The early medieval period was the age of “feudal warfare,” which means the primary method of enlisting men was through obligatory service. Vassals were obligated to devote 40 days of military service per year, without monetary return (Bentley 1997, 204; Goodwin 2012). This system meant armies were disorganized and made up of men with limited military experience (Goodwin 2012).

By the time the War of the Roses began in the 15th century, problems with military organization had improved greatly. Society was still loosely feudalistic (bastard feudalism), but improvements in “bureaucracy” and “techniques of literate administration” led to a more professional army (Goodwin 2012, 16). Emphasis on the importance of governmental administrations meant that one of the primary means of enlisting men in the late medieval period utilized legal contractual agreements where lords provided land or monetary reimbursement for a vassal’s allegiance and military assistance upon request (Goodwin 2012, 124; Gravett 2002, 14). Many of these vassals would have been proficient military men with training, such as knights, mercenaries, and men-at-arms (who were often kept as retainers by nobles or knights) (Bell et al. 2013; Goodman 2006; Goodwin 2012). Outside of these contractual agreements, English commoners were also enlisted through active campaigning carried out by influential members of society (magnates, gentlemen, and religious leaders) who had the connections, wealth, and charisma needed to convince people to take up arms (Goodwin 2012, 18–19). In sum, the men who fought in the War of the Roses would have been from mixed social and economic backgrounds, with varied degrees of military experience.

Increased emphasis on contractual agreements amplified the ability to tax subjects on a large scale, which allowed military operations to become larger and more organized (Goodwin 2012, 17). Despite these new advantages, armies were still very expensive and difficult to supply for extended periods of time, so campaigns during the War of the Roses continued to be relatively short endeavors that occurred during the winter when farming was not at its peak (Goodwin 2012). This means that late medieval armies are unique when compared with modern standing armies; standing armies would have been placed under the strenuous conditions of the military lifestyle for extended periods of time, but medieval armies would have been exposed to these condition in short bursts. Though the period of exposure would have been shorter for medieval armies, the struggles would have been much the same: overcrowding, unsanitary/inadequate living conditions, and being undersupplied (Royle 2008). Indeed, fatal diseases did occur among armies during the War of the Roses. For instance in the summer of 1485, a highly fatal infection labeled as the “English sweating sickness” took hold across England. Whether this disease originated within the army is debated, but nearly 50 years after the outbreak of the disease, Polydore Vergil wrote that the disease first appeared in August of 1485 among the French mercenary army of Henry VII (Sadler 2013, 16; Smallman-Raynor and Cliff 2004, 81).

Despite the fact that medieval armies do not fit the traditional concept of a standing army, issues such as overcrowding and unsanitary conditions were still recorded among 15th century armies, along with outbreaks of disease (Sadler 2013; Smallman-Raynor and Cliff 2004). For this reason, it was decided that examination of the Towton assemblage would still be appropriate. It was further speculated that this distinction would later prove to be an interesting comparison for the military assemblages obtained from 18th- and 19th-century conflicts. Outside of the historical background of this assemblage, the small number of individuals in the Towton assemblage made it ideal for a pilot study.

Operational definitions were designed to categorize erosive arthropathies. These definitions and the references used to create them can be found in Table 1. Clinical and bioarchaeological research was utilized to create these definitions and the primary categorizations included: erosive arthropathy (EA), rheumatoid arthritis (RA), erosive osteoarthritis (Erosive OA), gout, spondyloarthritides (SpAs), reactive arthritis (ReA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS). Use of these definitions throughout the study insured conformity in the identification of pathology.

Prevalence is the statistical measure of frequency used in this research (Waldron 2007). Prevalence is calculated by taking the number of individuals with the condition of interest and dividing it by the total number of individuals in the defined assemblage. For the purpose of this pilot study, prevalence was used to provide general figures of the total number and types of EA present among skeletal adults of the Towton and All Saint’s assemblages that could then be compared. Statistical analysis utilized the Confidence Interval Analysis Software to calculate prevalence and provide 95 percent confidence intervals.

Identification of EA requires the presence of several skeletal elements for confident diagnosis. The typical method of inclusion into the study dictates that all of the skeletal elements used for diagnosis must be present (Waldron 2007). This method would produce a drastically underestimated prevalence of EA because a large number of skeletons would have to be excluded from the study; archaeological skeletons are rarely complete, meaning it is highly unlikely that all of the key skeletal elements needed for diagnosis of erosive conditions (feet, hands, spine, and pelvis) would be present (Roberts and Manchester 2010; Waldron 2007).

