In September 2017, 28-year-old Caroline Witford vanished without a trace on a challenging trail in the San Juan Mountains.

For 5 years, she was presumed dead, believed to be the victim of a wilderness accident or some sudden mishap amid the Colorado backount.

But in October 2022, she walked into an ordinary gas station in the town of Montrose, alive, but unrecognizable.

What she told police once she was able to speak left even seasoned investigators stunned.

Where she had been for those 5 years and what really happened to her, you will find out in this video.

On the morning of September 12th, 2017, early autumn weather in the San Juan Mountains was clear but cold and dry with sunlit Rocky peaks creating thin bands of light mixed with mist and the Blue Creek Spur Trail enveloped in the characteristic silence of the Colorado high country.

Caroline Witford, 28 years old, a medical volunteer with Silverton Wilderness Aid, was the first to arrive at the starting station that day.

As an experienced medic with nearly four years supporting local search and rescue teams, Caroline was familiar with the mountainous terrain, old mining roads, and the harsh weather of the San Juan region.

Her assignment that day was to check the safety conditions of Blue Creek Spur, a narrow trail running along the north slope of Gladstone Basin, after rangers received reports of rockfall in the area.

Following protocol, Caroline left the Silverton aid station early in the morning carrying basic medical supplies and communication equipment, then drove an off-road vehicle to the trail head near Gladstone Mine.

A Forest Service camera at the entrance captured her stopping to check her harness, adjusting her jacket, then stepping onto the relatively dark, steep, rocky section shaded by the mountain.

That was the last data the monitoring system provided.

According to the plan, Caroline was expected to complete the check and return to the station by early afternoon that same day, but by nearly 300 p.m., the coordination team had received no communication from her radio.

Colleagues at the station tried contacting her multiple times, but the signal remained completely silent, an anomaly since Caroline always checked in periodically.

When teammates confirmed that Caroline had not returned to the main road and no other vehicles were parked near the trail head except hers, they immediately notified her family.

By late afternoon, after all contact attempts failed and with no sign of Caroline at any other ranger station or trail, the family officially reported her missing to the San Juan County Sheriff’s Office.

From that moment, Caroline Witford’s file was opened as missing, marking the beginning of one of the strangest disappearance cases ever recorded in the Gladstone Mountain area.

As soon as the report was received, the San Juan County Sheriff’s Office, in coordination with Silverton Wilderness Aid, immediately shifted to initial response mode, starting by establishing the LKP, last known point.

Investigators retrieved footage from the Forest Service camera and cross-referenced it with the last communication timestamp in the morning.

The camera at the trail head near Gladstone Mine recorded Caroline entering the trail at 720.

And from that moment onward, no further signals came from her radio or any tracking device.

Establishing this LKP became the focal point for deploying the entire search force as Blue Creek Spur lay amid numerous unofficial side paths created by hikers with no data indicating Caroline had left the area under her own power.

The SR search and rescue team was activated within less than an hour of the missing person file being opened.

The lead coordinator quickly divided personnel into three main teams.

One following the initial section of Blue Creek Spur from the trail head, another approaching from the Gladstone Basin direction to sweep the Rockfall area, and a third sweeping secondary paths leading toward the eastern slope.

The goal of the initial phase was to determine whether Caroline had left the main trail to avoid rockfall or had slipped into one of the deep rock gullies common in the area.

Search teams deployed K9 units trained for tracking missing persons.

Each dog handled by an experienced San Juan County handler.

The initial path was relatively clear, but from 300 to 500 m past the trail head, the terrain grew more complex with clusters of brush, rock slides, and broken ground, typical of old mining areas.

These factors made tracking difficult, especially with no information on her direction of travel after the camera’s last capture.

In the air, a Colorado Rescue helicopter scanned along the mountainside, focusing on ridge lines where hikers might become lost or fall.

Pilots used wide sweeps to look for reflections from high visibility clothing or the medical pack, while also checking crevices and openings into narrow canyon bottoms, though visibility was limited by heavy shadows from the Gladstone cliffs.

Meanwhile, the drone team deployed three small flying units, scanning from above in a grid pattern to cover areas the helicopter couldn’t reach.

Drones proved especially effective for surveying talis fields, steep slopes of loose rock debris, where visual searches were extremely difficult.

However, even after expanding the radius to nearly a mile around the LKP, teams found no movement or signs of survival.

High-risk areas such as the creek bed, a dry stream below Blue Creek Spur were also thoroughly checked.

Sarah teams followed every segment, looking for slide marks, scrapes on rocks, or drag marks from boots.

But all that was noted were fresh rockfall layers with no clear evidence of Caroline’s presence.

By the end of the first search day, the summary report stated no signs of foul play, no clear signs of accident, no evidence Caroline left the trail intentionally.

