Uncited Press › Star Trek Open the interactive journal →
Trek · Physics & Instrumentation

Time to Close and Time to Escape: Strand Geometry and Subspace Variance in 31 Recorded Tholian Webs, 2268–2375

Dr. Sharel Vo'thass1
1 Daystrom Institute, Subspace Theory Division
Received 1 Sep 2026 · Revised 9 Oct 2026 · Accepted 24 Oct 2026

Abstract

Tholian webs have trapped Federation and allied vessels since 2268, and a trapped crew needs to know how long a passable gap will remain. We tested whether that interval is set by the progress of the lattice, measured as closure time, or by local subspace instability. Sensor logs of 31 recorded constructions held by the Starfleet Fleet Telemetry Archive were used. For each web the "radius", the number of "ships", the "subspace variance" and two times were extracted, the time to closure and the time at which the last passable gap was lost. Closure time followed the strand geometry, with an exponent of 1.94 on radius (95% CI 1.85 to 2.03) and an exponent of −0.94 on ship count (95% CI −1.07 to −0.81). A web of 50 thousand km radius needed about six ships to close in two hours. The last passable gap was lost by subspace reclosure, not by a laid strand, in 25 of the 31 webs (80.6%). In those 25 webs the escape window fell almost in inverse proportion to variance (slope −1.03 on the log scale, 95% CI −1.19 to −0.87, R² = 0.89) and was unrelated to closure time (slope 0.15, 95% CI −0.35 to 0.64). The typical cross-validated error in window length was 26% from variance alone, against 98% from closure time. The escape window can be calculated by a trapped crew within ten minutes of the first strand.

1. Introduction

In the 31 Tholian webs analysed here, the median time from the first strand to the sealing of the last gap was 301 minutes. The median time for which a gap remained wide enough for a trapped ship to pass was 46 minutes. In the longest web the lattice needed 704 minutes to close, and in the shortest-lived the passable gap was lost after 16.

The difference between these two times is the problem which this paper addresses. Tholian webs are laid by Tholian ships which hold station on a closed surface around a target and extend energy strands between them until no aperture remains. The first web to be recorded by a Federation starship, around USS Enterprise in 2268, was escaped only through the spatial interphase which had already taken USS Defiant, at a moment calculated for that interphase and not for the web, and no estimate was made of how long the gaps would last. In fact, the archive holds 38 recorded constructions since then, and in none of them did the trapped vessel have a figure for its remaining time before the lattice was complete. A crew which assumes that the web will be finished at the time predicted from the strand-laying rate will act too late in most unstable regions of space, because by the time the lattice looks nearly complete from the outside, the gaps through which a ship could pass have already been closed from within by the drift of the strands themselves.

Two quantities can be separated. The time to closure depends on how many strands are needed and how fast they are laid. The escape window depends on how long a gap stays open, and strands which are held in unstable space may drift and bond to each other before the last strand is placed. Earlier work has described the energy structure of a single strand (Marrek, 2368), the sensitivity of coupled energy filaments to ambient subspace noise (Ilvane, 2374) and the bonding of adjacent filaments held under strain (Venn, 2371). In addition, the tactical record since 2268 contains accounts of escapes, but these were collected for a different purpose and give no aperture data (Tharnok, 2369). No study has measured both times in the same webs.

The aim of this paper is to solve this problem. Closure time has been modelled from the logged radius and ship count of each web, and the escape window has been tested against two candidate controls, the progress of the lattice (measured as closure time) and the local subspace variance. The work is written in 2376 from archive records which end in 2375.

2. Methods

The Starfleet Fleet Telemetry Archive holds the sensor logs of vessels which were enclosed by Tholian webs, and the Archive's Spatial Hazards Section has indexed them as a single series (Starfleet Fleet Telemetry Archive, 2376). The series lists 38 constructions. Seven were excluded, five because the log ended before the last passable aperture was lost and two because the hull envelope of the trapped vessel was not recorded. The remaining 31 webs were analysed. No web was selected or rejected after its values had been examined.

