 seconds

San Francisco, 5:28 p.m. The track was wet and I could feel my shoes slipping. The session was intended for October 20, but an event blocked access when I arrived and I could not return later. I completed it today instead.

The third repetition needs a distance correction. COROS recorded 31.02. I ran that repetition in lane three, from the 200 m mark to the opposite-side 200 m mark, covering an extra half-lap arc relative to lane one. Using my reported 42-inch (1.0668 m) lane width, the estimated extra distance is π × 2 × 1.0668 = 6.702902… m. At the same average speed, 31.02 over 206.702902… m corresponds to 30.014092… over 200 m, displayed as ~30.01. This is a geometry-based comparison; the original recorded time stays 31.02. Only this repetition needed the correction; the other seven covered 200 m.

The mean of the eight 200 m comparisons is ~30.41. The second set was about 0.18 seconds per repetition slower than the first, and the full range was 0.91 seconds. I opened in 30.47 and finished in 30.46.

Recorded recoveries:

• First set: 2:50.86 · 3:00.52 · 3:00.85.
• Between sets: 4:56.05.
• Second set: 1:32.29 · 1:55.52 · 1:25.59.

The 1:55.52 segment before repetition seven includes an interrupted start. I began at exactly 90 seconds, accidentally double-clicked the lap button, noticed the problem about 10 m in, stopped, returned to the start and went again. The remaining 25.52 seconds contains that effort and return. It was an extra short acceleration as well as a longer gap before the full rep.

The first repetitions were nice and controlled. By the fourth, my legs felt stiff. The second set became brutal. In repetition seven’s last 50 m, my legs felt completely stiff, with heavy, painful squeezing contractions. In the last repetition, that feeling arrived before I had even left the bend. I found a way to hold the speed by trying to put more force into the ground and using more arm action, while feeling my legs give way. I finished absolutely spent and out of breath; the burning in my legs lasted about ten minutes.

I think this was a good session, and holding the speed was impressive. The consistent times also conceal how much the effort and movement changed. Compared with October 12’s controlled opening four, today’s first set was about 0.94 seconds quicker per repetition; the later four stayed close to that speed with shorter rests. The effort, five-minute set break, wet surface and interrupted start all belong beside the comparison.

The 19 January 2022 version averaged 30.08 then 32.93. Today’s opening speed was close, with much less slowing in the second set. The old set-break duration and effort are not established, and that winter reference is distinct from June 2022’s roughly 27–28-second version.

Afterwards, I felt tension at the back of my right ankle, similar in location to the earlier left-sided tension. There was no ankle pain, but I changed my walking step as a precaution for a couple of hours. It felt normal again by the discussion.

I will rest on October 22. LT followed by VO₂ is the next intended pair, subject to recovery. I want another 4+4×200 in X, with under thirty seconds on average for its opening four as my prediction. Today’s first four totalled about 2:01.254 in 200 m equivalents; the ambition is to go under 2:00. That is a future target, not a new completed result or formal gate.

This puts the proposed connection to the 900s into sharper focus. I could keep producing fast running while very fatigued; I also described a severe late breakdown. Whether repeating this helps me preserve movement and find a finish in longer efforts remains a question to follow.

23 October — LT, 3×1600 m / nominal 1 min

San Francisco, 12:43 p.m. COROS recorded:

Mean: 6:18.46, or 3:56.54/km. Total work: 4800 m in 18:55.37. Actual recoveries were 1:02.68 and 1:11.25, against the nominal minute. From the first repetition’s start to the last finish, including both recoveries, the sequence took 21:09.30. It includes breaks rather than continuous running.

Compared with October 16, the average was 5.43 seconds quicker per repetition, or 3.39 seconds/km. The third repetition was 11.02 seconds quicker. The whole sequence improved by 30.27 seconds: 16.28 seconds of faster work and 13.99 seconds less recovery. Conditions and the way the effort developed matter alongside that change in time.

Composure and posture were good. I felt comfortable through the laps. It was still hot, though less hot than last time, and I did not burn out. I felt on the edge of overheating during the fourth lap of repetition two; by the end of the session I felt all good.

These were not metronomic laps. I remember many in the 1:35–1:37 range, with faster closing laps around 1:33–1:34. In the last repetition I remember passing 1200 m in about 4:45. Combined with the recorded 6:16.05 finish, that implies a final lap of roughly 1:31, potentially the fastest of the session. The intermediate split is recalled, so that closing lap is an estimate, not a separately recorded 400 m time.

The third rep therefore appears quicker through both the opening 1200 m and the finish than last time’s recalled 4:52 plus roughly 1:35. Holding my posture and feeling comfortable through more of the work is an important change from the previous heat-limited session. Cooler conditions helped; they do not let me isolate how much of the improvement came from adaptation.

