REACH OF A READOUT -- 10 chip-days of layout A, 2 of layout B, 4 four-cluster chips, 3 floor chips 1. Fall-off with distance. Ratio of reported/stated (measured arm) to reported/laid (idle arm), by spacing band. data: ibm_pittsburgh, 7 Oct 2026, runs [1031, 1032, 1033]; 73 spectators 3 or more spacings from every measured qubit. Three arms in one job per read: idle (identical idle time, nothing measured), measure (the chip's own 2.18 us readout), measure_2 (a 1.38 us readout). ratio = (reported/stated on the measured arm) / (reported/laid on the idle arm); reads 1 and 2 of the job in the first table. measure / idle, reads 1+2 value 3-5 spacings (n=14) 1.031 6-8 spacings (n=18) 1.025 9+ spacings (n=41) 1.009 idle control, all 73 0.991 <- idle arm, absolute (reported/laid) The same quantity as more of the data is admitted (measure / idle): 3-5 6-8 9+ Pittsburgh, read 1 only 1.062 1.021 1.024 Pittsburgh, reads 1+2 1.031 1.025 1.009 Pittsburgh, all 3 reads 1.014 1.027 0.999 1.034 0.985 1.000 (measure_2 arm, all 3 reads) read-to-read, measure/idle: read 1 (run 1031) 1.062 1.021 1.024 read 2 (run 1032) 1.001 1.028 0.995 read 3 (run 1033) 0.981 1.033 0.980 ibm_fez, all reads 1.017 1.017 0.987 (runs [1081, 1082, 1083]; read 1 alone: 1.038 1.023 0.991) ibm_aachen, all reads 0.998 1.011 1.001 (runs [1091, 1092]; read 1 alone: 0.999 1.048 1.009) -> Over all three reads of the Pittsburgh job the near-band excess is +0.014 and the 6-8 band is larger than the 3-5 band; the read-to-read spread of the 3-5 band is 0.035, as large as the effect. The two other chips run the same design show no near-band excess. NOTE the idle-control value is an absolute reported/laid value (idle arm), while the band ratios already divide by it; the two are not on one scale. 2. Three predictors ranked against each spectator's reading (Spearman rank correlation with reported/stated). Predictors: intensity (sum of 1/r^2 over all measured qubits), count of measured qubits within two spacings, nearest measured-qubit distance. n | count intens. nearest layout A pool (5 chips, 8 chip-days, no Phoenix) 1283 | +0.120 +0.157 -0.085 best by |rho|: intensity. bootstrap 95% interval (resampling spectator-reads, 1000 draws) of rho_intensity - rho_count: [+0.004, +0.070]; of rho_intensity - |rho_nearest|: [+0.046, +0.100] four-cluster chip (Fez), read 1 only 52 | +0.486 +0.501 -0.496 best by |rho|: intensity. bootstrap 95% interval (resampling spectator-reads, 1000 draws) of rho_intensity - rho_count: [-0.145, +0.187]; of rho_intensity - |rho_nearest|: [-0.093, +0.101] four-cluster chip (Fez), all 3 reads 155 | +0.327 +0.348 -0.311 best by |rho|: intensity. bootstrap 95% interval (resampling spectator-reads, 1000 draws) of rho_intensity - rho_count: [-0.076, +0.109]; of rho_intensity - |rho_nearest|: [-0.022, +0.096] -> intensity has the largest |rho| in each set. In the pool all three correlations are weak (|rho| <= 0.16, i.e. under 3% of rank variance) and the intervals above show whether the margin over the other two predictors is distinguishable from zero. 3. Dose-response: spectators binned by number of measured qubits within two spacings (0, 1, 2, >=3). Layout A pools 6 processors (10 chip-days, 24 reads): the mean over processors of each processor's mean of per-read medians. 