Five attributes were extracted for each web. The "radius" is the distance in thousand km from the position of the trapped vessel at the first strand to the lattice at closure, and it is a radius, not a diameter. The number of "ships" is the count of Tholian vessels whose emissions were logged on the lattice. The "subspace variance" is the root-mean-square fluctuation of the local subspace field in millicochranes, measured by the trapped vessel in the ten minutes after the first strand was detected, by the field-noise procedure of Ostrovane (2373). The "time to closure" ran from the first logged strand to the moment the aperture log showed no gap at all. The "escape window" was calculated from the first strand to the last moment at which any aperture exceeded the logged hull envelope of the trapped vessel by 20%.

The terminal event of each web was also recorded. The last passable aperture was classed as lost by "placement" if a Tholian ship laid a strand across it, and as lost by "reclosure" if neighbouring strands drifted together and bonded with no strand being placed. This class was fixed from the aperture log before any model was fitted, and it does not use the variance.

Closure time was modelled on the logarithmic scale (natural logarithms throughout) by ordinary least squares, with log radius and log ship count as predictors. If strands are laid at a constant rate by each ship, and the number of strands needed is proportional to the area of the enclosing surface, the exponents should be 2 and −1, and the closure time becomes T = K R2/N. The constant K was then estimated with the exponents fixed. It combines the strand count per unit area and the laying rate per ship, and these two cannot be separated by closure times alone. Because a single K fits all 31 webs, the laying rate per ship is treated as constant, and the progress of each web is represented by its closure time.

The escape window was analysed in the webs lost by reclosure, because in a web lost by placement the window and the closure time are the same quantity by definition. In the reclosure webs, log window was regressed on log variance and on log closure time, first separately and then together. Predictive accuracy was measured by leave-one-out cross-validation, as the root-mean-square error on the log scale. Because the samples are small, intervals use the t distribution with the residual degrees of freedom of each model (28 for the closure model, 30 for K, and 23 and 22 for the window models), and all intervals are 95%. No correction for multiplicity was applied, since the comparison between the two controls is the only planned one.

3. Results

The 31 webs spanned radii from 14 to 120 thousand km, ship counts from 2 to 6, variance from 0.05 to 1.10 millicochranes and closure times from 32 to 704 minutes, and Table 1 gives them by terminal event, with the last passable gap lost by reclosure in 25 webs (80.6%, 95% CI 63.7 to 90.8) and by placement in 6.

Closure followed the lattice geometry closely. The exponent on radius was 1.94 (95% CI 1.85 to 2.03) and that on ship count was −0.94 (95% CI −1.07 to −0.81), with R2 = 0.986 and a residual standard deviation of 0.109 on the log scale, or about 11%. Both intervals include the values of 2 and −1 expected from the geometry. With the exponents fixed, K was 0.294 (95% CI 0.282 to 0.306) when time is in minutes and radius in thousand km.

Ship numbers follow from this fit within the observed range of two to six ships. A web of 50 thousand km radius closes in about 120 minutes with six ships (0.294 × 502/6 = 122.5), and a web of 100 thousand km radius with six ships, the most observed, needs about 490 minutes.

Variance separated the two groups. The six webs lost by placement had variance from 0.05 to 0.13 millicochranes, and all of the 25 webs lost by reclosure had variance of 0.15 or more, so that the two groups do not overlap. In fact, the six placement webs were also the smaller webs, with radii from 14 to 35 thousand km, and in them the window was equal to the closure time, a median of 60 minutes.

In the 25 reclosure webs the window was a small part of the closure time, with a median ratio of 0.14 (range 0.02 to 0.57). The window fell with variance with a slope of −1.03 on the log scale (95% CI −1.19 to −0.87) and R2 = 0.89. The fitted line gives a window of 114 minutes at 0.2 millicochranes, 44 minutes at 0.5 and 22 minutes at 1.0. Closure time had no detectable relation to the window, with a slope of 0.15 (95% CI −0.35 to 0.64) and R2 = 0.016.

When both predictors were entered, the slope on variance was −1.04 (95% CI −1.20 to −0.88) and that on closure time was −0.05 (95% CI −0.22 to 0.12). The correlation between the two predictors was −0.18, so neither concealed the other. Cross-validated error was 0.228 log units for variance alone, 0.682 for closure time alone and 0.236 for both. In ordinary terms, the typical cross-validated error was 26% of the window from variance and 98% from closure time, and adding the closure time did not improve on variance.