This left investigators challenged because in most San Juan disappearances, associated items usually left some small trace, a scrap of fabric, a slide mark, bent grass, but in this case, everything was completely blank.

On the second day, the search area expanded further northeast where a long slope led into Gladstone Basin.

K9 teams were again deployed here, but the dog showed only very weak reactions in a few spots before quickly losing the scent.

This led to the initial hypothesis that Caroline may have moved across hard rock surfaces that retained no odor or footprints.

Remaining teams continued upward to higher ridgeel lines using thermal binoculars to detect heat signatures, but all equipment registered no signals consistent with a living person or a body.

By the third day, after completing standard search patterns for a wilderness missing person case, the team recorded one single reality, Caroline appeared to have left this trail segment without leaving any physical evidence.

This forced S to reassess and expand to smaller side branches, old mining roads, and sparse forest near the Gladstone mine.

However, nothing specific pointed to where Caroline might have gone.

At the close of phase 1, the search team’s report stated clearly no victim, no signs of foul play, no evidence of accident.

The only option was to expand the search farther into more inaccessible terrain as Caroline may have moved beyond the originally anticipated routes.

In the following days, after completing the entire initial search perimeter with no clear signs found, the Asair team and San Juan County Sheriff’s Office were forced to significantly scale up operations, prioritizing Gladstone Basin and nearby abandoned mines.

Gladstone Basin features complex terrain with uneven steep slopes, rock slides stretching hundreds of meters and partially collapsed old mining roads.

Always a major challenge for any missing person search to access these locations.

Ser divided into small teams, each including a guide, medic, terrain specialist, and at least one K-9 handler coordinating with local rescuers familiar with the mining road network around Red Mountain.

As teams entered the basin from the southeast, some groups found potential signs related to Caroline, though none were strong enough to conclude anything.

A small piece of blue fabric caught on brush on a steep slope with unclear tear patterns, possibly from natural forces or friction, was documented, but could not be proven to belong to Caroline due to the common color and lack of distinct identifiers.

At another spot near a partially collapsed mine road, an old piece of medical tape blown into a dirt gully was collected.

However, SAR could not confirm it was Caroline’s or left by another hiker, as the entire area had been a popular trekking spot before weather worsened in recent years.

Even with hope these items were connected, their scattered nature and lack of accompanying footprints or slide marks prevented teams from determining Caroline’s next direction of travel.

teams continued expanding into scattered abandoned mines around Gladstone, including shallow tunnels and old shafts that SAR had to access by rope and collapse prevention gear.

Yet, across all check sites, no signs indicated anyone had entered near the time of Caroline’s disappearance.

Air quality, floor moisture, and collapsed structures in many places ruled out use as shelter.

The complete absence of fresh footprints, survival items, or any distress signals made the situation increasingly baffling as typical missing person cases usually leave at least some direct or indirect signal for searchers.

Meanwhile, weather deteriorated significantly.

From day 18 onward, Gladstone Basin saw repeated small hail storms, sharply dropping nighttime temperatures, strong winds making drone flights unstable, and helicopters limiting access to canyons due to turbulence.

Steep rock fields became slippery.

Rockfall risk rose, forcing SAR to shorten daily work hours for minimum safety.

Terrain experts brought in to assist assessed that after three continuous weeks of sweeping all key areas with no signs of human activity, the likelihood of Caroline being in Gladstone Basin was near zero.

An internal report stated clearly, “If Caroline had an accident, her body should have left secondary traces such as impacts on rocks, altered flow in the creek bed, or disrupted rock debris distribution, but none appeared.

There was also no evidence she left the area in another direction as all exit routes had been flagged and thoroughly checked.

On day 32, after the disappearance, SAR submitted a campaign performance evaluation report to the Sheriff’s Office and Colorado Search and Rescue Board.

The report concluded that the operation had covered all areas where the victim could plausibly be, including those classified as low probability but necessary to check.

Remote hazardous spots that protocol required ruling out.

Yet, every result circled back to the starting point.

No victim, no signs of movement, no evidence of death, no evidence of crime.

The final phase of the search took place on day 40, mainly to re- sweep Gladstone Ridge and the western portion of Blue Creek Spur to ensure no oversightes.

Long range drones with infrared cameras made a final pass over narrow valleys and hidden rock gullies, but the data still revealed no anomalies.

With worsening weather and rising terrain hazards, SAR was forced to suspend the campaign per federal rescue protocols.

At this point, the San Juan County Sheriff’s Office held a meeting with Caroline’s family to update the case status.

With no new data emerging and all search methods fully exhausted, police officially classified Caroline Witford’s case as presumed dead, meaning presumed deceased, based on the prolonged disappearance and the wilderness environment where it occurred.

Search operations ended here.

The file was closed and it was added to the list of unsolved missing person cases in the Gladstone area.

