Civil War Incident History

The Raid at New Creek

A surprise Confederate cavalry attack on a fortified Union railroad depot, November 28, 1864.

New Creek, West Virginia Baltimore & Ohio Railroad Fort Kelley / Fort Fuller About 700 Union prisoners
Camp of the 22nd Pennsylvania Cavalry near New Creek with the valley visible behind it
Camp of the 22nd Pennsylvania Cavalry near New Creek. The unit helped construct Fort Kelley, one of the Federal works guarding the Baltimore & Ohio Railroad supply depot Camp photograph.
Introduction

A fortified depot on a vulnerable railroad line

New Creek, present-day Keyser, West Virginia, occupied a strategic point in the upper Potomac region because it sat directly on the Baltimore & Ohio Railroad. During the Civil War, the B&O served as a vital Union transportation corridor, carrying troops, military stores, animals, food, weapons, and intelligence across the mountains between the Ohio Valley and the eastern theaters of war B&O context.

The Federal post at New Creek was therefore more than a local garrison. Fort Kelley, also called Fort Fuller in some later accounts, protected the depot, the railroad shops, the wagon stores, and the surrounding valley approaches. Camps and picket lines spread across the nearby hills and roads. The post's strength depended not only on earthworks and artillery, but on warning, scouting, and a clear understanding of Confederate movement through Hardy and Hampshire Counties New Creek map.

On November 28, 1864, Confederate cavalry under Maj. Gen. Thomas L. Rosser turned that system against itself. Using speed, concealed movement, and a deception at the picket line, Rosser's command overran the position, captured artillery and hundreds of Union soldiers, and destroyed valuable railroad and quartermaster property Official Records.

The War Department reports also reveal that the raid continued beyond the fort itself. As New Creek collapsed, nearby Union detachments tried to understand whether they could reinforce, retreat, or defend the railroad line. Reports from Maj. Peter J. Potts, Col. William C. Starr, and Capt. John Fisher show that the incident widened into a moving crisis along the New Creek-Piedmont corridor Official Records.

Primary-Source Method

Reading the Official Records as evidence

The War Department reports are not neutral summaries. They are battlefield explanations written by officers who were assigning responsibility, defending decisions, reporting losses, and claiming credit. For that reason, this page reads the reports comparatively: Union accounts by Sheridan, Crook, Latham, Potts, Starr, and Fisher are placed beside Confederate accounts by Lee, Rosser, and Payne Official Records.

Several facts are stable across the reports: Rosser's cavalry reached New Creek on November 28, 1864; men in Federal uniforms helped capture the pickets; the fort and artillery were taken rapidly; hundreds of Union soldiers became prisoners; and railroad, quartermaster, and commissary property was destroyed. The disagreements are equally important. Sheridan and Crook framed the event as a disgraceful failure of resistance, while Rosser and his superiors presented it as a well-executed cavalry raid.

"The conduct of this garrison is a disgrace to the service."

Maj. Gen. Philip H. Sheridan, November 29, 1864 report

"By dressing my advance in blue [I] completely surprised the garrison."

Maj. Gen. Thomas L. Rosser, Confederate report report
Prelude To Disaster

Warnings, uncertainty, and the failure of readiness

In the days before the raid, Union commanders received warnings that Confederate cavalry might be moving toward Moorefield and the Baltimore & Ohio Railroad. Maj. Gen. Philip H. Sheridan directed Bvt. Maj. Gen. George Crook to maintain vigilance toward New Creek; Crook in turn ordered Col. George R. Latham to scout the roads and prepare the garrison Prelude documents.

The problem was not the total absence of intelligence, but its uncertainty. Latham later insisted that reports of Confederate concentration near Moorefield had circulated for some time, with enemy numbers estimated anywhere from 1,200 to 3,000. He also emphasized that scouts had been kept on the roads and that morning reports on November 28 still failed to locate the enemy near Burlington Latham report.

"For some time I had been receiving what I regarded as reliable information of a concentration of rebel troops at Moorefield. Their numbers were variously estimated at from 1,200 to 3,000."

Col. George R. Latham Official Records

That intelligence gap coincided with a second vulnerability: Potts had left New Creek at 10:30 a.m. on November 27 with about 150 cavalrymen to scout toward Moorefield. The detachment's absence reduced the forces immediately available at the post and placed Potts in a difficult position once firing began behind him at New Creek Potts report.

