Statements about the data as received, each with the numbers it rests on, F01–F08. A finding describes the files; it does not interpret the sample. What the findings suggest doing is on the Ideas page; what they cannot settle without Irene or the beamline is on the Questions page. Every number is recomputed from originals/ on every build.
2θ (degrees), intensity, esd; no header in any of the 101 files; step 0.001° everywhere (min = max over every file); every file ends at 26.999°. Twenty files start at 0.008°, eighty-one at 0.009°, so a file of one class has one more point than a file of the other, and that is the whole difference in length. The files can be stacked into one matrix after dropping the first point of the 0.008° files, or after interpolation, with no resampling otherwise.
The 20 files with 26 992 points have a median relative error e/y above 2° of 0.0396–0.0414; the 81 with 26 991 points have 0.0463–0.0490 (one exception, F05). With Poisson statistics the ratio of counting times is about (0.047/0.040)² ≈ 1.4. The long class is spaced 620–900 s apart in time, the short class ~262 s. Every original tagged with a C is in the long class; three bare-tagged files (015, 016, 017) are long too, so the C suffix and the exposure class are related but not the same thing.
The integrated intensity over the whole 2θ range is 4.49–4.66 million counts for every sample scan, long or short, and 4.16 million for the empty capillary. Scans of two exposure classes giving the same integral means the reduction normalised them (to the monitor or to time); the files are comparable as they are, and the roughly 10 % the background lacks is the sample's own scattering plus absorption.
Peaks above 1° (local maxima more than 6 esd above a 301-point rolling median): 2, 6, 4, 13 at 900, 925, 950, 975 °C; 51 at 1000 °C; 45–63 from 1000 to 1250 °C with the strongest peak at 600–680 counts; then, through the 1270 °C hold, 54 → 138 peaks and 630 → 990 counts; 150–210 peaks and up to 1480 counts at 1275–1280 °C; 230 → 300 peaks and 1800 → 2630 counts along the 1300 °C series; 322 peaks in the closing long scan; 0 in the background. Two regimes: a first phase forming near 1000 °C and staying roughly constant, and a second growth of crystallinity (new phase, grain growth or both) starting during the 1270 °C hold and continuing with temperature. The count is crude (it includes shoulders and the tail of strong peaks) but monotonic over 60 scans, so the trend is real.
Its median relative error is 0.0661, against 0.0463–0.0490 for every other short scan and 0.0396–0.0414 for the long ones, and its peak count falls to 160 between 227 (scan 084) and 228 (scan 086), while its integral is normal (4.64 million). A shorter exposure, a partial scan or a beam dip, then normalised like the others. To be treated with care in any kinetic analysis, and asked about.
Modification time minus elapsed seconds is 2026-06-10 18:09:48 for all 101 files. Clock time of every scan is therefore known: first scan 18:09:48 on 10 June 2026, background 04:46:20 on 11 June. Whether the elapsed seconds mark the start or the end of a scan is not known from the files.
The ramp is 900 → 1225 °C in 25 °C steps; then 1240, 1245, 1250, 1266, 1270, 1272, 1275, 1280, 1290, 1299, 1300 °C: five-degree and one-degree steps as the experimenter approached and tuned the hold temperature. The values are those typed into the scan names; no thermocouple record is in the files.
Below 0.1° every file has between 43 and 91 points with exactly zero intensity and a large esd; the direct beam's tail peaks at 0.023–0.026° (1400–1700 counts) and the pattern proper begins around 0.03°. Any plot or fit should start above that, and the "maximum intensity" of a file means nothing unless the first degree is excluded (P06).
In the fifteen projection figures: the pattern is nearly flat to 950 °C; from 1000 °C broad maxima near 7.4, 8.2, 8.7, 10.2 and 14.5° rise and then change little to 1240 °C; from the 1270 °C hold, sharp reflections at 5.79, 7.89, 8.47, 9.66, 11.59, 12.59, 14.37 and 14.70° appear and grow scan by scan, through the hold and on to 1300 °C, while the broad maxima remain under them. Two stages, then: a poorly crystalline or nanometric one, and a well-crystallised one growing out of or beside it. Seen, not yet measured: the integrated intensities of named reflections against time and the identification of the phases are the quantitative form.