diff --git a/nimbus/core/chain/persist_blocks.nim b/nimbus/core/chain/persist_blocks.nim
index 7a2dd2db6..b7665fe0a 100644
--- a/nimbus/core/chain/persist_blocks.nim
+++ b/nimbus/core/chain/persist_blocks.nim
@@ -174,7 +174,7 @@ proc persistBlocksImpl(c: ChainRef; headers: openArray[BlockHeader];
   dbTx.commit()
 
   # Save and record the block number before the last saved block state.
-  c.db.persistent(headers[^1].blockNumber - 1)
+  c.db.persistent(headers[^1].blockNumber)
 
   if c.com.pruneHistory:
     # There is a feature for test systems to regularly clean up older blocks
diff --git a/nimbus/db/aristo/README.md b/nimbus/db/aristo/README.md
index c2717fbcc..75fc8c8fa 100644
--- a/nimbus/db/aristo/README.md
+++ b/nimbus/db/aristo/README.md
@@ -24,7 +24,7 @@ Contents
   + [4.4 Leaf record payload serialisation for account data](#ch4x4)
   + [4.5 Leaf record payload serialisation for RLP encoded data](#ch4x5)
   + [4.6 Leaf record payload serialisation for unstructured data](#ch4x6)
-  + [4.7 Serialisation of the list of unused vertex IDs](#ch4x7)
+  + [4.7 Serialisation of the top used vertex ID](#ch4x7)
   + [4.8 Serialisation of a last saved state record](#ch4x8)
   + [4.9 Serialisation record identifier identification](#ch4x9)
 
@@ -351,13 +351,11 @@ maximum value of that byte is 0x99.
           marker(8) is the eight bit array *0110-1011*
 
 <a name="ch4x7"></a>
-### 4.7 Serialisation of the list of unused vertex IDs
+### 4.7 Serialisation of the top used vertex ID
 
-        0 +-- ..
-          ...                                -- recycled vertexIDs
-          +--+--+--+--+--+--+--+--+
-          |                       |          -- last unused vertex IDs
-          +--+--+--+--+--+--+--+--+
+        0 +--+--+--+--+--+--+--+--+
+          |                       |          -- last used vertex IDs
+        8 +--+--+--+--+--+--+--+--+
           |  |                               -- marker(8), 0x7c
           +--+
 
@@ -399,7 +397,7 @@ i.e. the last byte of a serialised record.
 |   0xxx 0yyy | (x(3)<<4) + y(3) | Account payload      | [4.4](#ch4x4)       |
 |   0110 1010 | 0x6a             | RLP encoded payload  | [4.5](#ch4x5)       |
 |   0110 1011 | 0x6b             | Unstructured payload | [4.6](#ch4x6)       |
-|   0111 1100 | 0x7c             | List of vertex IDs   | [4.7](#ch4x7)       |
+|   0111 1100 | 0x7c             | Last used vertex ID  | [4.7](#ch4x7)       |
 |   0111 1111 | 0x7f             | Last saved state     | [4.8](#ch4x8)       |
 
 <a name="ch5"></a>
diff --git a/nimbus/db/aristo/aristo_blobify.nim b/nimbus/db/aristo/aristo_blobify.nim
index fd7dd6f9a..b35dcf850 100644
--- a/nimbus/db/aristo/aristo_blobify.nim
+++ b/nimbus/db/aristo/aristo_blobify.nim
@@ -153,18 +153,32 @@ proc blobify*(tuv: VertexID): Blob =
   ## Variant of `blobifyTo()`
   tuv.blobifyTo result
 
-
-proc blobifyTo*(lSst: SavedState; data: var Blob) =
+proc blobifyTo*(lSst: SavedState; data: var Blob): Result[void,AristoError] =
   ## Serialise a last saved state record
-  data.setLen(0)
-  data.add lSst.src.data
-  data.add lSst.trg.data
+  case lSst.src.len:
+  of 0:
+    data.setLen(32)
+  of 32:
+    data.setLen(0)
+    data.add lSst.src.data
+  else:
+    return err(BlobifyStateSrcLenGarbled)
+  case lSst.trg.len:
+  of 0:
+    data.setLen(64)
+  of 32:
+    data.add lSst.trg.data
+  else:
+    return err(BlobifyStateTrgLenGarbled)
   data.add lSst.serial.toBytesBE
   data.add @[0x7fu8]
+  ok()
 