Document card preview for intelligence reports before the New Creek raid
Compiled intelligence and correspondence from the days before the raid. The documents frame New Creek as a failure of interpretation as much as a failure of warning Prelude documents.
Moorefield Engagement

The screen between Rosser and New Creek collapsed

Sketch of Rosser's cavalry action near Moorefield, Virginia, November 27, 1864
A. Rosser's cavalry advances north up the valley, driving back the Union screen.
B. Union cavalry near Heiskel's and Moorefield attempts to observe or block the advance.
C. The map notes captured Union artillery on the Romney road.
D. River crossings and road junctions opened the route north toward New Creek.
E. McNeill's Company operated on the eastern flank, screening Confederate movement.
Rosser's cavalry action near Moorefield, November 27, 1864. The map shows the clearing of the southern approach toward New Creek Moorefield map.

Confederate movement Union positions

On November 27, Confederate cavalry advanced along the South Branch of the Potomac and encountered Union screening forces near Moorefield. These Federal horsemen were supposed to detect, delay, or identify Confederate movement before it reached the railroad corridor.

Potts's report adds the human problem behind the map. After leaving New Creek, he learned from civilians that Rosser had left Mount Jackson and was moving toward Moorefield. He turned back, heard firing at New Creek, and concluded that his small command could not assist the garrison because his retreat to the post had been cut off Potts report.

"I could distinctly hear the firing at New Creek, and knew that the post was attacked."

Maj. Peter J. Potts Official Records

The map clarifies why the engagement mattered. Rosser's force pressed north up the valley (Callout A), while Union cavalry near Moorefield and Heiskel's (Callout B) failed to stop or fully identify the scale of the advance. Road junctions and river crossings (Callout D) opened the route toward New Creek, and McNeill's Company operated on the flank (Callout E), limiting Union visibility Moorefield map.

The defeat at Moorefield removed the final warning line between Rosser's cavalry and the New Creek garrison.
Attack At New Creek

Speed, deception, and a sudden collapse

Map of New Creek and vicinity showing Fort Kelley and the Federal camp captured by Rosser's cavalry
F. Fort Kelley and the main Federal camp anchored the Union position.
G. One Confederate column advanced north from the Ridgeville direction.
H. A second Confederate column moved up the valley east of the fort.
I. The Baltimore & Ohio Railroad made New Creek a critical supply hub.
J. The map marks captured artillery during the collapse of the Union position.
New Creek and vicinity, showing Fort Kelley, the Federal camp, the railroad corridor, and Confederate attack routes converging on the post New Creek map.

Confederate movement Union positions

On November 28, Rosser's cavalry struck New Creek with speed and precision. The official map shows Confederate columns converging from the south and southeast (Callouts G and H) toward Fort Kelley and the Federal camp (Callout F). The railroad line (Callout I) explains the strategic value of the target: New Creek guarded a major Union transportation and supply hub New Creek map.

The attack began with deception at about 11:30 a.m. Confederate troopers wearing captured Federal uniforms approached the pickets on the Moorefield road. Latham wrote that the pickets, supposing the riders to be Union men, allowed them to get so close that they were captured without firing a warning shot Latham report.

"The advance guard was dressed in Federal uniforms. The pickets, supposing them to be our own men, permitted them to approach so close that they were captured without firing a gun."

Col. George R. Latham Official Records

Latham's description gives the attack a precise geography. The Confederate column moved along the road at the foot of the hill, between the fort and the mountain. The first clear warning came from firing at the creek crossing roughly 250 yards from the fort; by then, the advance was already within thirty yards of Latham's tent and firing on him as he tried to reach the fort Latham report.

With only about 160 armed men available, Latham's command was quickly overwhelmed by a force he estimated at roughly 2,000 cavalry. He later claimed to have rallied about 30 men and held the creek crossing on the Moorefield road until the Confederates retired, but he also noted that the artillery lacked ammunition except what remained in the limber boxes Latham report.

"The panic and stampede became in a few moments hopelessly general." Col. George R. Latham, Official Records report
Prisoners And Captivity

Hundreds of Union soldiers moved from battlefield to prison system

The fall of New Creek produced a mass capture, though the official reports do not give a single identical number. Rosser claimed 700 prisoners; Lee's dispatch reported 26 commissioned officers and 400 or 500 enlisted men; public marker accounts summarize the result as more than 700 Union soldiers captured. The reports agree on the essential point: the garrison suffered a large-scale capture rather than a limited battlefield loss Official Records historical marker.

"He took 8 pieces of artillery, 26 commissioned officers, and 400 or 500 enlisted men."

Gen. Robert E. Lee, reporting Early's account Official Records

The captured men were not the only prize. Rosser reported taking eight pieces of artillery and about 1,500 horses and mules, while Lee emphasized artillery, prisoners, wagons, stores, and a railroad bridge. These details show that the raid was designed as a logistical strike as much as a prisoner-taking operation Official Records.