I was still a little sore from Wednesday’s 4+4×200 m, mainly in the glutes and slightly in the thighs. Afterwards, both ankles felt under tension. It did not change my walking this time, and the tension was already easing by the discussion. The soreness and ankle response belong in the session’s recovery account as well as the faster times.

3×1600 m remains the current LT format. The 6:08 average (3:50/km) graduation gate is about 10.46 seconds per repetition ahead of today’s mean. At the time, two LT slots remained before the cycle’s time trial, and the October 24 900s were provisional. They were subsequently postponed; the completed return is recorded below.

27 October — VO₂, 3×900 m / nominal 5:30

Stanford track. COROS recorded the three efforts and intervening recoveries:

Mean: 2:58.80, or 3:18.67/km. Total work: 2700 m in 8:56.40. Actual recoveries were 5:00.20 and 5:04.03. The first start to the last finish took 19:00.63, including those recoveries. Paces use the intended 900 m distance and recorded lap times, keeping watch-estimated distance separate. The programme’s nominal recovery remains 5:30.

I did not run on October 24, 25 or 26 while unwell. Before this return, I felt well again and reported that my ankles felt fine.

The distinctive feeling was present from the beginning. I felt good overall, composed, and able to finish strongly. Yet my legs did not feel fast. I tried to put more into the running without finding the response: a numb, slow feeling, as if this was the pace I could sustain and a faster gear was simply unavailable. I did not feel I was deliberately holding back or that I could have chosen to run 2:50.

That differs from October 19, when opening in 2:53.74 felt like flying, with the possibility of going quicker, before heavy legs and deteriorating form took over. Today was 1.80 seconds slower per repetition on average, or 2.00 seconds/km, with slightly shorter recoveries. The respective differences were 3.51, 1.36 and 0.52 seconds: most of the lost time was in the first two reps. The third still slowed, but I did not experience the previous breakdown and finished strongly. The sessions were on different tracks, and matching weather conditions have not been established.

I see this as a return after three days off and a short illness. I expect the next efforts to feel more like myself, possibly quicker again. That is an expectation to test; neither the times nor the sensation identify a physiological cause. 3×900 m remains the current format, with the 2:55.5 average gate still ahead.

28 October — LT, 3×1600 m / nominal 1 min

San Francisco. COROS recorded:

Mean: 6:18.77, or 3:56.73/km. Total work: 4800 m in 18:56.30. Actual recoveries were 57.42 seconds and 1:06.01. From the first start to the last finish, including both breaks, the sequence took 20:59.73. Paces use the intended track distance and recorded times.

I played this conservatively. The difficult final repetitions in the 900s had left me unsure whether my legs would hold, so I wanted to feel good in each rep. I generally kept my posture and good composure. On the last rep I was less cautious and ran quicker. Perhaps that was the pace I could have used earlier, but I had been reluctant to commit to it.

The average almost matched October 23: 0.31 seconds slower per rep, with a final rep 0.19 seconds quicker. The whole sequence was 9.57 seconds shorter, because the breaks were 10.50 seconds shorter while the running took 0.93 seconds longer. Today adds a controlled repeat after the illness interruption and yesterday’s 900s. My sense that more pace was available remains an observation to test; I did not run three repetitions at the final rep’s pace.

3×1600 m remains the LT format. The 6:08 average (3:50/km) gate is still 10.77 seconds per repetition ahead of today’s mean.

I lightly felt tension on the inner side of my right lower leg, just above the inner ankle bump, both while running and after finishing. It ran along a length of soft tissue rather than one spot on the bone itself, and had gone by the later discussion. I do not know which structure was involved. I decided to rest on October 29 before doing 4+4×200 m on October 30. The remaining sequence moves forward by one day, preserving the workout order and the rest days that follow X and the next VO₂/LT pair.

30 October — X, 4×200 m / 3 min + 4×200 m / 90 seconds

San Francisco, 1:59 p.m. I repeated the FA format in the cycle’s experimental X slot. COROS confirms these recorded times:

The first and third finishes include button-press delays. On the first I needed a second press; I estimate at least 0.15 seconds, more likely 0.20. On the starred third rep, I needed three attempts, including a pause when I realised the second had not registered. I estimate at least 0.50 seconds, more likely 0.70. These are recalled estimates; the recorded times remain above.

Subtracting the more likely 0.20 and 0.70 seconds gives illustrative times of ~30.35, 30.28, ~30.09, 29.81: ~30.13 average for the first set and ~30.27 over all eight. The unadjusted eight-rep average is 30.38. I thought the opening average could have been around thirty; the original laps and estimated delays stay separate.

Recorded recoveries were 2:56.16, 2:57.93 and 3:00.39 in the first set, 3:03.23 between sets, then 1:29.33, 1:29.74 and 1:30.71 in the second set. Late finish presses also affect the adjacent recorded recovery boundaries. The set break was 1:52.82 shorter than October 21’s 4:56.05; today’s second-set recoveries closely followed ninety seconds throughout.