-proc blobify*(lSst: SavedState): Blob =
+proc blobify*(lSst: SavedState): Result[Blob,AristoError] =
   ## Variant of `blobify()`
-  lSst.blobifyTo result
+  var data: Blob
+  ? lSst.blobifyTo data
+  ok(move(data))
 
 # -------------
 
@@ -343,6 +357,7 @@ proc deblobifyTo*(
   ## De-serialise the last saved state data record previously encoded with
   ## `blobify()`.
   if data.len != 73:
+    echo ">>> deblobifyTo got size=", data.len
     return err(DeblobWrongSize)
   if data[^1] != 0x7f:
     return err(DeblobWrongType)
diff --git a/nimbus/db/aristo/aristo_desc/desc_error.nim b/nimbus/db/aristo/aristo_desc/desc_error.nim
index b69864832..8e72d52ac 100644
--- a/nimbus/db/aristo/aristo_desc/desc_error.nim
+++ b/nimbus/db/aristo/aristo_desc/desc_error.nim
@@ -28,13 +28,14 @@ type
     SerCantResolveStorageRoot
 
     # Data record transcoders, `deblobify()` and `blobify()`
-    BlobifyNilFilter
-    BlobifyNilVertex
     BlobifyBranchMissingRefs
     BlobifyExtMissingRefs
     BlobifyExtPathOverflow
     BlobifyLeafPathOverflow
-    BlobifyFilterRecordOverflow
+    BlobifyNilFilter
+    BlobifyNilVertex
+    BlobifyStateSrcLenGarbled
+    BlobifyStateTrgLenGarbled
 
     DeblobNilArgument
     DeblobUnknown
diff --git a/nimbus/db/aristo/aristo_desc/desc_identifiers.nim b/nimbus/db/aristo/aristo_desc/desc_identifiers.nim
index 458dfbeb0..658336fc5 100644
--- a/nimbus/db/aristo/aristo_desc/desc_identifiers.nim
+++ b/nimbus/db/aristo/aristo_desc/desc_identifiers.nim
@@ -29,6 +29,19 @@ type
     ## keys, the vertex component will be called a node. On the persistent
     ## backend of the database, there is no other reference to the node than
     ## the very same `VertexID`.
+    ##
+    ## Vertex IDs are generated on the fly and thrown away when not needed,
+    ## anymore. They are not recycled. A quick estimate
+    ##
+    ##   (2^64) / (100 * 365.25 * 24 * 3600) / 1000 / 1000 / 1000 = 5.86
+    ##
+    ## shows that the `uint64` scalar space is not exhausted in a 100 years
+    ## if the database consumes somewhat less than 6 IDs per nanosecond.
+    ##
+    ## A simple recycling mechanism was tested which slowed down the system
+    ## considerably because large swaths of database vertices were regularly
+    ## freed so recycling had do deal with extensive lists of non-consecutive
+    ## IDs.
 
   HashKey* = object
     ## Ethereum MPTs use Keccak hashes as node links if the size of an RLP
diff --git a/nimbus/db/aristo/aristo_init/memory_db.nim b/nimbus/db/aristo/aristo_init/memory_db.nim
index d5dc60c8f..b9eed8159 100644
--- a/nimbus/db/aristo/aristo_init/memory_db.nim
+++ b/nimbus/db/aristo/aristo_init/memory_db.nim
@@ -168,7 +168,14 @@ proc putLstFn(db: MemBackendRef): PutLstFn =
     proc(hdl: PutHdlRef; lst: SavedState) =
       let hdl = hdl.getSession db
       if hdl.error.isNil:
-        hdl.lSst = some(lst)
+        let rc = lst.blobify # test
+        if rc.isOk:
+          hdl.lSst = some(lst)
+        else:
+          hdl.error = TypedPutHdlErrRef(
+            pfx:  AdmPfx,
+            aid:  AdmTabIdLst,
+            code: rc.error)
 