The author's ancestor, Hamilton McGarvey, has a service record that connects him to New Creek. This incident page, however, treats individual named prisoners as part of a larger military event: the collapse of the fortified supply depot, the scale of the capture, and the competing interpretations produced by Union and Confederate commanders.

Historical marker titled Capture of New Creek
The Capture of New Creek marker preserves the public-memory version of the event: surprise, enemy uniforms, more than 700 prisoners, captured artillery and horses, and destroyed supplies marker.
Piedmont And The Railroad

The raid widened after New Creek fell

The Official Records make clear that the disaster did not end at the fort. Col. William C. Starr, commanding at Piedmont, reported that he received a telegram from Latham on the morning of November 28 saying the enemy was advancing on New Creek and that an attack was expected. Around 1 p.m. Starr heard artillery fire in the direction of New Creek; by 2 p.m. a telegraph operator reported that the fort had been captured and the garrison made prisoners Starr report.

By about 4 p.m., Rosser's main body appeared near Piedmont. Starr ordered Capt. John Fisher's company to fall back to the Maryland side of the river and take position on the hills. Fisher later reported forming his men after word came that New Creek had fallen, skirmishing with the advance guard, and withdrawing in good order when the main body appeared Fisher report.

"The enemy advanced into the town, and immediately commenced the work of destruction."

Col. William C. Starr Official Records

Starr's report records the infrastructure damage in concrete terms: the B&O machine shops, engine houses, roundhouse, cars, depot building, quartermaster and commissary stores, and telegraph office were burned. This destruction confirms that the expedition aimed to damage the railroad system supporting Union operations, not merely to seize a fort Starr report.

Union And Confederate Reaction

One event, two sharply different meanings

Union reaction: disgrace, arrest, and blame

Union commanders treated the loss of New Creek as a humiliation. Sheridan reported that Rosser had captured the garrison and that it had "made no resistance," concluding that the conduct of the post was a "disgrace to the service" Official Records.

Crook likewise reported that Rosser's command, numbering from 2,000 to 3,000, came down on New Creek about 1 o'clock, "completely surprising the garrison," and took the fort and artillery "without the firing of a single gun." He ordered Latham into arrest and directed that escaped officers and men be collected at Cumberland Crook report.

Confederate reaction: skill, boldness, and success

Confederate reports presented the raid as an efficient cavalry success. Rosser wrote that he "captured the fort and 8 pieces of artillery, 700 prisoners, and about 1,500 horses and mules," while destroying wagons, stores, and railroad property with a loss of only two killed and two wounded Official Records.

Confederate endorsements praised the expedition's skill and boldness. Fitzhugh Lee forwarded the report as an operation conducted with "great skill and boldness," Jubal Early called it "very brilliant," and Robert E. Lee concluded that its results reflected credit on Rosser and his command indorsements.

Conclusion

The raid was a chain failure, not a single surprise

The disaster at New Creek was not caused by one moment alone. The picket-line deception mattered, but it succeeded because earlier safeguards had already weakened: intelligence remained uncertain, Potts's scouting detachment was away from the post, the Union screen near Moorefield collapsed, many men at New Creek were not ready for immediate combat, and Confederate cavalry used terrain, timing, and disguise to compress the defenders' reaction time.

The Official Records also broaden the event beyond a single fort. New Creek's fall created a crisis at Piedmont, damaged B&O infrastructure, and generated sharply different interpretations from Union and Confederate commanders. To Sheridan and Crook, it was a disgraceful collapse; to Rosser and his superiors, it was a model cavalry raid that seized prisoners, artillery, animals, wagons, and supplies while damaging a vital railroad line.

Timeline

Primary-source chronology

The sequence below follows the times and movements reported by Latham, Potts, Starr, Fisher, Rosser, and Payne. It is placed after the narrative because the reports make the most sense once the geography and command perspectives are established.

Rosser's division reportedly leaves camp near Mount Jackson in the Shenandoah Valley and moves toward Moorefield Potts report.

Potts leaves New Creek with about 150 men of the Sixth West Virginia Cavalry to scout toward Moorefield Potts report.

Rosser reaches Moorefield; Payne's brigade attacks and drives off the Union scouting party, opening the route toward New Creek Rosser report.

Latham's scouts return from the Moorefield road and report no enemy as far as Burlington Latham report.

Confederates in Federal uniforms approach and capture the pickets on the Moorefield road without a warning shot Latham report.

Starr hears artillery firing from the direction of New Creek while commanding at Piedmont Starr report.

A telegraph operator reports to Piedmont that New Creek has fallen and the garrison has been made prisoner Starr report.