4. Discussion

The escape window in a Tholian web is set by local subspace variance. In unstable space it is not set by the progress of the lattice, and the data show this in three separate ways. The window is almost inversely proportional to variance, with a slope close to −1. It bears no relation to the closure time that the geometry of the lattice determines, and the typical cross-validated error from variance alone is 26% against 98% from the closure time.

Closure time and window therefore need separate models, and each is now available. Closure time follows the geometry of the lattice. A larger web needs more strands, in proportion to the area enclosed, and each added ship shortens the work by the expected share. This is the quantity which a Tholian commander controls by the number of ships assigned, and it answers, for webs of two to six ships, the question of how many ships a web requires. The window is the quantity which a trapped commander needs, and it follows the field in which the web is laid, because strands in a disturbed field drift together and bond before the last strand is placed.

For a crew the consequence is direct. Variance is measured by the trapped vessel in the ten minutes after the first strand, with standard sensors, and the window can then be calculated as about 22 minutes divided by the variance in millicochranes. A vessel at 0.5 millicochranes has some 44 minutes, whatever the size of the web or the number of ships building it. In addition, a crew which estimates its time from the visible progress of the lattice will overestimate it, because the window was a median of 14% of the closure time in the webs where variance was 0.15 or above.

The progress of the lattice is therefore the wrong quantity to watch from inside a web.

Calm space shows the other case. Where variance was below 0.14, the last gap was sealed by a strand, and the window and the closure time were the same. These webs were small and few, and the model makes no claim about the range between them and the smallest unstable webs, but within the data the division between the two cases fell at a variance of about 0.14.

5. Limitations

All six placement webs were small, with radii of 35 thousand km or less, so calm space and small size are confounded in them. The statement that rate sets the window in calm space therefore rests on six webs, while the finding about unstable space rests on 25 and is not affected. Variance was measured by the trapped vessels with sensors of several classes between 2268 and 2375, and calibration has been reviewed only within a single sensor class (Sarrick, 2372). Any such differences add scatter, and they would weaken the slope, not produce it. Seven of 38 constructions were excluded because the record was incomplete, and webs of which no log survives are not represented at all. Variance was fixed at the value measured in the first ten minutes, which is the only value a crew would have in time to act on, and any later change in the field during construction was not modelled. The terminal event was read from the aperture log by the archive's own classification, and a web in which both processes acted within the same minutes would have been assigned to the one that closed the final gap.

Radius was taken as the distance from the trapped vessel to the lattice, which is a radius. If a later study defines the web by its diameter, K falls to a quarter of the value given here, to 0.073, and the exponents, the window and every result on variance are unchanged. The Tholian ships were observed only through their emissions, and the count of ships is the count of distinct emitters.

It follows from the data that the problem of the escape window has been solved for webs laid in unstable space. In future work, the aperture criterion will be tested in containment lattices of subspace field engineering, where the same drift and bonding should set the lifetime of any gap, and the open question is whether variance can be measured from outside the lattice.

Tholian weblattice closuresubspace varianceescape windowclosure timefield measurement

References

  1. Starfleet Fleet Telemetry Archive, Spatial Hazards Section (2376). Aperture and emission logs of Tholian web constructions, 2268–2375. Starfleet Fleet Telemetry Archive, series TW-4.
  2. Marrek, T. (2368). Energy structure and tension of a single Tholian strand. Journal of Applied Field Dynamics, 8(1), 17–44.
  3. Ilvane, S. (2374). Sensitivity of coupled energy filaments to ambient subspace noise. Subspace Engineering Quarterly, 20(2), 91–118.
  4. Tharnok, P. (2369). Accounts of escape from energy lattices in the sectors bordering Tholian space, 2268–2368. Starfleet Tactical Analysis Reports, TAR-69-117.
  5. Venn, A. (2371). Bonding of adjacent energy filaments under lateral strain. Subspace Engineering Quarterly, 17(4), 233–259.
  6. Ostrovane, L. (2373). A field-noise procedure for short-interval subspace variance on small vessels. Journal of Applied Field Dynamics, 13(2), 65–90.
  7. Sarrick, M. (2372). Within-class calibration of Fleet subspace sensors, 2360–2371. Starfleet Corps of Engineers Reports, SCE-SN-72-09.
Read this article inside the full journal experience — browse by faculty, search across universes, and filter by discipline.
Open in Uncited Press →