 proc putEndFn(db: MemBackendRef): PutEndFn =
   result =
@@ -287,7 +294,7 @@ iterator walk*(
   if be.mdb.tUvi.isSome:
     yield(AdmPfx, AdmTabIdTuv.uint64, be.mdb.tUvi.unsafeGet.blobify)
   if be.mdb.lSst.isSome:
-    yield(AdmPfx, AdmTabIdLst.uint64, be.mdb.lSst.unsafeGet.blobify)
+    yield(AdmPfx, AdmTabIdLst.uint64, be.mdb.lSst.unsafeGet.blobify.value)
 
   for vid in be.mdb.sTab.keys.toSeq.mapIt(it).sorted:
     let data = be.mdb.sTab.getOrDefault(vid, EmptyBlob)
diff --git a/nimbus/db/aristo/aristo_init/rocks_db.nim b/nimbus/db/aristo/aristo_init/rocks_db.nim
index 845e1d929..2a029f536 100644
--- a/nimbus/db/aristo/aristo_init/rocks_db.nim
+++ b/nimbus/db/aristo/aristo_init/rocks_db.nim
@@ -215,7 +215,13 @@ proc putLstFn(db: RdbBackendRef): PutLstFn =
     proc(hdl: PutHdlRef; lst: SavedState) =
       let hdl = hdl.getSession db
       if hdl.error.isNil:
-        db.rdb.putByPfx(AdmPfx, @[(AdmTabIdLst.uint64, lst.blobify)]).isOkOr:
+        let data = lst.blobify.valueOr:
+          hdl.error = TypedPutHdlErrRef(
+            pfx:  AdmPfx,
+            aid:  AdmTabIdLst,
+            code: error)
+          return
+        db.rdb.putByPfx(AdmPfx, @[(AdmTabIdLst.uint64, data)]).isOkOr:
           hdl.error = TypedPutHdlErrRef(
             pfx:  AdmPfx,
             aid:  AdmTabIdLst,
diff --git a/nimbus/db/core_db/backend/aristo_db.nim b/nimbus/db/core_db/backend/aristo_db.nim
index 1d7d84e88..60313b611 100644
--- a/nimbus/db/core_db/backend/aristo_db.nim
+++ b/nimbus/db/core_db/backend/aristo_db.nim
@@ -11,7 +11,7 @@
 {.push raises: [].}
 
 import
-  std/[tables, typetraits],
+  std/tables,
   eth/common,
   results,
   ../../aristo as use_ari,
diff --git a/nimbus/db/core_db/core_apps_newapi.nim b/nimbus/db/core_db/core_apps_newapi.nim
index dac5a7476..dc513ba2d 100644
--- a/nimbus/db/core_db/core_apps_newapi.nim
+++ b/nimbus/db/core_db/core_apps_newapi.nim
@@ -353,18 +353,15 @@ proc getSavedStateBlockNumber*(
   ## the `relax` argument can be set `true` so this function also returns
   ## zero if the state consistency check fails.
   ##
-  var
-    header: BlockHeader
-  let
-    st = db.ctx.getMpt(CtGeneric).backend.toAristoSavedStateBlockNumber()
-    # The correct block number is one step ahead of the journal block number
-    bn = st.blockNumber + 1
-  if db.getBlockHeader(bn, header):
+  var header: BlockHeader
+  let st = db.ctx.getMpt(CtGeneric).backend.toAristoSavedStateBlockNumber()
+  if db.getBlockHeader(st.blockNumber, header):
     discard db.ctx.newColumn(CtAccounts,header.stateRoot).valueOr:
       if relax:
         return
-      raiseAssert "getSavedStateBlockNumber(): state mismatch at #" & $bn
-    return bn
+      raiseAssert "getSavedStateBlockNumber(): state mismatch at " &
+        "#" & $st.blockNumber
+    return st.blockNumber
 
 proc getBlockHeader*(
     db: CoreDbRef;