The finish was very hard. In repetition seven, I tried to stay relaxed on the straight rather than muscle through the last eighty metres. In the final rep I gave everything: it felt like all muscles, and I finished completely destroyed. The 31.13 to 30.11 change includes that deliberate increase in effort: the quicker finish took everything I had.

I reproduced roughly the earlier session’s speed with substantially less between-set recovery. The estimated timing corrections suggest a modest improvement, while the all-out finish keeps its cost visible. This completes X and the eighth of twelve workouts in the cycle.

1 November — VO₂, 3×900 m / nominal 5:30

Stanford track, at night. COROS recorded:

Mean: 3:02.77, or 3:23.07/km. Total work: 2700 m in 9:08.30. Recorded recoveries were 5:11.14 and 10:48.17, against nominal 5:30. Times and paces use the intended 900 m track distance, separately from the watch’s distance estimate.

I felt disgusting. I had eaten far too much at lunch and felt that the food had not settled. I felt close to throwing up, urgently needed the toilet, and stopped around the second repetition to use it. That interruption belongs beside the unusually long recorded recovery. It was cold, the stadium had no lights on, and I felt out of breath and uncomfortable while running.

The mean was 3.97 seconds per rep slower than October 27 and 5.76 seconds slower than October 19. Unlike October 27’s composed running with a missing faster gear, this was an effort made while feeling physically awful. The times remain part of the record, but the session gives a poor comparison of fitness after the recent illness. No adjusted performance is inferred from the stomach symptoms, cold or darkness.

On November 2, I decided to repeat 3×900 m that day, then do LT on November 3. November 1 still counts as completed training; the repeat adds a VO₂ session to the original twelve-workout cycle. The repeat was subsequently completed, as recorded below; LT remains planned. The numerical 2:55.5 average VO₂ gate is unchanged.

2 November — extra VO₂, 3×900 m / nominal 5:30

San Francisco. I repeated the 900s after the previous night’s session, then joined another group for three fast 200s. COROS recorded the 900s:

Mean: 2:55.70, or 3:15.22/km. Total work in the 900s: 2700 m in 8:47.09. Recorded recoveries were 4:59.86 and 6:03.16; the programme’s nominal recovery remains 5:30.

There were massive crowds before the last repetition. I waited roughly an extra minute hoping they would clear. They did not, so I started. I made many passes, mostly on the outside, including on the bends with excursions as far as lane three. The 2:54.84 remains the recorded finish; the pace above uses nominal 900 m, without estimating the extra distance from those passes.

I felt good through the last rep and even afterwards. The mean was 1.31 seconds per rep quicker than October 19, 3.10 quicker than October 27 and 7.07 quicker than November 1. The fastest and slowest repetitions were only 1.98 seconds apart, and the fastest came last. That final rep was 6.21 seconds quicker than October 19’s finish. The sequence, traffic, unequal rests and felt effort belong alongside the quicker average.

Afterwards, I joined some fast runners doing 200 m repetitions with walk-back recovery, for three repetitions. They timed their efforts, but I did not ask for the times. My estimate of around 30 seconds each is based on feel, not a supplied stopwatch reading. Those add approximately 600 m of fast running, for 3300 m of nominal work in the session. My ankles felt good.

Status clarification, 2 November: close to the target, not graduated. The recorded average is 0.20 seconds per rep above the 2:55.5 gate (0.59 seconds across the set, calculated before rounding). It is a good result, and I want to cement the graduation target next cycle. 3×900 m remains the current format; moving to 4×900 m still requires earning the gate. LT on November 3 remains the next plan.

What the next sessions will test

November 2’s 3×900 m is the strongest set so far: a 2:55.70 average (3:15.22/km), progressively quicker reps, and a good finish despite crowding. It adds a much more encouraging reference than the stomach-disrupted effort the night before. It is close to the 2:55.5 target, not a graduation. I aim to cement that target next cycle, with no traffic-adjusted time inferred.

Ten workouts are complete in the cycle begun October 15: nine of the original twelve, plus the November 2 repeat. The intended total is now thirteen. The three extra 200s belong to that same session, not a separate workout count. LT on November 3 remains planned at 3×1600 m. I want to move the last 6:18.77 average (3:56.73/km) towards the low 6:10s. The intention is a more confident opener under 6:15, without going crazy, then trying to accelerate across the set. Around 1:33 per lap gives 4:39 through 1200 m, just under the 4:40 checkpoint, and 6:12 for 1600 m (3:52.5/km). These are targets; the 6:08 average graduation gate remains ahead.

The earlier November 4 800 m TT, November 5 6×200 m and November 6 closing chill day remain provisional. I am treating the 800 as a checkpoint within normal training, without a formal taper. The programme keeps the current calendar separate from the original template. A possible next-cycle move to five LT and three VO₂ sessions remains unadopted.