In early October 2022, more than 5 years after Caroline Witford’s missing person file was classified as presumed dead, a completely unexpected event occurred in the town of Montrose, about 60 mi from the Gladstone area by Mountain Road.

Around 6:00 a.m. the onduty employee at a small gas station on Townsend Avenue spotted a woman appearing at the edge of the premises, staggering, her body shockingly emaciated, and her clothing tattered and worn in a way unlike any transient the area had ever encountered.

She stood close to the wall, avoiding the headlights of a parked truck, repeatedly shielding her eyes with her hand and showing clear signs of panic when anyone approached.

The employee immediately called 911, reporting a woman in critical condition, apparently completely disoriented.

When the first Montrose Police Department patrol car arrived, officers noted that the woman had no identification, did not respond to direct questions, and only nodded or shook her head very slowly.

Her gaunt face, long tangled hair, and extremely thin body revealing prominent bones led them to suspect she might be the victim of prolonged abuse or severe starvation.

After several minutes of trying to calm her, they were finally able to get close enough to observe identifying facial features.

One of them recognized a resemblance to the photo in the missing person bulletin that had been circulated internally in Southwest Colorado.

Though initially no one dared draw a conclusion because so much time had passed and her condition differed so drastically from the 2017 images.

The shift supervisor decided to use descriptive data from the NCIC system for a quick comparison of basic identifiers.

Height, facial bone structure, mole, and eyebrow characteristics.

The matches were sufficient for police to make a preliminary determination that this woman could be Caroline Witford, the individual presumed missing and deceased in the Gladstone Basin area.

This coincidence was immediately reported through internal channels along with a request for emergency medical assistance.

Recognizing the victim’s extreme exhaustion, officers did not wait for an ambulance, but promptly placed her in the patrol car and transported her directly to Montrose Regional Health Hospital, just a few minutes drive away.

N route, Caroline remained unable to answer any questions, only gripping the seat edge tightly and continuously avoiding light shining on her face, indicating pronounced sensory disturbance after a long period of disappearance.

Upon arrival at the hospital and handover to the emergency team, initial support procedures were implemented immediately while authorities activated the protocol for cases of reappearance after long-term disappearance.

Immediately after Caroline Witford was moved to the secure treatment area, the hospital coordinated with investigators to carry out the first step, identity confirmation, using mandatory biometric methods.

DNA samples were collected within 2 hours of admission, including blood and oral mucosal cells and rushed to the Colorado Bureau of Investigation Laboratory for comparison with the archive 2017 missing person sample.

Simultaneously, investigators took Caroline’s fingerprints using a specialized electronic kit.

Because her fingertips were dry, cracked, and scaly from malnutrition, scanning had to be repeated multiple times to achieve acceptable clarity.

The investigative agency also requested the dental records previously submitted by the family to the sheriff in 2017, including dental x-rays and treatment notes from the Silverton Dental Clinic.

These steps aim to establish three fully independent layers of verification, DNA, fingerprints, and dental, to completely eliminate any possibility of misidentification.

While these procedures were underway, the emergency medical team assessed Caroline’s initial physical condition and documented several clear abnormalities.

First was her BMI falling below the minimum threshold required to sustain basic bodily functions accompanied by severe wasting in the muscles of the lower and upper limbs.

The leg muscle structure showed prolonged atrophy consistent with infrequent ambulation over an extremely long period.

Her pallet skin and flaking and thin patches indicated a lack of natural light exposure and likely severe vitamin D deficiency, conditions impossible in someone living outdoors or lost in the wilderness, as the Colorado mountain environment, though harsh, still provides abundant sunlight.

More notably, doctors observed numerous scars encircling her wrists and ankles, darkened and deeply embedded in the skin with characteristics compatible with prolonged restraint using hard materials.

Some scars were old and fully healed from years earlier, while others were more recent, indicating that the restraint was not limited to an initial phase, but continued through multiple cycles.

X-ray examination of her bones revealed mild osteoporosis, a condition commonly associated with prolonged immobility and lack of light exposure.

Additionally, Caroline’s cardiovascular and respiratory functions were reduced compared to the average for a healthy 28-year-old woman, consistent only with restricted nutrition and an unventilated living environment.

Taken together, all these factors clearly indicated that she could not have survived in a natural environment, even a harsh one, during her disappearance.

Instead, she had been confined in an enclosed dark space with restricted movement and likely provided minimal food on a cyclical basis.

While the medical team evaluated her physical condition, results from the identification tests began returning.

Fingerprint verification came back first.

Despite damage to her fingertips, the APHIS system was able to identify key features in the prints, matching Caroline’s 2017 record at a level sufficient for preliminary confirmation.

Next, the CBI Dental Laboratory sent a report confirming that features on the X-ray films, including the filled tooth hash4, the eruption angle of tooth hash8, and an old crack in an incizer from a 2012 accident, matched Caroline’s records completely.