With these details in mind, a method for inclusion and exclusion into the study was developed. Rather than considering every element singularly in each skeleton, a method was designed to measure the preservation of the key elements as a whole. Skeletons were assigned to one of four potential categorizations: poor (the majority of the key elements were unobservable due to absence and/or poor preservation, elements present <45 percent); moderate (approximately half of the key elements were observable, but absence and/or poor preservation was still considerable, key elements present 46–65 percent), good (the majority of the key elements were observable, elements present 66–85 percent), and excellent (a near complete skeleton, key elements present >86 percent).

Assigning skeletons to these categories allows the data to be analyzed in levels. A “comprehensive” prevalence would consider all of the skeletal material in the defined population (adult males/males?) regardless of their completeness; this measure certainly underestimates the true prevalence since it assumes all of the missing elements did not have evidence of erosive pathology. Critical analysis would exempt skeletons without relevant erosive pathology assigned to the “poor” or “poor” and “moderate” categories. This analysis accepts that there is an increased probability of underestimating the prevalence when skeletons with a low percentage of observable elements are included in the study, but decreases the probability of underestimating the prevalence if only categories with a high percentage of observable elements are allowed into the study. This method of dealing with missing elements does not eliminate the issue of prevalence underestimation, but it is a critical method that is more inclusive than the alternative and is capable of providing comparable results.

Currently the best method of ageing adult skeletal material is to have a multifactorial approach which uses several methods to determine a likely age range (Bedford et al. 1993). The key features examined for age estimation in this study were: epiphyseal fusion (only an option in young adults, based on McKern and Stewart 1957), dental attrition (compared to Brothwell 1981), changes to the pubic symphysis (using Brooks and Suchey 1990), and changes to the auricular surface of the pelvis (using Lovejoy et al. 1985). After comparing the ages assigned by the various methods, individuals were placed into one of four potential categorizations based on ten year increments, which were: Young Adult (16–25), Lower Middle Adult (26–35), Upper Middle Adult (36–45), and Mature Adult (46+).

As with age estimation, sex estimation was done with a multifactorial approach. Both morphological and metric data was utilized. The human skull and pelvis express the largest amount of sexual dimorphism, so these morphological features were accessed using Buikstra and Ubelaker (1994). Sex was also determined through consideration of metric measurements. The primary measurements taken were the maximum diameter of the femur head (FHD1) and the maximum diameter of the humeral head (HuD1) (Bass 1971). Measurements >46.5 mm were considered to be male, unknown between 43.5–46.5 mm, and female in measurements <43.5 mm.

Towton 25 had two erosions. The first erosion was on the right first metatarsal on the distal end, medial surface. This erosion was para-articular, round, and deep. The walls were made of thickened and smooth trabecular bone. The second erosion was on the right scaphoid. This erosion was round and marginal, with uneven scalloped margins. The interior surface of this erosion exposed irregular and dense trabecular bone. These characteristics were not enough to clearly understand the nature of these erosions.

The skeleton of Towton 41 had erosions on the distal end of the right fifth proximal phalanx. These erosions were marginal and made of irregular bone on the para-articular lateral surface. This was the only location where erosions were observed, but the erosive changes were extensive and eroded enough bone to distort the natural shape of the joint margin. Once again, these erosions were not characteristic enough to be categorized beyond the EA categorization.

Towton 50 was the last skeleton to have unspecifiable EA. Two of the erosions were suspect since they were not characteristic of what is normally observed in EA or pressure lesions such as bunions. These lesions occurred on the distal end of the right and left first metatarsals on the medial surface. The holes were deep and round in shape with the interior walls being made of dense trabecular bone. The lesion on the left metatarsal had an overhanging bridge of bone around the margin, but it did not appear to be a Martel hook as would be seen in gout. Though these mystery lesions could not be easily identified as true erosions, clearly defined erosions were also present. The first true erosion was observed on the superior surface of the right navicular. Finally, erosions were observed on the distal end of the left second proximal phalanx and affected the margins of both the medial and lateral surface and was accompanied by subcortical resorption on the joint surface.

The first skeleton fitting with the SpA classification was Towton 8 (see Fig. 1), which had several foot erosions, characteristic bone formation in the feet, and sacroiliac joint involvement. The focus of the erosions and bone formation in the feet is suggestive of ReA, but minimal spinal involvement and lack of sacroiliac joint fusion prevented confident diagnosis based on the operational definition. Two vertebrae (L1 and S1) both had para-vertebral bone formation that seemed more characteristic of ossification of ligamentous tissue than the marginal osteophytes observed in spinal degeneration; the surface of the vertebral body did not show pitting and the bone growth originated inferior to the joint’s true margin. The foot erosions affected numerous elements, including: the distal end of the left fifth proximal phalanx (also associated with ankylosis of the distal interphalangeal joint), the distal end of the left fourth intermediate phalanx, the distal end of the left first proximal phalanx, and the inferior margin of the right navicular. In addition to these erosions, both the right and left shaft of the fifth metatarsals had ReA-characteristic entheses associated with the plantar interossei muscle. Lastly, there was unilateral sacroiliitis of the right sacroiliac joint, which was defined by: marginal erosions on the sacrum (superior), an area of subcortical resorption on the joint surface, and lipping suggestive of early ossification of the anterior sacroiliac ligament.