Rosser's main body appears near Piedmont, prompting Starr and Fisher to withdraw to the Maryland side of the river Starr and Fisher reports.

Documents

Document cards and source evidence

These cards follow the Hamilton archive style: each item identifies the type of evidence, explains how it supports the reconstruction, and links to the source image, transcription, or public archive.

Official records

The Disaster at New Creek: November 28, 1864

Official Records transcription of the capture of New Creek
Union and Confederate reports describing the surprise raid, Latham's account of the picket-line deception, Potts's detached cavalry command, Starr and Fisher at Piedmont, the capture of the fort and artillery, disputed prisoner totals, and the destruction of railroad and quartermaster property. This is the central primary-source cluster for the incident.
Open transcriptionView original archive
Official records

Prelude to Disaster: Intelligence Before the Raid

Official Records correspondence and intelligence before the New Creek raid
Correspondence and intelligence reports from the days before the raid, showing warnings, uncertainty, scouting reports, and the failure to establish a reliable picture of Rosser's movement.
Open transcription
Battle map

Map of New Creek and Vicinity

Map of New Creek and vicinity showing Fort Kelley and Federal camp
Jedediah Hotchkiss's map shows Fort Kelley, the Federal camp, the B&O Railroad corridor, and the Confederate avenues of approach. It is essential for understanding how the attack converged on the post.
Open image
Battle map

Sketch of Action Near Moorefield

Sketch of Rosser's cavalry action near Moorefield
Map showing the cavalry engagement one day before New Creek. It explains how Union screening forces were driven back and why the route north opened to Rosser's command.
Open image
Historical marker

Capture of New Creek

Historical marker for Capture of New Creek
Marker text summarizing Rosser's surprise attack, the use of enemy uniforms, more than 700 captured Union soldiers, captured horses and artillery, and destroyed supplies.
Open HMDB entryOpen Clio entry
Primary source

Official Records, Series I, Volume 43

Official Records of the War of the Rebellion, Series I, Volume 43
The published War Department compilation that preserves Union and Confederate reports on the affair at New Creek and related operations in the Shenandoah and West Virginia theaters.
Open Cornell archive
Historical photograph

Camp of the 22nd Pennsylvania Cavalry at New Creek

Camp of the 22nd Pennsylvania Cavalry near Fort Kelley
Rare wartime view of a cavalry camp near New Creek. The photograph gives visual context for the military landscape around Fort Kelley and the railroad depot.
Open image
Secondary source

Private Uriah "Duck" Alley and the New Creek Raid

Zac Cowsert's article provides a modern narrative of the raid through another captured Union soldier, helping connect the Official Records to the lived experience of prisoners taken at New Creek.
Open source

Bibliography

Primary Sources

Hotchkiss, Jedediah. "Map of New Creek and Vicinity Showing Position of Fort Kelley and the Federal Camp Captured by Rosser's Cavalry Division, November 28, 1864." In Atlas to Accompany the Official Records of the Union and Confederate Armies. Washington, DC: Government Printing Office, 1891-1895.

Hotchkiss, Jedediah. "Sketch of the Action Near Moorefield, Virginia, November 27, 1864." In Atlas to Accompany the Official Records of the Union and Confederate Armies. Washington, DC: Government Printing Office, 1891-1895.

Parsons, George. "Camp of 22nd Pennsylvania (Ringgold) Cavalry, Union Army, New Creek (Keyser), W.Va." 1865. West Virginia & Regional History Center, West Virginia University. West Virginia History OnView. Accessed March 31, 2026. https://wvhistoryonview.org/catalog/000853.

United States War Department. The War of the Rebellion: A Compilation of the Official Records of the Union and Confederate Armies. Series I, vol. 43, pt. 1. Washington, DC: Government Printing Office, 1893. https://collections.library.cornell.edu/moa_new/waro.html.

Historical Markers and Public History

Historical Marker Database. "Capture of New Creek." Accessed 2026. https://www.hmdb.org/m.asp?m=152788.

The Clio. "Fort Fuller / Capture of New Creek Highway Historical Markers." Accessed 2026. https://theclio.com/entry/96505.

Secondary Sources

American Battlefield Trust. "Baltimore and Ohio Railroad in the Civil War." Accessed 2026. https://www.battlefields.org/learn/articles/baltimore-and-ohio-railroad-civil-war.

Cowsert, Zac. "Private Uriah 'Duck' Alley: The Story of West Virginia's Last Civil War Veteran." Civil Discourse, November 24, 2020. https://civildiscourse-historyblog.com/blog/2020/8/6/private-uriah-duck-alley-the-story-of-west-virginias-last-civil-war-veteran.