Finally, the DNA results from the lab returned that same afternoon, confirming an absolute match with no discrepancies whatsoever.

Thus, all three independent methods produced the same result.

The woman who appeared in Montrose was Caroline Witford, missing for over 5 years in the Blue Creek Spur area.

Immediately after identity confirmation, the sheriff’s and CBI investigative teams compiled all medical and identification data, preparing an urgent report that explicitly noted the following.

Caroline exhibited multiple signs consistent with a victim held in a controlled environment.

Her injuries, restraint scars, and physiological condition were incompatible with survival in the wild.

There were no signs of impact from wild animals or terrain accidents, and the prolonged damage indicated years of restricted movement and limited light exposure.

The report concluded that Caroline could not have survived in the forest or surrounding areas of Gladstone Basin.

Instead, she had been held in an enclosed environment, a form of artificial space or underground structure throughout her entire disappearance.

This was the first medical basis to establish that the 2017 disappearance was not an accident or natural incident, but bore clear signs of prolonged deprivation of Liberty.

During the postidentification evaluation phase, Caroline was asked to provide any sensory information she still remembered with the goal not of reconstructing events, but of collecting raw data to aid the investigation.

The CBI investigative team recorded three main categories of consistent sensory information: sounds, smells, and surface characteristics that Caroline could describe reliably despite fragmented event memory.

Regarding sounds, Caroline described a small engine noise that occurred cycllically, not from large trucks or industrial equipment, but with a tone similar to an ATV or off-road vehicle engine, commonly used in old mining areas of San Juan.

Investigators noted that this engine type has a distinctive frequency profile and is widely used for travel over rough terrain, allowing access to areas without clear trails.

The second sound Caroline described was prolonged metallic reverberation with a characteristic echo of metal objects striking hard rock surfaces.

Acoustic analysis concluded that this sound profile fits granite tunnels or deep abandoned mines where sound can reverberate in multiple layers and linger for several seconds.

This was significant data because Galena and granite in the Hawkins and Gladstone areas produce similar reverberation effects, but only in sufficiently deep and enclosed tunnels.

Smell information was recorded with even higher consistency.

Caroline stated that the dominant odor she remembered was damp cement characteristic of hydraulic mortar used to reinforce old tunnels.

This smell differs from natural clay or rock and is rarely found outside areas previously repaired or reinforced against collapse during mining operations.

From this, investigators ruled out natural caves or fractured rock passages lacking human intervention history.

The third category concerned surface characteristics.

When asked to describe the sensation underfoot during movement, Caroline described a rough but even surface distinct from natural soil or loose rock slides.

This matches leveled tunnel rock floors or partially reinforced mine roadways where the surface is hard and maintains a more stable incline than external terrain.

Additionally, Caroline clearly recalled a sensation of descending slopes during many movements, not ascending or flat terrain.

From this, investigators concluded a high likelihood that her location was within a tunnel system with an entrance at a higher elevation than the holding area or a mine with inclined shaft levels common in old mines around Red Mountain and Gladstone.

This movement data helped eliminate enclosed cabins or natural caves in the area as they lack the sustained incline Caroline described.

The investigative team continued synthesizing all three sensory categories to establish a probable location range.

The small engine sound was consistent with areas accessible by ATV, ruling out extremely steep or inaccessible regions.

Damp cement odor indicated reinforced structures, excluding most fully collapsed mines and natural caves.

Prolonged metallic reverberation required a narrow, deep space with hard rock walls and a relatively intact tunnel shape.

The descending slope sensation indicated an entrance at higher elevation with a consistent incline along the travel path.

Combining these four factors, the team built a mapped range based on San Juan Mountains geological data.

Specialists narrowed down candidate mines by cross-referencing mining history, tunnel maps, ATV accessibility, entrance locations, and areas previously reinforced with hydraulic cement.

The results showed only a very small number of mines fully met all criteria, including deep tunnels with preserved structure, concrete patch repair history, and ATV accessible trails.

These sites were assigned priority codes for ground survey.

The conclusion from the sensory information collection was clearly documented in the record.

Although the victim had limited memory of events and timelines, her descriptions of sounds, smells, and surfaces provided sufficiently strong technical grounds to identify a potential confinement location range.

The identified factors were consistent and highly locationational in value, allowing investigators to significantly narrow the search area from the entire Gladstone and Red Mountain region down to a small group of mine structure sites.

At the end of this analysis phase, CBI proceeded to the next step in the process, cross-referencing terrain features with accessible abandoned mines to screen locations for direct field investigation.

Based on the three groups of sensory information collected from Caroline, the investigative team moved to the terrain narrowing step using technical data.