0 1 2 >=3 layout A (mean over 6 processors) 1.039 1.045 1.083 1.181 boston 1.070 1.078 1.113 1.216 (10-04 x3) fez 1.019 1.032 1.040 1.349 (10-04 x3, 10-06 x1) kingston 1.055 1.111 1.122 1.127 (10-04 x3, 10-06 x1) marrakesh 1.074 0.976 1.096 1.197 (10-04 x3) phoenix 1.005 1.023 1.051 1.100 (10-05 x3, 10-06 x3) pittsburgh 1.010 1.051 1.077 1.098 (10-04 x3, 10-05 x1) processors where >=3 exceeds 0: 6 of 6. Monotone 0<1<2<3+ on: boston, fez, kingston, phoenix, pittsburgh same, rate cut applied 1.026 1.032 1.049 1.066 (only spectators with |rate| > 4 deg/us, the cut the duration table uses; spectator-reads kept 244/380/315/155 of 430/633/520/218) Layout B (an independent second random draw of which qubits are measured), 7 Oct: ibm_pittsburgh layout B runs [901, 902, 903]: 1.032 1.083 1.131 1.176 0 measured in range: 1.032 >=3 measured in range: 1.176 ibm_pittsburgh is also one of the six processors pooled in layout A (layout A there, 4 Oct and 5 Oct: 1.011 1.047 1.056 1.040; 1.010 1.055 1.098 1.156). On that same processor, layout A on 4 Oct gives >=3 -> 1.040 (0 -> 1.011); layout B three days later gives 1.032 -> 1.176. ibm_marrakesh layout B runs [911, 912, 913]: 0.979 1.114 1.196 0.688 (layout A, same processor, 4 Oct: 1.074 0.976 1.096 1.197) -> layout B on ibm_marrakesh gives >=3 -> 0.688, below its zero bin 0.979. Marrakesh layout B has 7 spectators with >=3 measured qubits in range; read-to-read: 0.696 / 0.685 / 0.684 Change in reading per additional measured qubit in range: table, (>=3 bin - 0 bin) / 3 rungs (layout A) +0.047 table, layout B Pittsburgh, (>=3 - 0) / 3 +0.048 least-squares slope, every spectator-read in layout A, 0..4 in range +0.042 (n=1283) least-squares slope, every spectator-read in layout B Pittsburgh +0.083 (n=203) pittsburgh: 44 spectators present in both layouts: least-squares change in reading per extra measured qubit in range +0.044; median change of the 12 that gained exactly one: +0.214 marrakesh: 39 spectators present in both layouts: least-squares change in reading per extra measured qubit in range +0.218; median change of the 10 that gained exactly one: +0.027 Pittsburgh median change for spectators that gained exactly one measured qubit: +0.214 (12 spectators); this is a subgroup median, not a slope. 4. Four clusters of measured qubits around four central spectators. data: ibm_fez, 7 Oct 2026, runs [921, 922, 923] (same layout, three consecutive jobs); stated readout 1.660 us; centres (spectator qubit numbers) [43, 131, 102, 69]; 25 measured qubits. reported/stated at each centre (nan = the spectator's phase signal was too weak to fit): centre read 1 (921) read 2 (922) read 3 (923) 43 2.066 nan 0.513 131 nan nan 1.429 102 2.285 nan -0.424 69 nan 3.252 1.463 readable: 7 of 12 (centre, read) cells; values -0.424, 0.513, 1.429, 1.463, 2.066, 2.285, 3.252; median of readable 1.463 The 4th centre (131) was unreadable on reads 1-2 and read 1.429 on read 3. read 3 of the same job: 43: 0.513, 131: 1.429, 102: -0.424, 69: 1.463. Per-read mean of the readable centres: 2.175 / 3.252 / 0.745 2 of the readable centre-values are below 1 or negative: 0.513, -0.424 (a negative elapsed time is not a duration; this is the estimator failing near its limit). Open field: the same job's spectators with no measured qubit within two spacings (median of reported/stated): read 1 (run 921): fit estimator 1.003 (n=33) | 3-and-9 estimator 1.044 read 2 (run 922): fit estimator 0.978 (n=29) | 3-and-9 estimator 0.973 read 3 (run 923): fit estimator 0.983 (n=31) | 3-and-9 estimator 1.002 three reads pooled (fit estimator): 0.983 The same four-cluster layout, same day, three other processors (reported/stated at the centres, reads 1/2/3): ibm_pittsburgh runs [931, 932, 933] 43: 1.51/nan/nan | 131: -0.37/-0.37/-0.41 | 102: 0.21/0.23/0.18 | 69: 0.78/0.79/0.68 ibm_marrakesh runs [951, 952, 953] 43: 0.04/0.04/0.06 | 131: nan/nan/nan | 102: 0.72/2.26/2.50 | 69: nan/nan/nan ibm_aachen runs [961, 962, 963] 43: 1.38/1.39/1.52 | 131: nan/nan/nan | 102: nan/nan/nan | 69: 2.16/2.75/1.76