Towton 9 also fit the SpA classification (see Fig. 2) and was characterized by: extensive lower limb enthesopathy, asymmetric fusion of three consecutive thoracic vertebrae (T6-T8), ligamentous ossification in T9 –T12, sacroiliitis, and small erosions in the feet. The lower limb enthesopathy and small erosions in the feet are characteristic of ReA, but the spinal fusion did not have observable skip lesions (unfused vertebrae separating segments of fused vertebrae). Though there were no skip lesions, the pattern of ligamentous ossification in several other vertebrae suggested that further fusion may have eventually occurred. Though sacroiliac joint fusion was not observed in Towton 9, bilateral sacroiliitis was present and defined by enthesopathy of the anterior sacroiliac ligament and a thin layer of porous bone growth on the joint surface.

The final SpA identified was Towton 13 (see Fig. 3). The features used to categorize Towton 13 were: enthesopathy (primarily in the lower limbs), left and right para-vertebral ligamentous ossification in several thoracic vertebrae (T7 –T12) suggestive of eventual fusion, sacroiliitis, and small erosions in the feet. Once again, lower limb enthesopathy and erosions are a distribution pattern seen in ReA. This pattern was observed in Towton 13 apart from entheses on the humeri. The erosions observed in Towton 13 occurred in the left foot, including the distal end of the first metatarsal and the distal end of the second proximal phalanx. Though there was no spinal fusion in Towton 13, ligamentous growth was observed in the left and right para-vertebral position and was suggestive of eventual fusion most characteristic of ReA or PsA; the position of the bone growth observed in the spine is not characteristic of the anterior fusion seen in cases of AS and was not characteristic of the bone growth seen in Diffuse Idiopathic Skeletal Hyperostosis (DISH), which is confined to the right side in the thoracic region of the vertebral column (Waldron 2009). The sacroiliac joint was not fused in Towton 13, but sacroiliitis was present and defined by bilateral growth in the anterior sacroiliac ligament and pitting on the joint surfaces.

The spinal fusion observed in 2705 was uncharacteristic of DISH because skip lesions were present; thoracic vertebrae T6-T8 were fused in the right paravertebral position, T9-T10 were unfused, and T11-L1 were fused in the right and left paravertebral position. A SpA would explain the observed pathology of the spine in skeleton 2705, but erosions, enthesopathy, and sacroiliac joint involvement were all absent. Since no pathology was observed outside of the spinal involvement, skeleton 2705 could not be classified as any form of EA.

The problem of a SpA diagnosis in skeleton 3114 is a mix of contradicting characteristics, post-mortem damage to the spine, and moderate preservation (the majority of the left hand and all off the left foot were absent). Skeleton 3114 exhibited extensive spinal fusion, one foot erosion, upper and lower limb enthesopathy, and extensive sacroiliitis.

The observed spinal fusion was confined to the right side and involved T5-L1. The fusion in the thoracic and lumbar vertebrae appeared to have been continuous, but post-mortem fragmentation made it imposable to determine if this was truly the case. If continuous para-vertebral fusion could be confirmed, then DISH would be considered a reasonable explanation for the observed spinal pathology, but there are other features of the spinal bone growth that are uncharacteristic of DISH as well. The spinal bone growth of skeleton 3114 was not limited to the ligament, but had spread onto the joint surface in some of the vertebral bodies, which is not a feature of bone formation in DISH (Resnick 1996), but can be a feature of SpAs (Waldron 2009). Though sacroiliac joint fusion has been reported in DISH (Waldron and Rogers 1990), inflammatory changes (sacroiliitis) are not normally linked to the condition (Resnick 1996). The presence of sacroiliitis in skeleton 3114 in addition to the questions of the spinal fusion made DISH a questionable diagnosis and suggested that SpA may be a more satisfactory explanation for the observed pathology, but there is also the question of the erosions located on the distal end of the right fifth metatarsal. These erosions had clear undercut edges with exposed trabecular bone, but they were not marginal. The erosions occurred on the articular surface, which is a characteristic feature of gout. The presence of erosions in combination with the other pathology seems suspiciously in favor of a SpA, but it is not impossible for the individual to have had DISH and gout simultaneously, making the arthropathy of skeleton 3114 highly complicated. Though an argument could be made for a SpA categorization (spinal fusion of a questionable nature, enthesopathy, sacroiliitis, and erosions), skeleton 3114 was classified as unclassifiable EA since this would account for both gout and SpA.