The Colorado Bureau of Investigation coordinated with the US Bureau of Land Management and Forest Service to retrieve the complete list of old mines, abandoned tunnels, and underground mining structures within a 15-mi radius of Caroline’s LKP at Blue Creek Spur.

This data system included information on depth, reinforcement history, drift direction, host rock type, and collapse condition.

A total of more than 40 mine sites were recorded within this range, stretching from Gladstone Basin to Red Mountain Pass.

However, to filter out the mines that could match Caroline’s sensory description, the investigative team established a set of mandatory criteria.

The first criterion was depth.

Caroline described a prolonged echoing metallic sound which only occurs in mine tunnels with sufficient depth and rock wall systems that maintain acoustic reflection.

Shallow mines were eliminated from the start because they could not produce this sound effect.

The second criterion was slope direction.

The information Caroline provided about continuously moving downhill ruled out horizontal addits or room and pillar or storage tunnels while retaining only those tunnels with a clear incline heading into the mountain.

The third criterion was rock texture.

Descriptions of the sound and surface suggested hard granite or metamorphic rock structures as soft rock or clay walls would not produce the type of metallic echo Caroline recalled.

The fourth criterion was a history of concrete reinforcement since Caroline noted the smell of damp cement in the confinement space.

This meant the tunnel must have been repaired with artificial materials to prevent collapse, reducing the number of matching mines to about 1/3.

In addition, the investigative team excluded mines recorded as fully collapsed before the year 2000 or mines with no ATV access because all the sounds Caroline described suggested the perpetrator used motorized vehicles for movement.

When the full set of criteria was applied to the database from BLM and Forest Service, the initial 40 mine sites were narrowed down to only seven points in the potential group.

From there, the investigative team proceeded to a more detailed analysis by cross-referencing 3D terrain maps with the ATV trail systems used by miners and locals.

They assessed accessibility for each mine site based on factors including entry road gradient, ground stability, distance from main roads, and ATV operability under limited visibility conditions.

Some locations, although structurally suitable, were eliminated because access roads had been filled in or were impassible throughout 2017 2022 or because satellite data showed no signs of use on those mine roads for many decades.

After this filtering step, only three mine sites fully met the requirements.

An old tunnel east of Gladstone Basin, a deep tunnel near the edge of Red Mountain Pass and Hawkins, no do seven, a mine with a long history of extraction and multiple phases of reinforcement.

The investigative team reviewed records for each mine and found that the tunnel east of Gladstone had been eroded by water since 2012, causing the entire internal structure to collapse in the middle section.

This ruled out the possibility of a stable artificial room for holding a person for years.

The tunnel near Red Mountain Pass was eliminated because it lacked the extended incline sections Caroline described and most of the internal rock walls were slate, a type of rock that does not produce the characteristic metallic echo.

With these two sites ruled out, only two candidates fully met the technical criteria.

Hawkins number Seven and White Swan Tunnel.

Hawkins number Seven was notable for its multi-level tunnel system connected by old inclines with some sections previously reinforced with concrete to prevent rockfalls.

While White Swan Tunnel featured thick granite structure, significant depth and recorded occasional motorized vehicle access in the past 10 years.

Both locations were at a distance but could be reached by ATV departing from Gladstone Basin.

After narrowing the list to these two sites, the investigative team decided to prepare for in-depth survey forces, they developed an approach plan, including a geology team, forensics team, K9 unit, mine technicians, and support security group.

Technical documents for both mines were printed for analysis of entry structures, tunnel ceiling stability, segment depths, and collapse risk locations to ensure safety during the survey.

ATV usage history around the two mines was also cross-cheed with Forest Service GPS data to determine the feasibility of motorized activity in the area.

With all these parameters, the sheriff’s office and CBI jointly agreed to direct field surveys at the two target mines, starting with Hawkins number 7 due to its structure best matching Caroline’s sensory description.

After deciding to select Hawkins number 7 as the first survey site, the investigative team coordinated with geological engineers and mine safety specialists to conduct a full external structural inspection before entering Hawkins.

Number seven is an old mine with multiple interconnected tunnel levels.

Some sections had collapsed since the 1980s, but most tunnels still retained their hard granite structure.

The technical team used seismic vibration measurement equipment, pressure sensors, and long shaft cameras to assess entry stability.

Initial inspection results showed that the main entry, although lightly eroded, remained sufficiently stable to allow the survey team to proceed deeper, if following the rock wall edges and avoiding the steep incline area near the mouth.

Once acceptable safety conditions were confirmed, the in-depth survey group began approaching the internal tunnels.

They split into two teams.

The first team followed the main tunnel heading southwest, where old maps recorded a long incline leading down to lower levels.

The remaining team followed the eastern auxiliary tunnel branch, which had been partially mined for ore, but abandoned decades ago.

