diff --git a/shardwidth/helper.go b/shardwidth/helper.go
new file mode 100644
index 000000000..c04e2e72b
--- /dev/null
+++ b/shardwidth/helper.go
@@ -0,0 +1,70 @@
+// Copyright 2021 Molecula Corp.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package shardwidth
+
+import (
+	"math/bits"
+)
+
+// FindNextShard returns the index of the first item which is not in the
+// same shard as i. The index it returns may be equal to the length of the
+// haystack, indicatincg that the rest of the list is in the same shard.
+func FindNextShard(i int, haystack []uint64) int {
+	// compute the last thing that's in the same shard as haystack[i].
+	if i >= len(haystack) {
+		return i
+	}
+	// current shard:
+	shard := (haystack[i] >> Exponent)
+	// last value in shard:
+	shardEnd := ((shard + 1) << Exponent) - 1
+	j := i
+	// We want to do a binary search of the haystack. For any length of
+	// haystack, its topmost bit gives us a reasonable halfway point; it may
+	// not actually be halfway, but the number of steps it'll take to search
+	// it will be the same as if it were. sort.Search has interface overhead
+	// and makes us sad.
+	for incr := 1 << (bits.Len64(uint64(len(haystack) - i))); incr > 0; incr >>= 1 {
+		if j+incr < len(haystack) {
+			if haystack[j+incr] <= shardEnd {
+				j += incr
+			}
+		}
+	}
+	// we've found the last item that is in the same shard as i, so...
+	return j + 1
+}
+
+// FindShards finds the shards in a given haystack
+func FindShards(haystack []uint64) (shards []uint64, endIndexes []int) {
+	if len(haystack) == 0 {
+		return nil, nil
+	}
+	index := 0
+	// the steady state of this loop is that shards contains the current
+	// shard, but not its ending index; each time we find a new ending
+	// index, we record that index as the end for the current shard, and
+	// the new shard, until we reach the end and append len(haystack)
+	// as the last index.
+	shards = []uint64{haystack[index] >> Exponent}
+	index = FindNextShard(index, haystack)
+	for index < len(haystack) {
+		shards = append(shards, haystack[index]>>Exponent)
+		endIndexes = append(endIndexes, index)
+		index = FindNextShard(index, haystack)
+	}
+	endIndexes = append(endIndexes, index)
+	return shards, endIndexes
+}
diff --git a/shardwidth/helper_test.go b/shardwidth/helper_test.go
new file mode 100644
index 000000000..cd1806cd8
--- /dev/null
+++ b/shardwidth/helper_test.go
@@ -0,0 +1,103 @@
+// Copyright 2021 Molecula Corp.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package shardwidth_test
+
+import (
+	"math/rand"
+	"testing"
+
+	"github.com/molecula/featurebase/v2/shardwidth"
+)
+
+type nextShardTestCase struct {
+	name         string
+	haystack     [][2]uint64 // stored as shard, offset pairs
+	shardIndexes []int
+}
+
+var nextShardTestCases = []nextShardTestCase{
+	{
+		name: "all-in-one",
+		haystack: [][2]uint64{
+			{0, 0},
+			{0, 1},
+		},
+		shardIndexes: []int{2},
+	},
+	{
+		name: "split",
+		haystack: [][2]uint64{
+			{0, 0},
+			{1, 1},
+		},
+		shardIndexes: []int{1, 2},
+	},
+	{
+		name: "two-and-one",
+		haystack: [][2]uint64{
+			{0, 0},
+			{0, 1},
+			{1, 1},
+		},
+		shardIndexes: []int{2, 3},
+	},
+}
+
+func TestFindShards(t *testing.T) {
+	for _, c := range nextShardTestCases {
+		haystack := make([]uint64, len(c.haystack))
+		for i, h := range c.haystack {
+			haystack[i] = (h[0] << shardwidth.Exponent) + h[1]
+		}
+		_, indexes := shardwidth.FindShards(haystack)
+		if len(indexes) != len(c.shardIndexes) {
+			t.Fatalf("%s: expected %d, got %d", c.name, c.shardIndexes, indexes)
+		}
+		for i, expected := range c.shardIndexes {
+			if indexes[i] != expected {
+				t.Fatalf("%s: expected index %d to be %d, got %d", c.name, i, expected, indexes[i])
+			}
+		}
+	}
+	// fake up some more test cases
+	for i := 0; i < 100; i++ {
+		haystack := make([]uint64, 100)
+		shard := uint64(0)
+		bit := uint64(0)
+		shardIndexes := []int{}
+		for j := range haystack {
+			if rand.Intn(30) == 0 {
+				if j > 0 {
+					shardIndexes = append(shardIndexes, j)
+				}
+				shard++
+				bit = 0
+			} else {
+				bit += uint64(rand.Intn(30))
+			}
+			haystack[j] = (shard << shardwidth.Exponent) + bit
+		}
+		shardIndexes = append(shardIndexes, len(haystack))
+		_, indexes := shardwidth.FindShards(haystack)
+		if len(indexes) != len(shardIndexes) {
+			t.Fatalf("trial %d: expected %d, got %d", i, shardIndexes, indexes)
+		}
+		for idx, expected := range shardIndexes {
+			if indexes[idx] != expected {
+				t.Fatalf("trial %d: expected index %d to be %d, got %d", i, idx, expected, indexes[idx])
+			}
+		}
+	}
+}