Bodies differ
The 3D models on this site are single bodies. Real spines vary, and the differences matter in surgery and in reading a scan. Here each difference is shown as the spread a study measured, with the study beside it. Averages describe no one exactly.
Where the spinal cord ends
In 629 healthy adults scanned by MRI, the lower end of the cord (the conus) sat most often opposite L1, but anywhere from the T11–12 disc to L2. The spine viewer’s model ends opposite L1.
Percent of participants by the vertebral level opposite the conus. Highlighted: this site’s model. Nakashima et al. 2021, Results, first paragraph.
“Multiple regression analysis showed that the pelvic incidence (p < 0.0001) and height (p < 0.0001) were significant factors affecting the conus medullaris level.”
The artery of Adamkiewicz
The largest artery feeding the lower cord arises at different levels in different people, usually on the left. Its level is checked before aortic and spinal surgery because cutting it can harm the cord.
Pooled percent by vertebral level of origin, 43 studies, 2834 people with the artery. Taterra et al. 2019, Table 4 (pooled prevalence by vertebral level of origin).
“Our main findings revealed that the AKA was present in 84.6% of the population, and patients most frequently had a single AKA (87.4%) on the left side (76.6%) originating between T8 and L1 (89%).”
Five lumbar vertebrae, or six, or four
At the bottom of the column the last lumbar vertebra can be partly fused to the sacrum, or the first sacral segment can look like a sixth lumbar vertebra. Counting levels then goes wrong unless someone counts from the top. The female reference body of the Human Reference Atlas, used for this site’s body models, has six lumbar vertebrae; the male has five.
Radiographs of 5941 people. French et al. 2014.
“The prevalence of LSTV in the study population was 9.9%. Lumbarized S1 and sacralized L5 were seen in 5.8 and 4.1% of patients, respectively.”
Female and male lumbar spines overlap
A CT study of 81 men and 77 women traced the lumbar curve along the spinal canal. Some measures differ on average; all overlap widely.
Lumbar segment length
Difference in standard deviations (Cohen’s d, women − men): −0.85. If both groups were normally distributed with equal spread, their curves would overlap by about 67%. Reported p value < 0.01.
Lordosis cranial peak height
Height of the curve measured from its upper end, as a fraction of segment length.
Difference in standard deviations (Cohen’s d, women − men): 0.71. If both groups were normally distributed with equal spread, their curves would overlap by about 72%. Reported p value < 0.01.
Lordosis caudal peak height
Height of the curve measured from its lower end.
Difference in standard deviations (Cohen’s d, women − men): 0.2. If both groups were normally distributed with equal spread, their curves would overlap by about 92%. Reported p value 0.310.
Means, SDs and ranges from Hay et al. 2015, Table 1. Effect size and overlap are computed here from those numbers; the bell curves are drawn as normal distributions, which the study does not claim.
“Males and females manifest different lumbar curve shape, yet similar amount of inward curving (lordosis).”
Even nerve roots vary
The phrenic nerve, which drives the diaphragm, is usually described as coming from C3, C4 and C5. A cadaver study found two-root and one-root patterns too.
“Classical three root (C3,4,5), double root with varying contribution from C3 or C5 (C3,4/C4,5), and single root (C4/C5).” Sharma et al. 2025
See the phrenic nerve on the Body page.
Composite model Distributions come from different populations and methods; compare within a chart, not across charts.