Outside, the technical team continued monitoring vibration and structural changes via equipment placed at the tunnel mouth to signal alerts for any localized collapse.

On the first level, the granite surface clearly showed mechanical tool extraction marks with no signs of recent repairs.

However, as the survey team moved deeper into the southwest branch, humidity increased and uneven broken rock fragments appeared.

At this point, the K9 team was deployed to detect any residual human scent traces.

One of the K9 suri acted strongly when passing a narrow section initial tail wagging followed by the characteristic alert behavior for detecting biological traces.

The handlers noted a cool air current blowing out from a low height rock crevice, suggesting a void or tunnel connection to another area inside.

Since abandoned mines often have collapsed sections causing stagnant air, the technical team used airflow measurement equipment to determine wind stability.

Results showed the air current had an unusually stable cyclic pattern inconsistent with natural mine conditions and suggesting the presence of an artificial passage or a controlled ventilated enclosed space.

When examining the rock wall where the K9 reacted most strongly, specialists discovered a patch of rock with a different color from the rest.

While natural granite in the tunnel had a characteristic gray color and rough surface, this area was lighter gray, smoother, and flatter, suggesting the wall had been patched with artificial material, most likely a concrete patch mixed and applied to cover a passage or isolate an area behind it.

By lightly tapping the area with a geological hammer, the survey team recorded a hollowower, more resonant sound compared to natural rock, indicating a possible open space behind the patch.

Surface temperature equipment placed against the patch showed a slight temperature differential, suggesting the interior air had a stable heat source, a sign often seen in areas with past survival activity or mechanical equipment.

The technical team marked this area as a high-level point of interest and prepared an official report recommending breaching the patch following safety protocols.

Before proceeding with breaching, the investigative team continued gathering supplementary data by inserting small diameter tube cameras into natural crevices around the wall.

The captured images, though limited, showed a low ceiling void behind the patch with relatively intact rock surfaces and a metallic object lying against the floor function not yet identifiable.

The cool air current from the rock crevice was also confirmed to have regular oscillating rhythm consistent with human created or maintained ventilation rather than natural air flow in a collapsed tunnel.

After synthesizing the factors, including the K9 reaction, abnormal air flow, artificial patch, and camera data, the survey team concluded that this location had a high probability of leading to a human modified or intentionally used space.

The team leader drafted a document to CBI command and the sheriff’s office recommending elevation of this area to critical point of interest status, meaning a site requiring direct opening and inspection.

With full safety conditions met, the technical team marked breaching points just sufficient to avoid collapsing surrounding structures while preparing ventilation equipment and deep lighting for the next phase of the survey.

When the technical team completed marking the breaching points on the material suspected to be a concrete patch, CBI’s mine specialized forensics unit was dispatched to Hawkins number seven to carry out the concealment breach following risk control procedures, ensuring no collapse of surrounding structures.

A low power core cutting device was used instead of conventional breaking hammers to avoid strong vibrations in the already weakened tunnel space.

After approximately 40 minutes of continuous drilling and cutting, the artificial material was revealed to be a panel composed of hydraulic cement mixed with fiberglass fixed to a thin steel frame to create a uniform surface with the rock wall.

When the panel was removed, a thick steel door approximately 5 cm thick appeared behind it without an external lock, but using an internal sliding bolt, indicating the creator intended the door to operate one way only.

The forensics unit used wide-angle lighting to inspect the door surface and recorded inward opening slide marks, proving it had been used multiple times.

After confirming safe gas levels and tunnel ceiling stability, the team pushed the steel doors bolt inward forcefully.

And as the door moved, a cold air rush escaped from the space behind, accompanied by heavy moisture carrying metallic and cement odors.

This space was the point the investigative team had been trying to identify for many weeks.

A concealed confinement room inside Hawkins number 7.

When full lighting was activated, the confinement room structure became clearly visible.

The interior space measured approximately 3 and 1/2 m in length, 2 1/2 m in width, and just over 2 m in height, with the rough granite ceiling still bearing traces of extraction activity from decades earlier.

Both sidewalls were reinforced with concrete patches at varying heights, indicating multiple repairs or structural modifications over time.

The floor of the confinement room was leveled rough rock, not natural soil, proving the space’s creator had ground the rock surface to create a more stable standing or lying area.

The ventilation system consisted of two small diameter metal pipes attached to the right wall, venting air to a higher position in the tunnel.

The pipes showed mixed fresh and old scratches and rust, indicating prolonged manual adjustment and maintenance.

In the left corner of the room was a single metal bed frame with legs deeply embedded into the rock floor, showing it had been fixed in place to prevent movement.

The bed’s position relative to the ventilation path supported the hypothesis that this was used as a sleeping or holding area for a victim over an extended period.

As the forensics team began collecting samples, numerous signs of prolonged human activity emerged.

On the rock floor near the foot of the bed, they found long reddish brown hair strands adhered to the surface due to condensed moisture.

Some strands retained intact hair follicles suitable for DNA analysis and comparison.

On the left side wall, the investigative team discovered a dark brown staining patch forming a vertical streak.

When tested with peroxidase reagent, it gave a positive reaction, confirming dried human blood.

In the lower corners of the floor, embedded in rock crevices, the forensics team found very small tissue fragments, possibly skin shed from rubbing or scratching.

These samples were collected using low pressure suction techniques to avoid destroying DNA elements.

Next to the bed were two metal rings drilled directly into the rock floor with clear wear marks on their inner surfaces.

The wear pattern was consistent with ropes or chains hooked in for extended periods, and the metal’s oxidation level suggested they were used not just once, but over multiple cycles.

On the bed edge, the forensics team recorded another rust streak about 10 cm long, likely caused by water or other liquid flowing along the surface during use of the confinement room.

Hand tools were also found at the scene.

a small coiled piece of copper wire, a loose screw, and a stiff frayed cloth piece with abrasion stains, items that could have been used for temporary repairs to the room structure during confinement.

A key point in the examination was determining the time frame of this space’s use.

On the concrete patch layers reinforcing the walls, the investigative team found uneven dust accumulation.

newer dust layers over older ones allowed estimation that structural intervention in the mine occurred within approximately the last 5 years.

Rapid on-site chemical analysis showed the hydraulic cement type was commonly sold in the 2016 2021 period in the Colorado area supporting the hypothesis that this confinement room not only existed long-term but was maintained until recently.

The forensics team continued sampling from the ventilation pipes where they found degraded fabric fibers and epithelial cells, indicating prolonged human occupancy and sufficient ventilation to maintain minimum oxygen levels.

Under preliminary UV light analysis, they detected faint fluorescent streaks running along the base of the walls, consistent with repeated body contact marks under restricted movement conditions.

All samples were sealed and immediately transferred to the CBI laboratory for DNA and tissue structure analysis.

DNA analysis results returned a few days later, confirming that all collected hair samples, including those with intact follicles and mechanically broken segments, matched Caroline Witford’s genetic profile.

Skin tissue samples also matched, proving the victim had been present in this confinement room during multiple periods.

The blood sample on the wall showed long-term oxidation but originated from the same source with DNA comparison consistent with Caroline.

These data combined with analysis of the confinement room structure showed that this space had been used continuously for approximately 5 years precisely matching the period of Caroline’s disappearance.

The investigating agency concluded in the scene report that Hawkins number seven contained a secret confinement room created and deliberately maintained and victim Caroline Witford had been held there throughout her absence.

This examination marked a critical turning point as it was the first time the investigation campaign identified a specific location directly linked to Caroline’s prolonged confinement, providing concrete physical evidence for the entire case file.

After completing the examination of the cell at Hawkins number 7 and collecting sufficient physical evidence, confirming Caroline’s prolonged presence in this space, the investigative team proceeded to analyze the modus operandi to determine how the perpetrator abducted, transported, and maintained her captivity for many years without detection.

Based on field data and the sensory information provided by Caroline, investigators developed the hypothesis that the perpetrator most likely approached Caroline at or near Blue Creek Spur via an ambush.

This is a narrow area with obstructed visibility due to rocks and brush, allowing the perpetrator to lie in wait while remaining difficult to detect.

The possibility of using anesthesia was considered highly likely because Caroline had no clear memory of the initial moment of being subdued.

This aligns with the perpetrators intent to minimize noise or dragging marks while facilitating victim transport without resistance.

Based on Caroline’s description of repeated small engine sounds across multiple phases, the investigative team determined that the perpetrator likely used an ATV to move from the abduction site to the entrance of Hawkins number seven.

An ATV is well suited to the steep rocky terrain of the San Juan region and also minimizes human footprints left behind.

The fact that Caroline left no traces during the initial search further supports this hypothesis.

The perpetrator most likely placed Caroline on the ATV immediately after subduing her, removing her from the trail without passing through areas where Sarah teams could find biological or physical signs.

Regarding why the holding location went undetected during the 2017 search, the investigative team concluded that the cell structure was concealed with cement panels and a steel frame skillfully integrated with the natural rock wall, making it difficult for any search team lacking specific scent data or signals to recognize.

Additionally, most search efforts focused on surface terrain, rock crevices, and natural trails.

While Hawkins number seven is a deep abandoned mine with no signs of recent collapse, so it was not in the initial SR priority sweep group.

The perpetrator’s use of artificial panels matching the color of the surrounding granite demonstrates a clear understanding of camouflage techniques in a mining environment.

Not only did the perpetrator conceal the cell entrance, but he also selected a location deep in the second suble where air flow and odors had reduced escape potential, making K-9 detection difficult unless approached very closely.

Furthermore, the perpetrator maintained the cell using a manual ventilation system consisting of small metal pipes installed at an inclined angle to channel air from higher level tunnels down into the room.

This mechanism ensured minimal oxygen for the victim while avoiding strong airflow currents that search teams could detect.

The lighting inside the room was not fixed, but most likely portable based on burn marks and rust traces found on the brackets.

This indicates the perpetrator only brought equipment in when needed, minimizing the chance of leaving traceable items.

Regarding the locks and restraint system, two metal rings drilled directly into the stone floor proved the perpetrator had knowledge of working with hard materials and possessed specialized tools.

The uniform drilling depth and symmetrical placement on both sides of the bed demonstrate careful calculation to secure the victim without compromising the room’s structural integrity.

The combination of concealment panels, ventilation pipes, fixed bed, and restraint rings proves the cell was designed as a long-term holding space rather than a temporary confinement site.

From all the above data, the investigative team constructed a preliminary profile of the perpetrator.

First, the perpetrator possessed deep knowledge of mine structures, including rock composition, granite behavior under vibration, and reinforcement techniques.

Second, the perpetrator had skills in using hydraulic cement, reinforced mortar, and drilling cutting tools in confined spaces.

Third, the perpetrator knew how to operate an ATV in steep rocky terrain, and had knowledge of unmarked mine paths.

Fourth, the perpetrator was capable of sustaining long-term resupply to the cell without leaving surface traces.

All these factors indicate that the perpetrator was not an ordinary hiker, but someone with a technical background related to mining operations or maintenance of old tunnels.

From this, the investigative team reached an initial conclusion.

The criminal behavior was deliberately planned, not impulsive, and the perpetrator possessed sufficient specialized knowledge to create and operate a secret cell for many years without detection.

During the examination of all evidence recovered from the cell at Hawkins no 7, the forensic team paid special attention to a rare detail with tracing potential, the locking mechanism attached to the steel door found behind the cement panel layer.

Unlike the rest of the cell, which was largely handmade, this lock had a clear serial number on the inner body.

It was an industrial-grade steel lock designed for high-pressure doors, not a common type sold in civilian hardware stores.

When the analysis team traced the serial number, they found this product line was distributed on a limited basis to certain mechanical equipment stores serving the mining industry in Colorado.

Manufacturer data showed the matching serial batch was shipped to three distributors between 201521 with only one store in Silverton maintaining retail sales records.

The investigative team contacted this store and requested invoices within the time frame matching Caroline’s disappearance.

The store’s manual sales record showed only one purchaser of this lock type during the relevant 2-year period.

A man named Elias Mercer, age 43.

Mercer paid in cash but provided contact information for warranty purposes as required for industrial lock purchases.

When the purchase data was forwarded to the investigation center, the background analysis team quickly identified that Elias Mercer had previously worked as a mine engineer on restoration and safety inspection projects for abandoned mines in the Silverton and Oay areas.

His employment records documented many years of experience in tunnel wall reinforcement, ventilation system installation, and deep tunnel incident handling, skills that directly corresponded to how the cell at Hawkins number 7 was constructed and maintained.

Further investigation revealed that Mercer had worked freelance after leaving a project in 2014 and held no long-term contracts with any mining company, yet he still owned tools and specialized knowledge.

This was corroborated by a series of invoices for hydraulic cement, steel, mesh, and hand tools purchased from various local stores, coinciding with the period of periodic cell maintenance.

The investigative team cross-cheed personal information and discovered Mercer had a registered ATV matching the sound Caroline described.

This ATV was a model commonly used in steep, rocky terrain and was also the type of engine noted near abandoned mines in the region.

Additionally, residency records showed Mercer lived in an isolated cabin on the outskirts of Silverton, only a few miles from the Hawkins Nober 7 entrance via off-road paths.

The cabin had no close neighbors and was not in an area with traffic cameras facilitating unmonitored movement.

Investigators also examined Mercer’s social circumstances.

He lived alone, had no close family in the area, and very limited community interaction.

Neighbors described Mercer as reclusive, rarely interacting with anyone, and often traveling by ATV at night.

Forest Service data recorded multiple instances of unidentified ATV tire tracks near Hawkins number 7 during the period of Caroline’s disappearance, suggesting regular movement activity.

When all this information was synthesized, it formed a consistent pattern.

Mercer had access to the area where Caroline disappeared, possessed the technical skills to build and maintain the cell, owned a compatible vehicle matching the description, and had a purchase history aligning with evidence found at Hawkins number seven.

With these elements, CBI officially placed Elias Mercer on the primary suspect list for the case.

An internal report was prepared stating that no other individual in the area combined similar technical capability, means of transport, residence location, and material usage behavior matching what was discovered in the cell.

Continue reading….
Next »